Print This Post Print This Post

[This is intended as a rapid, rough blog response. It is unedited and unrevised and no doubt contains many errors of spelling, grammar and argument.]

In this chapter, Frances Kamm contrasts her own non-consequentialist ethical theory with Peter Singer’s version of consequentialism, utilitarianism. She examines his general ethical theory and its implications for killing/letting die, treatment of the disabled and animals, and famine relief. I will only offer a skeletal summary of the first half. I will focus on the second half, which for me is incredibly rich and stimulating and more oriented to practical ethics. I will attempt to give a consequentialist response to her arguments in the latter half of the chapter. The first half, however, is of great interest to those involved in normative theory and its foundations. For those interested in more detailed analysis and response to Kamm’s arguments, please move straight to the section 1. The Principle of Irrelevant Goods.

The chapter begins with an analysis of Peter Singer’s, or strictly, Richard Hare’s two level approach to moral thinking. The higher level is impartial consequentialism. The lower level consists in various rules or character traits which make the outcome go best. Kamm believes Singer is committed to

“the possibility that we should develop certain character traits or patterns of thinking that make it impossible to think of breaking rules that maximize good consequences overall …”

She then exposes several problems with this two level account. Firstly, she explores his “motivational argument.” She claims that Singer is committed to the view that the motivation to be moral at the higher order derives from self-interest (not rationality) and that being moral is in an individual’s self-interest because she has an interest in a meaningful life. It is not clear (to me at least) that Singer would or should commit himself such a motivational argument based on self-interest. He could adopt a dualistic account, like Sidgwick’s, and claim that there are two reasons which motivate people: morality and self-interest. For Singer, according to Kamm, moral conscientiousness or acting for morality’s own sake, occurs at the lower level.

Kamm argues that if the justification for an impartial morality is that it is necessary for individuals to lead meaningful lives, then this justification fails. For individuals may lead meaningful lives without any commitment to or requiring any morality – simply devoting themselves to personal relations and work.

Secondly, Kamm gives a version of Williams’ “government house utilitarianism” objection that a two level model of normative theory works best if only a few people know the impartialist consequentialist justification and most merely know the rules, virtues and desirable character traits.

I have only sketched some of her critique against a 2 level view. Of course, consequentialists need not sign up to a schizophrenic two level account.

The following section deals in greater detail with Singer’s moral point of view: impartial consequentialism. Impartiality involves

1. equal consideration of interests, where interests are defined by what an individual would rationally prefer
2. maximizing the overall satisfaction of interests
At this level, there is no place for partiality, say giving special weight to one’s own family.

Kamm considers various ways in which non-moral considerations on this kind of account can legitimately outweigh or override moral considerations, such that we can still act morally overall. For example, it might be legitimate to tell a little lie to present a great work of art. So minor moral considerations can be outweighed by serious aesthetic or economic considerations, but serious moral considerations are never outweighed. Thus, according to Kamm, even those who disagree with Singer about the place of partiality within the moral point of view can agree that moral considerations are overrideable only in this limited way.

Kamm then goes on to analyse the implications of Singer’s account of interests and their strict maximization. For Singer, according to Kamm, our interests go beyond pleasure and avoidance of pain, to an interest in a meaningful life, even if we do not know we do or are too irrational to prefer it. The kind of being with interests is of no moral significance – species, for example, is an irrelevant consideration. The pleasure of a mouse is ot be weighed equally with the pleasure of a person.

One special interest is the interest in continuing to live. On Singer’s view, only self-conscious beings have an interest in not being killed. Non-self-conscious beings are replaceable. Kamm rightly points out that Singer’s philosophy would find no problem in killing non-self-conscious animals, provided this did not inflict suffering directly or indirectly.

One implication of this kind of view is that it is committed to a “great rescue effort” to deal with all pain that exists, in all sentient animals. One way to resist this is to adopt Scanlon’s suggestion that we only owe moral obligations to beings who can act for reasons. Such a view would allow us to give priority to our own suffering but it would also imply that we do not owe, on these grounds, a duty to relieve the pain of the severely retarded.

Kamm then asks whether there might be reasons to believe it wrong to kill non-self-conscious beings:

“If a being’s future time were to be filled with good experiences, these good experience would be its good experiences. Death would deprive it of this, and so it seems there is a reason not to kill it…”

One of the major problems with Singer’s account of interests, according to Kamm, is that much of morality is about “rational agency, responsibility, and respect for what someone wills rather than interests.”

The final critique of Singer’s moral point of view is directed towards is stated commitment to pure aggregation of interests. Small benefits to very many would outweigh enormously significant benefits to a few. Curiously, Kamm thinks that this weakens Singer’s arguments in support of aid and famine relief. While in theory the significant benefits to the starving and poor could be outweighed by massive numbers of much more well off people experiencing smaller benefits, in practice this si hardly likely to be the case. The situation is one where very large numbers of people could enjoy very great benefits at the cost of very large numbers of people forgoing much smaller benefits or luxuries. Singer’s approach is, in contrast to Kamm’s, tightly connected to the real world and facts about what would in practice make things go best. The fact that there is some possible world where utilitarians are committed to inflicting some great evil for the sake of small benefits to the many is not a concern for utilitarians like Hare and Singer, who are concerned with what their theory yields for the real world. Moreover, it is my experience that when confronted with such cases, they will often make the empirical speculation that the long term consequences of such practices (eg allowing slavery for the benefit of billions of slave owners) are for the worse. That is, they retreat to a rule consequentialist objection to such practices.

1. The Principle of Irrelevant Goods
Kamm draws attention to a much underscrutinised aspect of Singer’s maximizing theory that has important practical implications:

“even small differences in expected outcome always give us reason to choose the better outcome.”

For this reason, we should give greater priority in the distribution of limited resources to the able bodied rather than disabled. That is, while a disabled person might have a full right to life, faced with the choice of saving a disabled person or an able bodied person, Singer’s view commits us to saving the latter. Indeed, even if a person is prone to recurrent migraine, we have reason to save the life of a person who does not have migraines, other things being equal. Call this the Principle of Priority for the Able. We should agree with Kamm that this principle is strongly counterintuitive. It seems to violate a principle of equal respect for persons.

Kamm rightly points out that disability makes a life go worse. Call this the Badness of Disability. It is this consideration that means the outcome is better if we save able rather than disabled people. She then sets out to retain the Badness of Disability while rejecting the Principle of Priority for the Able which Singer takes to follow from it. She claims that

“a difference in outcome could be morally relevant in one context … but morally irrelevant in another.”

She calls this the Principle of Irrelevant Goods. She applies this to the Hand Case. In this case, we can save one of two men, A and B. They are relevantly identical except B has recently lost his hand. The Principle of Priority for the Able requires we save A because his life will go slightly better with the use of 2 hands.

Kamm rejects this conclusion and argues the Principle of Irrelevant Goods accounts for intuitions about this case. She argues that when what is on offer is a long, good quality of life, “what is reasonably held to be most important for each person” – then the benefit of the additional hand is irrelevant in this context. “…we should take seriously from an impartial point of view the fact that each person, from his personal perspective, wants to be the one to survive.”

The Principle of Irrelevant Goods is one way to resist the strict Principle of Priority for the Able. However, questions arise: how do we decide whether or when a good is irrelevant in a non-question-begging way? Are all goods irrelevant? Consequentialists can also hold the Badness of Disability but resist the strict Priority for the Able, as I will argue.

It is important to note that disability is not always an irrelevant good in such saving cases. In 2001, I wrote an editorial for the British Medical Journal on the Brompton Report which examined allegations that children with Down syndrome had been denied cardiac surgery on the basis of their disability. I pointed out that in Australia, about one third of children on heart transplant waiting lists die before transplantation because of the scarcity of hearts. With a severe shortage of hearts, transplanting a child with Down syndrome implies that a child without Down syndrome will die who would otherwise have received a transplant.

Should quality of life be a relevant factor in deciding how to allocate scarce resources like donor hearts? The Report was critical of reference to quality of life evaluations and

“value judgements – for example related to factors such as limited lifespan, inability to get the most out of life, or not being a burden upon others or upon society.”

Equality of access urges us to ignore quality of life considerations. So, too, does the law. The European Convention on Human Rights 1950 (which is now British law) states that:

‘Everyone’s right to life shall be protected by law …. No one shall be deprived of his right to life intentionally …’ and ‘The enjoyment of the rights and freedoms set forth in this Convention shall be secured without discrimination on any ground such as sex, race, colour …. or other status’ (Sect. 27.13)

Denying children with Down syndrome access to transplants is probably unlawful discrimination according to this law and the Disability Discrimination Act 1995.

Yet we do appeal to quality of life in deciding how much benefit people derive from scarce resources. For example, people with brain injury or dementia resulting in a severely impaired cognitive state (such that they cannot interact socially with other people) may be denied transplantation, not because it is against their interests, but because the scarce organs would do more good if directed to a person without such impairments. Quality of life is taken into account when allocating organs, such as when hearts are matched to recipients of a similar size to provide the best functional result.

If we appeal to equality of access and fail to give weight to the magnitude of the benefit a person expects to derive from treatment, absurd consequences follow. Equality of access requires that we ignore not just quality of life but also length of life Imagine George is one year old and has a metabolic abnormality. He will die in one year. He needs a heart transplant if he is not to die in the next few weeks. John is one and has cardiomyopathy and could expect up to 20 years of life if he receives a heart transplant. According to equality of access, both need a heart transplant to live, so both should have an equal chance. This seems absurd. We should give priority to John.

The idea of equality of access is that each of us has a life of equal value and arbitrary factors like disability should not be considered in determining our eligibility for treatment. John Harris claims that each rational person wants at least 3 things from health care: (i) the maximum possible life-expectancy for him or her; (ii) the best quality of life for him or her; (iii) the best opportunity or chance for him or her of getting both (i) and (ii). Treating people as equals involves giving equal weight to each person’s own claim. As Harris recognised, a principle of equality cannot only be selectively invoked by those with disability (on pain of itself being discriminatory) but also applies to those who happen to have poor prognoses or diseases which are expensive diseases to treat.

Equality of access thus requires that we ignore the probability of survival. This is inconsistent with accepted practice. Everyday older women in Britain are denied access to in vitro fertilisation because they have a lower chance of a successful outcome. We do not follow the principle that “anything goes” irrespective of likelihood of success (as judged in terms of survival and quality of life). In 1995 Jaymee Bowen (Child B) was denied a second bone marrow transplant for leukemia because it would cost £75,000 and there was little chance of success. Her father took Cambridge Health Authority to court. Sir Thomas Bingham, Master of the Rolls, appealed to a principle of maximising benefit when he reject the father’s appeal this way:

“Difficult and agonizing judgments have to be made as to how a limited budget is best allocated to the maximum advantage of the maximum number of patients.”

Indeed, the General Medical Council has stated that,

‘the clinical team in determining priorities and the utilisation of the resources made available to them by the NHS is entitled to take into account the likely success of the treatment proposed.’

Under conditions of sufficient resources, equality of access is ethically mandated. However, under conditions of insufficient resources to treat everyone (and these will always be with us), unthinking application of the principle of equality means fewer people will live, and those who do live will live shorter periods and lead worse quality lives. It also rejects a cost-effectiveness approach to maximising benefits to different but mutually exclusive populations of patients.

Quality of life, length of life and probability of beneficial effect (and cost of treatment) are relevant considerations in determining who should receive treatment. Severe disability in some circumstances should disqualify a person from access to scarce resources. With the current shortage of donor hearts, it would be wrong to transplant a child with severe intellectual disability who was barely conscious or a child with congenital myotonia dystrophica who was dependent on a ventilator.

It is probably unlawful to place lower priority on children with Down syndrome and other disabilities who need heart transplants. But is it unethical? Doctors cannot and should not be involved in fine grained evaluations of the worthiness of different lives. Someone is not a less worthy recipient of a heart transplant because he has a finger missing or a speech impediment. A tolerant and affluent society would strive to provide equality of access to everyone for as many beneficial interventions as possible. Whether disability such as Down syndrome should be considered relevant in the allocation of a scarce resource turns on how much the disability associated with it detracts from a good life.

Down syndrome is associated with intellectual disability, infertility, reduced opportunities for independent living and employment, shorter life span, early onset Alzheimer Disease. These all make those lives go worse. But there is also considerable variation in the quality of people’s lives with disability and with Down syndrome in particular. A person with intellectual disability can have a good, happy and worthwhile life. This is why it is essential to judge every case of heart transplantation on its merits, depending upon all of the factors, but including the likelihood of a good outcome in any particular case.

So, it appears that disability can be a relevant bad in rationing cases. How might she respond? Kamm might separate her justification for the Principle of Irrelevant Goods. While claims “what is reasonably held to be most important for each person” and “…we should take seriously from an impartial point of view the fact that each person, from his personal perspective, wants to be the one to survive” take her close to Harris’ extreme egalitarian position, she could retreat to claiming that the goal is not to give each person what that person would rationally desire or need to survive, but “a long, good quality of life.” She could claim a person with Down syndrome cannot expect to have a long, good quality of life and on that basis, should be denied a heart transplant in favour of a normal child. Or she could bite the bullet and claim that the life of a child with disability such as Down Syndrome is long and good enough, and so that child should be given an equal chance of receiving a heart transplant.

I am not arguing that Kamm does not have the armamentarium to deal with intuitions where disability does seem to be a relevant bad in rationing cases. Rather, I am arguing that consequentialists can respond to these cases and her challenges. Here are two responses.

Firstly, her objection is to maximizing consequentialism. Submaximizing versions would claim that if the benefit is significant enough, above some threshold, then people’s claims should be treated equally.

John Mackie once suggested, against utilitarians, that everyone has a ‘right to a fair go’ . This an attractive principle. According to a maximizing version of this principle, we should give as many people as possible a decent (reasonable) chance of having a decent (good) life. This has also been called “sufficientarianism.”

According to a version of suffientarianism which I hold, which I have called “a right to a fair go”:

1. each person has a legitimate claim to medical care when that care provides that person with reasonable chance of reasonable extension of life and/or a reasonable improvement in its quality.
2. comparable legitimate claims are those referring to similar needs.
3. as many comparable legitimate claims should be satisfied as possible.
4. provided as many comparable legitimate claims are being satisfied as possible, there should be equality of access to medical care, and no allocation on the basis of irrelevant factors such as socioeconomic status, race, gender, etc.

This, I believe is a version of consequentialism capable of dealing with the problem of irrelevant goods. What constitutes “reasonable” is determined in part by the scarcity of the resource. A reasonable prolongation of life in paediatric cardiac transplantation would be years, not months. It would be wrong to transplant children with severe intellectual disability with the result that children with much better intellectual functioning die. However, there is also great variation in the quality of people’s lives with disability and with Down syndrome in particular. It is true that Down syndrome is associated with intellectual disability, infertility, reduced opportunities for independent living and employment, shorter life span, early onset Alzheimer Disease. But a person with Down syndrome can have a good life. That might be enough to justify a claim on scarce community resources in a particular case, depending on the level of disability.

Secondly, strict maximizing consequentialists could claim that while the loss of a hand is bad, it is not possible to tell whether a whole person’s life with the loss of a hand is worse than a similar person with two hands. Thus, in the real world, all other considerations will not be equal so the presence of an extra hand is irrelevant to the decision. We simply cannot predict who will have a better life on the basis of that consideration. I do not favour this option, but it is a possibility.

Kamm is right that the Principle of Irrelevant Goods is a problem for some consequentialists. But is not a problem for all consequentialists. Moreover, some goods are relevant and her position will need to be expanded to account for this.

2. Killing/Letting Die
In these final sections, Kamm considers a central consequentialist tenet, and one prominent in Singer’s philosophy, that there is no intrinsic moral distinction between harming and failing to aid, or between killing and letting die, or, as it has also been put, between acts and omissions. These distinctions are an important part of common sense morality, so Kamm’s arguments have special practical relevance, especially in medical ethics.

Kamm’s defence of the distinction turns on conceptual analysis of what is intrinsic and what is extrinsic to an act. Singer claims that we mistakenly believe that killing is worse than letting die because of extrinsic factors and not due to intrinsic factors. He attempts to show this by constructing cases in which all apparent extrinsic factors, like motive and effects, are the same between killing and letting die. The famous example of this kind of thought experiment aimed at equalising all extrinsic factors is James Rachels’ case of Smith and Jones, one who kills his nephew to obtain an inheritance, and the other lets the nephew die for the same end, with the same intentions.

Kamm responds that it is possible to export a property that is necessarily intrinsic to (i.e., essentially true of) cases of killing to some cases of letting die. The example of such a property is “causing death.” If A stabs B killing B, A causes B’s death. If A removes B from a life support machine belonging to A, resulting in B’s death, A causes (in part) B’s death.

It is also possible, according to Kamm, to export properties “intrinsic to letting die (but not to killing) into a case of killing.” The candidate property here is “someone losing out on only life he would have had without my help.” This can be exported to some cases of killing, reducing its wrongness. For example, if I stab someone on life support which is mine and who would not have lived without my support.

Kamm argues that if some intrinsic property of killing is exported to letting dies cases and it makes them worse, this is evidence for a moral difference between killing and letting die. However, she does not believe that exporting “causing the death of” into letting die cases makes them worse.

[A short note: I fail to see how “causing the death of” is not necessarily intrinsic to all cases of letting die. If an instance of letting die failed to cause the death of someone, how could it be a case of letting die? It is essentially true of the claim “A let B die” that A causes (in part) the death of B. So I fail to see how “causing the death of” is an exportable property from killing to letting die; it seems (to me at least) intrinsic to both.]

However, if some intrinsic property of letting die is exported to killing cases and it makes them worse, this is evidence for a moral difference between killing and letting die. Kamm does believe that the property of “someone losing out on only life he would have had without my help” satisfies this criterion.

I believe that Kamm has identified the critical difference between killing and letting die cases, but that this property is extrinsic to both. It is the degree of “help” that is typically involved in cases of refraining from killing and in saving that accounts for intuitions about killing and letting die, I will argue.

For the present, I would like to suggest that the mere property of “someone losing out on only life he would have had without my help” does not itself reduce the badness of killing. Imagine that X’s father-in-law has a heart attack. X saves his life by performing mouth to mouth resuscitation. Later, after X’s wife dies, the father-in-law obstructs X seeing his children and X kills him. Even though he loses out on only life he would had without X’s help, X killing him would be equally wrong as if he had not had a heart attack at all. Certainly, courts would find X guilty of murder and punish him in the same way.

Kamm’s second strategy is to show that there is an intrinsic difference between killing and letting die by showing that the extrinsic factors that make letting die permissible do not make killing permissible. She uses Singer’s example of “burdensomeness”, which I will go on to argue is critical. One reason why letting die is often seen as less bad than killing is because it is more burdensome. Kamm argues that if there were no intrinsic difference between killing and letting die, then if it were very burdensome to avoid killing someone, this would reduce the badness of killing. But, she argues, it does not. Here is her example.

“If I drive down road A, I will run over someone stuck in the road. I have a duty to take road B, even if it is dangerous and will cause damage to my car.”

This example, however, is stacked in Kamm’s favour. Mere damage to one’s car is not a justification to take someone’s life. Imagine by “dangerous” we mean high probability of lethal threat by terrorists or bandits. It would be superogatory to take B, but permissible to take A. Or imagine the bandits on road B will not kill me, but abduct my children. My children will have a tolerable life, living at the sort of level Singer would have us live, but I will never see them again. Does not this fact make the wrongness of killing the person stuck on road less? When travelling on road B involves great sacrifices it is more permissible to travel road A.

Note that it is not necessary to show that it is right to travel on road A, but merely that as the sacrifice involved in travelling the road of the bandits increases, the wrongness of taking A decreases. This is enough to show that the extrinsic factor of burdensomeness does influence the wrongness of killing.

Kamm’s 2 strategies, at least as presented, do not conclusively establish her case against consequentialists. She then goes on to make the striking claim that “to rebut the general thesis that there is no intrinsic difference between killing and letting die per se, we need present only one case where the moral difference between the two shows up.” Consequentialists, however, must not present one striking case where there is no difference, but consider many cases. Here is her case of a moral difference:

(a) I must rush five dying people to the hospital to save them. To get there on time, I must speed over a road where I know one person is stuck, thereby killing him. This is impermissible.
(b) I must rush five dying people to the hospital to save them. To get there on time, I must pass someone drowning in a pond, thereby allowing him to die. This is permissible.

This case, again consequentialists might argue, is unclean, polluted by background extrinsic intuitions. For example, “on time” creates an aura of convenience, which could never justify killing. The kind of people who are dying and need to be taken to the hospital represent, in our minds, certain classes of people: the old, those with little time to live or likely to die anyway, those who have been responsible for their illness or accident, whereas the person stuck on the road is likely to be fit, healthy and innocent.

To cleanse our mind of these associations and biases, we need an example where these factors are clearly not present.

Imagine there is a dirty bomb in London. Many people are exposed to toxic levels of radiation and will die in 6 months from the effects of the radiation. They are now, however, apparently well. Treatment needs to be immediately implemented if the effects of the radiation are to be prevented. An ambulance is rushing to a hospital with 5 ten year old children who have been exposed to the radiation, but who could be saved. However, another ten year old child who has also been exposed to the radiation (and will die in 6 months) is stuck on a debris strewn road which is the only access to the hospital. To reach the hospital, the ambulance drivers must drive over the wreckage, killing the one child, in order to get the 5 to hospital.

Is it really impermissible to kill an innocent child in this example? Consequentialists have the intuition it is not. Some non-consequentialists will have the intuition it is impermissible. That does not matter. What matters is it is not clear that killing in this case is impermissible. I, and many others, believe it is permissible.

Kamm finishes this section with some insightful remarks on why there might be a difference between killing and letting die, based on “our entitlement to personal resources that are connected to what makes us separate persons.” She claims that “persons are entitled to certain things that are both related to their identity as separate persons and that they have independently of our help, for example, their bodies …” If we harm them and cause them to lose these things, we encroach their entitlements. However, if we fail to aid someone, she loses out only what she would have had with our help. This, I believe, is part of the correct analysis of the difference between killing and letting. However, it is not the whole story.

3. A Moderate Libertarian Consequentialism: Duty of Easy Rescue
Kamm focuses on the separateness of persons and the rights or entitlements of people not to be harmed. This is a rights-based or non-consequentialist or “negative” perspective of morality. Consequentialists have a more positive perspective, focussing on what we owe to others, and our duties to benefit. This is an important part, if not the whole part, of morality.

To my mind, the strongest critique of consequentialism is that it fails to respect our separateness as persons and it is overly demanding, requiring us to give up our deepest projects and commitments for the sake of benefiting others. We can share with Kamm a deep commitment to the importance of the separateness of persons and the idea that each of his is an individual who should have considerable scope to live our lives as we choose, free from interference by others or burdensome constraints, such as the demands of an excessive morality.

If we placed individual liberty at the centre of a consequentialist doctrine, as Mill attempted to do, we might get a very different consequentialism to that advocated by Peter Singer. For example, imagine that we made own individual liberty our primary consequentialist value. So for an action to be morally required of me, it must be consistent with my nature as an agent free to substantially choose his own life. It cannot require great sacrifices of me. We can still retain a commitment to promoting well-being, our own and others, within such a libertarian framework.

Sidgwick described the dualism of practical reason: self-interest and impartial morality. We could replace “self-interest” with “liberty”. Such as libertarian consequentialism would create only duties of easy rescue: that where the cost to onself of an act X is small, and the benefit to others is large, we are required to do X. Singer is attracted to the principle of a duty of easy rescue in motivating his aid arguments.

On this view, what accounts for the difference between cases of killing and letting die is typically the burdensomeness or degree of help that is required to benefit (or not harm) the person affected by our actions. That is, the degree to which saving or not killing requires we give up our liberty. Typically, refraining from killing someone is costless and saving someone requires sacrifice. It is the degree of sacrifice, not the mere fact of helping, that accounts for intutions about killing and letting die.

We can supplement Kamm’s view that what accounts for intuitions about killing and letting die are beliefs about the rights of the victims to non-interference, to their own liberty, with the consequentialist belief that we should benefit others, we have a moral obligation to, but only when the cost in terms of liberty and sacrifice is not substantial to the actor.

Jeff McMahan, in The Ethics of Killing, argues that killing fails to respect persons. He postulates that this failure of respect may be partly derived from convention –”the signs and tokens of respect” (390). Or it may derive its significance from the more general distinction and claim that it is worse to harm than to fail to prevent harm. This itself may be the product of

”a convention that has evolved as a tacit compromise between the insistence of the powerful on a norm of nonintervention and the insistence of the weak on a norm of mutual aid. If so, our belief that killing is worse than letting die is, in the first instance, a product of our conditioned or learned acceptance of norms that are necessary for social stability.” (390)

But the norm of a duty of easy rescue can also account for many intuitions AROUND killing-letting die and it builds on McMahan’s idea that the distinctions are built upon two competing norms which are necessary for social stability. On one view of the world, often favoured by contractarians, we are individuals with our own lives to lead in a society which exists to protect the liberty of the individual. This is not merely the claims of the powerful to nonintervention. We all have a fundamental interest in liberty. Society is organised to allow each individual significant liberty to lead his or her own life as he or she chooses. The fundamental unit of morality is the individual, and his or her own life plan. This kind of liberty is negative liberty – the freedom from interference of others or being called upon to perform certain actions for the benefit of others. According to libertarians, protection of liberty is absolute.

Consequentialists can accommodate such a view of the world by maintaining that we should act so as to maximise the liberty of individuals. According to any view, pure libertarian or libertarian consequentialist, that places absolute or high value on liberty, liberty must be preserved regardless of the extent of suffering which this causes. This kind of view is extremely harsh. According to this view, a man who sees a small child drowning in a wading pool has no obligation to pluck the child and save its life, though many think that he does have such an obligation. Or the businessman on the way to the office has no duty to stop at the roadside to call an ambulance for an injured road accident victim who would otherwise die.

What is plausible about views which place great value on liberty is that it is important to each of us that we be free to live as we choose in society. We benefit and society benefits from having a high degree of certainty in freedom, and not being compelled to sacrifice what is dear to us for the sake of others or State ideals. We can organise our lives knowing that we are their master. However, some curtailments of liberty seem justified. Taxes, seat belts, smoking warnings all impinge on liberty but because the costs to the individual are small, and the benefits to great, we accept them. A moderate libertarian or libertarian consequentialist position can accept a duty of easy rescue: when the costs to a person are zero or trivial of a course of action, and the benefits to others are significant, that course should be undertaken. Such a duty would impose only infrequent incursions of liberty. While duty of easy rescue generally implies a duty to act, duty of easy rescue here should be taken to mean a duty to undertake a course of action, including an omission to act by the agent.
What is plausible about the libertarian position is that large, frequent wholesale incursions of liberty are wrong. For example, being required to sacrifice one’s career or family for the sake of some cause or persons. The occasional minor infringement of liberty for the sake of society, others and the promotion of liberty and well-being is rational and desirable. Such a principle would provide great benefits and liberties to all, as we are all equally likely to be the beneficiaries as those called upon to perform some duty of rescue. For example, we are all possible victims of car accidents – we each stand to benefit from policy of requiring passers-by to stop to call an ambulance if the costs to them are not great. This is not merely ”insistence of the weak on a norm of mutual aid”- we all stand to benefit enormously from a rule of easy rescue. Even the powerful can be involved in car accidents.

For moderate libertarians and libertarian consequentialists, it is the cost to the liberty of the individual asked to perform some action which is the first consideration. If the cost is small, the action can be required. While rights as discussed in the previous section protect victims from harm, negative liberty protects individuals from the claims of others, including putative moral claims.

If we are moderate libertarians or libertarian consequentialists, and we accept a duty of easy rescue, we can make sense of the apparent difference between acts and omissions. Acts and omissions in general have a different impact on the constriction of an individual’s liberty. Being morally required to perform an act is absolutely liberty constraining – one can only perform one action or one course of action at one time, so the decision to undertake one course of action necessarily precludes many other courses of action. To be a free and autonomous agent is to choose how one is to act to shape one’s own life. Being required to act in a certain way is a severe curtailment of this. However, being morally required to omit to act or undertake a course of actions leaves many other courses of action open. Moral requirements for agents to act are thus more liberty constricting than requirements for them to refrain from acting. The costs, in terms of liberty, of acting are vastly greater than omitting. What is behind the normative difference between acts and omissions are considerations of personal liberty.

For example, consider the well known example of Philippa Foot’s intended to show that there is a moral difference between acts and omissions. There is an apparent moral difference between sending a poisoned food parcel to a person in a poor country and failing to send money which would save the life of a starving person in a poor country. The costs to the liberty of the sender are different in both these cases. If the would-be killer fails to send a poisoned food parcel, there are still many courses of action open to him. He still has great liberty. However, requiring that a person send money is liberty constraining – it requires that a person undertake a course of action to the exclusion of many others. You cannot use the money which you send for a range of other things in your life which you desire or which are good for you. Of course, in this example, failing to send money to save a starving person would be wrong if the amount of money was trivial (e.g. a cent), and the benefits to others significant.

While Kamm’s account focuses on the negative rights of the victim of actions or omissions, this account focuses on the agent and the costs of acting or omitting to act to the agent to determine the moral requirements of that agent. In particular, this account focuses on whether or not there is a duty of easy rescue in a context of a general presumption negative liberty.

This kind of account, call it a duty-based account, contrasts with the Kamm’s more rights-based account. What is wrong with killing is not that it violates a right to life but rather that we have a basic duty not to kill which is grounded in the duty of easy rescue: the cost to the agent of not killing is trivial and the benefit to the victim of not being killed is enormous. According to a rights-based account, while I may gravely harm somebody by winning when he would have won if I had not, I have not wronged him because nobody has a right to win. On a duty-based account, I may well have a duty to let him win if the cost to me is trivial and the benefit to him is enormous. Bill Gates would have a moral obligation to let the poor man win the lottery even if they both had equal and fair chances of winning, just as Bill Gates should (and does) money to benefit others. A duty-based account makes sense of the moral requirement for philanthropy.

A duty-based account also makes sense of the qualifications that are usually added to make a killing wrong. For example, self-defence is permissible because the costs to the agent of forgoing defending oneself are enormous – loss of life. One has no duty to die for another on the moderate libertarian or libertarian consequentialist duty-based account. And, as Kamm elsewhere suggests, the innocent person who finds himself in lethal conflict with you, also has no duty to sacrifice is life to you, and so can use liberty and lethal force to defend himself against you.

A duty-based account also makes sense of euthanasia. When life is full of pain and suffering and is no longer worth living, or when a competent person requests to die, then there is not a cost to the victim in death but rather a benefit. If the costs to the agent of performing active euthanasia are trivial, then there is a duty of easy rescue to perform euthanasia. However, if the costs to the agent of such killing are significant – psychological distress, social ostracism, legal risk – then there can be no duty to perform euthanasia. Indeed, in the only country to legalise euthanasia, the Netherlands, this is precisely the situation – doctors are permitted but not required to perform active euthanasia.

Such an account faces an obvious objection because it relaxes the absolute prohibition on killing of innocent people. Is an agent entitled to kill to protect his negative liberty from a significant but non-lethal threat? Say a kidnapper is about to abduct my children and the only way to prevent him is to kill him. Or a woman is about to be raped and the only way to prevent it is to kill her attacker. Here the cost to the agent of refraining from killing is significant. Morally, we may believe that agents may be entitled to protect themselves in whatever way is necessary from significant incursions of their liberties. Society, or government, or law makers, may choose to create laws to prevent such killings for social reasons to promote the social good. But it is not clear that the agent does wrong in protecting her liberty in these cases. At any rate, when lethal retaliation is justified to non-lethal threats is a complex issue that faces any account of the morality of killing. (This is not to endorse the shooting of would-be burglars. The threat to liberty of a burglary is relatively trivial compared to the loss of life.)

Thus, on the duty-based account, whether it is right or wrong to kill thus turns on the costs of killing and refraining from killing on the agent, and the benefits or harms to the victim.

Table 1. Is there a duty for an agent to undertake course of action x, where x can be an act or an omission to act?
Signficant harm caused
Signficant benefit caused/signficant harm prevented
Small harm caused/harm prevented/benefit
Yes (conflict – self-defence against innocent), if agent liberty preservation significant

There are significant differences between a rights-based account and a duty-based account. To take the standard killing/saving example, a trolley will kill two people unless it is diverted. You can divert it but it will then kill one person who would not otherwise die. If you act to divert it, you will kill one. If you omit to divert it, you allow 2 to die. The rights-based account says that you should not kill one to save two. On the duty-based account, you should divert the trolley and save the two. There may be reasons to do with your life and liberty of acting or omitting that favour one course of action. For example, if you would be sued or suffer grievous psychological harm from diverting the trolley, or if by delaying picking up your children from school would place them at serious risk of child abuse, these would mean the costs to your liberty are too great to require that you save the two. But absent such significant liberty effects, you have a duty to save the greatest number of lives. Once undertaken the duty of easy rescue requires we rescue as many as possible without significant liberty infraction.

There will be some cases in which an omission will be very liberty constraining. For example, imagine I am asked to omit to put in an application for a job to allow another candidate a better chance. Imagine that there are many jobs I could get. In this circumstance, omitting to apply is not very liberty constraining. But if this is the only job that will allow me to be in the city where my children reside, it would be very liberty constraining to omit to apply. However, in general, moral requirements to refrain from acting will be less liberty restricting than requirements to act.

4. Aid
The final section of this chapter deals with Singer’s arguments regarding aid. According to Kamm, Singer holds a stringent principle of aiding: unless something of comparable moral significance is at stake, we should help to relieve suffering.

We should notice that this much stronger than duty of easy rescue I have argued is consistent with libertarian consequentialism and which Singer originally used to motivate support for aid: the drowning pond example. This is essentially a utilitarian position and much stronger than a principle of easy rescue, where the cost to the actor is trivial or small.

Kamm argues that Singer has claimed that this principle “implies that one must bring oneself and one’s family down to a level that if one did any more in order to aid, one would be worse off than those one is trying to help.” She then argues that if one accepts aggregation, as Singer does, there is no reason not to drive oneself even lower (even to death) if it will greatly benefit many people. Such a view would require that we sacrifice our arm to save many lives, which seems counterintuitive. Kamm is right – this is a strict utilitarian conclusion but not one that the moderate, libertarian consequentialism which merely requires a duty of easy rescue is vulnerable to.

Kamm endorses as much when she says

“If it were morally required of people to make large sacrifices in order to prevent all sorts of suffering and death, … their status as beings with a certain type of independence and inviolability would be lost.” (416-7)

That is exactly right. But consequentialists can embrace this view if they only require a duty of easy rescue.

Kamm finishes with some interesting methodological observations about the role of intuitions, which I will not discuss. What I have tried to establish is that a more moderate consequentialism based on a duty of easy rescue, which was at one time espoused by Peter Singer, can address many of the objections which Kamm raises to Singer’s more stringent utilitarian version of consequentialism.

While Kamm may have scored good points against Singer, consequentialism remains as a contender for a plausible, intuitive moral theory.


  1. 1. Posted by S. Matthew Liao | October 4, 2007 2:20 am

    Julian, thanks a lot for your excellent summary and comments! I had a few thoughts regarding them:

    1. On your points against Kamm’s discussion of the Principle of Irrelevant Goods and disability, I don’t think that Kamm would necessarily disagree with you that quality of life could sometimes matter in determining distribution. I think her aim here, as you have pointed out yourself, is only to show that Singer’s maximizing theory has difficulty explaining disability.

    Also, you ask,

    The Principle of Irrelevant Goods is one way to resist the strict Principle of Priority for the Able. However, questions arise: how do we decide whether or when a good is irrelevant in a non-question-begging way? Are all goods irrelevant?

    I think Kamm would answer that it depends on the actual context. As she says just before introducing the Principle of Irrelevant Goods,

    “This is because a difference in outcome could be morally relevant in one context (for example, giving us a reason to prevent or cure disabilities) but morally irrelevant in another context” (p. 409).

    As I understand her, goods will be relevant in some contexts but not others depending on what else is at stake. For example, sore throats are irrelevant when lives are stake, but might be relevant when headaches are at stake.

    2. You said,

    [A short note: I fail to see how “causing the death of” is not necessarily intrinsic to all cases of letting die. If an instance of letting die failed to cause the death of someone, how could it be a case of letting die? It is essentially true of the claim “A let B die” that A causes (in part) the death of B. So I fail to see how “causing the death of” is an exportable property from killing to letting die; it seems (to me at least) intrinsic to both.]

    Consider the following: Suppose someone is drowning; you are passing by, and you fail to throw the person a life saver. Granted you have let the person die. Have you though caused this person to die? Ordinarily it seems that we would say that the cause of this person’s death is the water and not you. That is, it seems that the person would have died even if you had not been there, which suggests at the minimum, that there is another cause of death. If so, wouldn’t this be a case where “causing the death of” is not intrinsic to letting die?

    3. You said,

    On this view, what accounts for the difference between cases of killing and letting die is typically the burdensomeness or degree of help that is required to benefit (or not harm) the person affected by our actions. That is, the degree to which saving or not killing requires we give up our liberty. Typically, refraining from killing someone is costless and saving someone requires sacrifice. It is the degree of sacrifice, not the mere fact of helping, that accounts for intutions about killing and letting die.

    But it does not seem true that burdensomeness always accounts for the difference between cases of killing and letting die. Refraining from killing someone could have very high cost (imagine someone who has had her lover stolen from her); and saving someone could be really easy (e.g, sending a cheque to Oxfam). So your ‘typically’ is doing a lot of work here, and I’m not sure that is true.

    4. Minor point. You said,

    There are significant differences between a rights-based account and a duty-based account. To take the standard killing/saving example, a trolley will kill two people unless it is diverted. You can divert it but it will then kill one person who would not otherwise die. If you act to divert it, you will kill one. If you omit to divert it, you allow 2 to die. The rights-based account says that you should not kill one to save two. On the duty-based account, you should divert the trolley and save the two.

    If, as you say, a rights-based account entails that you should not divert to save the two, then Kamm’s account is not a rights-based one, since she believes that it is permissible to divert the trolley to save the two in these cases.

  2. 2. Posted by Julian Savulescu | October 4, 2007 8:44 am

    Thanks for the helpful responses. I will respond to each point.
    1. i agree that Kamm might respond that it is context dependent, but she needs to give how an account of what disabilities would count as relevant in life-saving situations. She gives no indication that she accepts any (in this chapter) and no principled way of dividing them. It would also be inadequate just to intuit on a case by case basis whether a disability was relevant or irrelevant to life-saving.
    2. You claim if I let a child die in a pond I am passing by, I do not cause the child’s death. But this would mean in no case of letting die would I cause the individual’s death. Say I am a doctor and I fail to resuscitate a patient. My role in the causal chain is exactly the same, but that is a paradigmatic case of “letting die”.
    Some people have mistakenly packed into the concept of letting die normative considerations, for example, that the doctor lets the patient die because he has a responsibility or duty to the patient. But Ingmar Persson and I have resisted this normative account of killing and letting die, arguing we should go for a purely non-normative account of killing/letting die. (Savulescu, J Persson I, (2005) McMahan & withdrawal of life-prolonging aid Philosophical Books Vol 46 (1): 11-22 Jan DOI: 10.1111/j.1468-0149.2005.0355c.x)If we do that, all cases of letting die will have the same relevant pattern of causation. If I did not cause (in part – as Kamm says) through my inaction the death, it would not be a case of letting die. If I have a part in the causal network of someone’s death, it would be a case of letting-not-die.
    3. I think I made the point that in some cases refraining from killing may have very costs, which speaks in favour of killing. It is just that our ordinary intuitions are grounded in teh fact that it is much easier to kill (say 100 people) than it is to save (say 100 lives). Toby Ord might disagree with this because it is easier now through donation of small sums of money to save a life, but through human history, killing has been easier for humans than saving lives.
    4. Yes, that was a mistake. Kamm does indeed say you can divert the trolley.

  3. 3. Posted by Fiona Woollard | October 4, 2007 8:48 am

    Julian and Matthew: I’ve very much enjoyed the thread so far. I have a comment on the question of letting die and causing death:

    Kamm’s example of a case of letting die where the agent does cause death involves A removing B from a life support machine. This suggests to me that Kamm is using something like the notion of agent causation suggested by Kadri Vihvelin and Terrance Tomkow (The Dif, Journal of Philosophy, 2005): an agent causes a person’s death if the agent’s basic actions cause the death.

    In the life support case, however detailed a description we give of the agent’s actions, the occurence of these actions will still be a cause of death. However, in the life-saver case Matthew described above, even if we think that “the agent’s failing to throw the life saver” is a cause of death, perhaps we should not think that “the agent’s leaning on the bridge” is a cause of death – for the death would have occured even if A had performed a cartwheel, jumped up and down, or performed many other actions apart from leaning on the bridge. (Vihvelien and Tomkow refine this account to avoid overdetermination counterexamples but I won’t go into those details here.)

    So we might say that while it is an intrinsic property of both killing and letting die that some aspect of an agent’s behaviour causes death, it is only killing that has the agent causing death as an intrinsic property.

  4. 4. Posted by Julian Savulescu | October 4, 2007 9:12 am

    Dear Fiona,
    I think you are right. However, Ingmar Persson and I argue in the paper mentioned in my previous comment, that this is a mistake to describe withdrawal of life-supporting aid as letting die. It is killing. We argue that all withdrawal of life-supporting aid is killing. All abortions and removal of life-prolonging medical treatment also constitute killings. This of course leaves open whether they are permissible or impermissible killings but it is an example of the mistake of building in normative considerations into the metaphysics of killing and letting die to describe an action to withdraw life-sustaining aid as letting die.

  5. 5. Posted by Guy Kahane | October 6, 2007 4:22 pm

    Just a point on that Principle of Irrelevant Goods. You might think that this is a non-consequentialist principle. But whether it is depends on what it is a consequentialist is trying to maximise/promote. If the object of promotion simply is value, and we admit that sore throats are goods, then of course we get counter-intuitive conclusions. But suppose the the aim (at least in this context) is to promote happiness or well-being. And you might think that beyond a certain threshold (or other condition) further benefits don’t further promote happiness. So while still goods, they are not the goods you’re aiming to promote. And I think that to some extent something like this view best explains our intuitions. We are reluctant to explain them in clear non-consequentialist terms: ‘yes, the outcome would be better, but it would be wrong to do that’. It rather seems that, from this particular perspective, these benefits do not really matter.

  6. 6. Posted by JP Sevilla | October 17, 2007 10:07 pm

    Dear Julian,

    I have not posted a comment on this blog before and I hope this one is not unwelcome. My two points concern what a reductionist about mind would make of the distinction between killing and letting die.

    1. One might be a reductionist who believes that mind has no causal efficacy over and above that of the underlying mechanical brain (for how could it without breaking the second law of thermodynamics?). If so, one would conclude that regardless of what kind of distinction we could make between killing or letting die, no one is ultimately more morally responsible (or ultimately more morally deserving of punishment or blame) for killing than for letting die. That is, no distinction between killing or letting die could have any differential implications for ultimate moral responsibility or desert.

    2. For such reductionists, there could be other potentially relevant differences between killing and letting die, and these could be moral differences. Such reductionists hold that actions are events. Killing and letting die are different events. They are constituted, for example, by different mechanical events in the neuroanatomy of the brain. The former may involve stronger engagement of the brain’s reward and sensorimotor systems than the latter, perhaps. Since outcomes are constituted by events, and actions are events, then differences in action constitute differences in outcome in a very natural and unforced way that I expect consequentialists should easily agree with. The outcome difference between a world in which a killing occurs and a world in which a death was not prevented consists at least of the difference in histories of neurobiological mechanical events between those worlds. Maybe upon closer inspection, we might conclude that these differences don’t matter morally, but we shouldn’t begin by assuming that they don’t. And wouldn’t most consequentialists say that all events matter?

Post a comment

Name: (required)

Email Address: (required) (will not be published)



(Spamcheck Enabled)

This work is licensed under a Attribution-NonCommercial-NoDerivs 3.0.