Non-consequentialism

Introduce the canonical pair of ethical dilemmas for our discussion:

Phillipa Foot’s trolley problem: At a minimum, most agree that you may pull the lever; that is, it would not be morally wrong to do so.

Transplant case: Everyone would agree that it is not permissible for you to proceed in killing the healthy man. Even consequentialists, such as utilitarians, deny that it would be the right thing to do and scramble to amend their theories in order to justify why.

It is unclear what differs the trolley case from the transplant case, when in both you are killing one to save five.

Introduce the DDE:

Non-consequentialists typically accept some constraint on what we may do to non-consenting people in pursuit of the good.

One of the most commonly proposed constraints is a prohibition against intending harm to others. This is grounded on the distinction between intending and foreseeing harm, and is famously articulated through the Doctrine of the Double Effect (DDE). This doctrine states that there is a moral constraint on intending evil, even when the evil will be a means to a greater good. However, we may be permitted to employ neutral or good means to pursue a greater good even though we foresee evil side effects. This holds as long as the good is proportionate to the evil and there is no better way to achieve this good.

Explain the two bomber cases:

The DDE's canonical illustration is the pair of bomber cases.

In Strategic Bomber, we bomb a munitions factory in an enemy state, in order to bring about the end of a war. However, the factory is situated in such a way that some civilians in a nearby children’s hospital will be killed.

In Terror Bomber case, the pilot bombs the children’s hospital directly in order to terrorize the enemy state into accepting a peace agreement.

Why is the DDE successful?

The DDE yields that the pilot in the first case may bomb the factory because the death of the children is a foreseen but unintended consequence. The intention is to end the war. By contrast, the pilot in the terror bomber may not bomb the hospital because the deaths of the children are an intended consequence.

Applied to the Trolley and Transplant cases, the DDE again seems to deliver correct verdicts. It is one thing to steer a trolley towards someone, foreseeing but not intending that you will kill him, and a very different thing to aim at someone’s death as part of your plan to save five other patients.

What is the first objection to the DDE?

However, it is difficult to formulate the doctrine in such a way that it avoids unnatural applications.

For example, the Terror Bomber might act intending only that the civilians appear dead until peace is declared. Of course, he foresees with certainty that the civilians will die but if by some miracle they came back to life after the war was over, he would not object. Under this interpretation, the DDE may see his act as permissible. The doctrine's verdict on the case therefore flips absurdly.

What is Quinn’s response to this objection?

Warren Quinn responds by reformulating the distinction as one between direct and indirect harmful agency.

The Tactical bomber undeniably intends that civilians be involved in a certain explosion precisely because their involvement in it serves his goal. He may not necessarily, as we have seen, intend their deaths. However, his purpose requires that they be violently impacted by the explosion of his bombs.

The tactical bomber, on the other hand, intends an explosion but not in order that any civilians be affected by it. He may foresee with certainty that his bombs will kill many of them, but he can honestly deny that their involvement in the explosion has anything to do with his purpose.

Explain direct harmful agency:

The person who harms by direct agency therefore takes up a distinctive attitude toward his victims. He must treat them as if they were there for his purposes.

Indirect harmful agency may treat the harm or deaths of agents as unimportant, and this is morally questionable. However, in cases of direct agency there is additional morally grievous weight that the victims are made to play a role in the service of the agent's goal that is not morally required of them. They have a right not to be pressed into the service of other people’s purposes and, though this right may be justifiably infringed in certain situations, in all cases they add their burden to the opposing moral argument.

So what is the DDE saying?

The DDE does not suggest that it does not matter what you bring about as long as you merely foresee and do not strictly intend the evil that follows. What it does say is that we need, other things equal, a stronger case to justify harmful direct agency than to justify equally harmful indirect agency, and that this distinction may sometimes affect the permissibility of an action.

What deeper criticism does this show?

It simply seems irrational that a state of mind, such as intending, could somehow change the permissibility or impermissibility of an action. Intending may well affect the moral worth of an action – whether the agent should be praised or blamed for the act – but not moral desirability. Thompson suggests that the doctrine fails to take into account this distinction.

Suppose that a doctor dislikes her patient and administers a lethal dose of pain killer, relishing the thought that this will be the last of him, but that the earlier death is in fact better for the patient. If she is moved by such reasons then she is a morally bad person. However, it does not follow that it is impermissible for her to administer the drug, or that the patient should have to wait until a different doctor – with better intentions – comes on duty.

What is another example to undermine intention?

Imagine Alfred who wishes to speed up the death of his dying wife. He buys a liquid, thinking that it is poison. However, it is actually the only existing cure for his wife’s illness. Is it permissible for Alfred to give the liquid to her, even with the intention of killing her? Surely yes. If it were not permissible for Alfred to give it to her with the wrongful intention, then morality would needlessly condemn her to die. Thus, we can conclude that, if an action is permissible, it should be permissible regardless of the agent’s intentions.

Instead of focusing on whether an agent intends a particular outcome, we should focus on the objective factors which make the action permissible or impermissible. The DDE’s focus on intention mistakenly shifts the focus from the objective wrongness of an action to the subjective state of mind of the actor.

What is one final criticism of intention?

Finally, any constraint centring on intention seems to forbid acts of self-defence because harming an attacker is intended as a means of self-preservation, and also self-sacrificial acts – such as jumping on a grenade to save your comrades - because the harm to oneself is intended as a means of benefiting others.

What is the first defense of the DDE?

Fitzpatrick suggests that critics fundamentally misinterpret the DDE. None of their objections hold because the DDE does not ever link moral permissibility of an act with the intentions with which an agent acts. Properly understood, it says that an action is permissible just in case there exists a justification for it in terms of a sufficiently worthy end that can be pursued without intending anything illicit as a means.

How does this solve our issues?

If we allow that there is a sufficiently good end (ending the patient’s suffering) to justify giving the morphine dose in question as a means without intending anything illicit, then the act is permissible.

Similarly, even if the particular pilot sent to bomb the munitions plant may also harbour the intention of killing innocent civilians, the fact remains that there exists a sufficient justification for bombing the munitions plant simply in terms of its military significance, such that one could act for the justifying end without intending any deaths.

This is sufficient to declare the act morally permissible as far as intention is concerned, and this permissibility is not further affected by the actual intentions of particular agents. This interpretation of the DDE avoids the charges of absurdity.

Further defense of intention:

Furthermore, critics of intention attempt to provide alternative explanations for moral intuitions that have traditionally been explained the relevance of intention. However, these explanations are offered on a case-by-case basis.

For example, with the cases of SB and TB, critics such as Scanlon and Thompson bring in concepts like military advantage or suggest that the requirements of morality alter during conditions of war. But this does nothing to explain why, for example, you may divert the trolley but not push a fat man in front of the trolley.

Ad hoc alternative explanations of this kind do not provide the same degree of explanatory power compared to intention which provides a unified and relatively simple explanation for a wide range of moral intuitions.

What is the new objection to the DDE?

Imagine now that the track in the trolley loops around. Therefore, if the trolley diverts onto the side-track it would simply loop back to kill the five, if not for the one large person whose body would stop the trolley. Now, the death of the one is not only a foreseen and unintended side-effect of saving the five but an intended means to save them.

The answer in this case may be less clear than in the original but it is quite plausible to maintain that if it is permissible to turn the trolley in the first case, then it is permissible to do so in the second case as well, even though it is an intentional killing. 

Defense of DDE to loop:

It is certainly true that the DDE, or related ideas about intention, explain many of our intuitions about certain cases. The rare exceptions do not seem intuitively powerful enough to erode our expectation that a general principle of some kind is at work here.

Thompson asserts that it is clearly permissible to turn the trolley in the Loop-variation. However, at least for me personally, I can’t say that I have particularly strong intuitions about the case. If a theory, such as DDE, tells us that it is impermissible to turn the trolley in Loop we should not regard this as an overwhelming reason to reject the theory.

Conclusion on DDE:

Quinn's observation that direct agency involves a distinctive use of persons captures something real. However, the feature it tracks is not the agent's intention; it is the structural role the victim plays in the act. Shifting from intention to structural role preserves the cases the doctrine was designed for while avoiding the Alfred-type counterexamples. So the DDE fails on its own terms - since the thesis that actual intentions matter for permissibility cannot be sustained - but leaves behind a serviceable successor where some acts have an objectionable structure, since they can succeed only by using persons as means.

What is there to say about the DDE and doing vs allowing?

Even on the most favourable construction, the DDE is not the fundamental non-consequentialist principle. Foot argued, I think rightly, that our judgements about the Trolley pair and its cousins are not driven primarily by intention.

Suppose I use poisonous gas to kill mosquitoes, foreseeing that the gas will kill my neighbour. Now, suppose the gas does not threaten my neighbour, but I could have donated the money for the gas to famine relief and saved an individual’s life. The magnitudes of good and harm are identical, and in neither case do I intend the death. Yet they are not morally equivalent.

What distinguishes them is that in the first I do harm while in the second I merely allow it. If this is right, then the distinction between doing and allowing - not the distinction between intending and foreseeing - does the fundamental work. The DDE captures a further consideration that sometimes matters on top of that distinction, but it cannot stand alone as the primary non-consequentialist constraint.

Introduce the distinction between doing and allowing:

In some cases, we intuitively recognise a clear contrast between doing and allowing harm.  We allow thousands of people to die of starvation in developing countries each year, but most of us think this is, at worst, a failure of charity. Yet sending those people poisoned food would be murder, even though in both scenarios our action (or inaction) is linked to their deaths. Similarly, it seems worse actively to push someone off a bridge than to refuse to rescue them from drowning, even where the harm foreseen is identical.

What suggests there is no distinction?

Yet, in other cases, killing and letting die can be just as bad as each other. James Rachels famously denies that any such distinction exists, with his bathtub cases providing the canonical challenge.  In one, you drown your small cousin to get the inheritance and in the other you find that he has hit his head and is already drowning so you simply watch as he dies. Here, there seems to be no great moral difference here between killing and letting die.

What is Quinn’s response to Rachels?

Warren Quinn's response is that the argument conflates two different claims: that killing is intrinsically worse than letting die, and that the constraint against killing is stronger than the constraint against letting die. The DDA needs only the second.

A constraint is stronger when it is harder to defeat - when it takes a weightier countervailing consideration to override. The fact that, in certain highly motivated and malicious cases, violating the weaker constraint is just as bad as violating the stronger one is no evidence against the difference in strength.

What is a simple explanation of the DDA?

A simple explanation is that doing and allowing harm run up against different kinds of rights, one of which is stronger in the sense that it is less easily defeated. Negative rights are claim rights against harmful intervention whereas positive rights are claim rights to aid or protection.

Negative rights are, ceteris paribus, harder to override than positive rights. It must be this way round or else we might permit killing two to save one, which seems bizarre. Strict negative duties clearly take precedence over even the strictest duties of positive aid. For example, it is clearly not permissible to commit murder in order to bring one’s starving children food.

The justification for this precedence is a complex one but boils down to this reality: the sense in which your own life is truly yours can only be provided by negative rights. Warren Quinn and, more recently, Fiona Woollard explain why

Explain Quinn and Woollard’s theory:

A person is constituted by her body and mind. For these to be hers in any substantive sense, she must have privileged normative status with respect to them: she must be able to refuse their use by others, to direct them towards her own ends, to count their loss as her own loss rather than as a cost society may offset against some greater aggregate good. If morality treated my body as a resource that could be deployed whenever doing so would maximise welfare, then my body would effectively be common property, and my existence as an independent agent would be undermined.

Continue the Quinn/Woollard argument:

Crucially, protection against unwanted effects is not enough for a person’s body to genuinely belong to him. Ownership, even ordinary material ownership, is not merely a right to possess but a right to use. If I can have something taken from me whenever someone else's use of it would be more efficient, I do not own it but I am a caretaker. The same applies, more intensely, to my body and life. If I am required to sacrifice myself whenever a consequentialist calculation favours it, I do not really have the privileged status over myself that genuine ownership requires.

What are the two ways ownership can be intruded on?

Such ownership can be intruded upon in two ways. Causal imposition occurs when another's behaviour intrudes into my sphere, substantially affecting what belongs to me – for example, when they harm my body. Normative imposition occurs when another's needs generate demands on my body or resources, requiring me to place them at her disposal. For something genuinely to belong to me, I need normative protection against both kinds of imposition. The DDA provides exactly this protection. The stronger constraint against doing harm resists causal imposition; the weaker status of positive duties resists unlimited normative imposition.

Why is this asymmetry needed?

Without the asymmetry, one of these forms of protection would collapse. If doing harm were no harder to justify than allowing it, then a sufficient aggregation of positive claims on my body would license causal imposition upon me; I could be killed to save several others. If allowing harm were no easier to justify than doing it, then every failure to rescue would be murder, and each of us would be obligated to ruinous sacrifice in response to the global aggregate of need. In either case, the normative structure that makes my life mine would dissolve.

Conclusion on Quinn-Woollard:

This is not merely an appeal to a general intuition about rights. It is an argument from the structural conditions under which persons can be independent moral beings at all.

The Quinn–Woollard argument does not entail that doing harm is always impermissible, nor that allowing harm is always permitted. What it establishes is a structural asymmetry, which has several consequences worth noting.

First objection, and response:

Firstly, some suggest that by allowing impositions in extreme cases, the argument of ownership collapses. However, small impositions for large gains need not violate the doctrine. I can still count as owning my car even if I am required to move it for an ambulance. Similarly, my authority over my body remains meaningful if I am required to bear minor costs to prevent serious harm to others. The DDA is not a licence for selfish refusal to help.

What is there to say about the level of precedence of negative over positive rights?

The precedence of negative over positive rights must be significantly, though not infinitely, strong. A weak thesis - that negative rights prevail when the goods at stake are equal - would permit killing one to save two, which is precisely the sort of aggregative trade that the ownership argument rules out. A very strong thesis - that certain negative rights are absolute whatever is at stake - may be implausibly rigid in the face of genuine catastrophe (would it really be wrong to kill one to save a million?).

An intermediate view, on which negative rights have considerable weight but eventually give way under sufficient pressure, is most defensible. Precisely where the threshold lies is a difficult question, and Woollard argues, in broadly Aristotelian fashion, that no algorithm will fix it; the matter must be left to moral perception. This is a limitation, but it is the same limitation faced by any plausible non-consequentialist view and is no special embarrassment for the DDA.

Is DDA the only relevant principle?

Third, the doctrine leaves room for other principles. It is matters whether a positive duty is strict (feeding one's own children) or a matter of beneficence (feeding distant strangers). An intentional and malicious letting die may seem worse than a killing which is the side effect of a benevolently motivated action. Similarly, rights of competition, of punishment, and of self-defence all legitimise forms of harmful positive agency that the DDA alone would appear to prohibit. The doctrine operates as one major prima facie principle among others, not as a complete moral theory. Any ethical theory that attempts to systematize all these elements will have to be quite complex, and will not be captured within simple principles.

How does Kamm explain why we must not pull the lever?

Francis Kamm provides a similar explanation to Woollard but frames it through the high inviolability of persons: the respect which we must give to each individual and their right not to be harmed.

It may be true that more people will get seriously violated if we do not kill the one person. However, this is acceptable because it does not undermine any of those people’s inviolability; after all, inviolability is a status that defines what we can permissibly do to people and not what actually happens to them. If it were permissible to kill the one person to save five then the inviolability of all six would be lower and this is worse.

The fact that the others have this same inviolable right does not diminish the constraint against violating the one’s right that I come up against. Thus, when we truly understand the inviolability of rights, one should not pull the lever.

What does Singer say about ethics and moral intuitions?

Every time a seemingly plausible justifying principle has been suggested, other philosophers produce variants on the original pair of cases that show that the suggested principle does not succeed in justifying our intuitive responses.

However, Singer argues that moral intuitions, often seen as the basis for ethical judgments, are unreliable due to their evolutionary origins and emotional basis.

What do brain scans show?

Greene proves, using brain scans, that people have a robust, negative emotional response to the personal violation proposed in the footbridge case, so they immediately conclude it is wrong.

On the other hand, people fail to have a strong negative emotional response to the relatively impersonal violation proposed in the original trolley case, and therefore revert to the most obvious moral principle “minimize harm.”

Those who concluded that it would be right to push the stranger off the footbridge took much longer on average to form their judgement, and displayed more activity in the parts of the brain associated with cognitive activity.

What is there to say about evolution?

Furthermore, our intuitions are shaped by the conditions in which our ancestors lived, and therefore may not be relevant to modern ethical dilemmas.

Singer points out that throwing a switch that diverts a trolley bears no resemblance to anything likely to have happened in the circumstances in which our ancestors lived. Hence the thought of doing it does not elicit the same emotional response as pushing someone off a bridge: a simple example of close, personal violence. This evolutionary reality holds no moral salience. Singer concludes that there is likely no morally relevant distinction between the cases.

What is projective separating?

Projective separating leads to inconsistent ethical judgements by causing us to view a problem as belonging only to those in a specific situation, and not to others. We consider it badly wrong to spare people the harms of “their” problem by imposing serious harms on others, even when the losses imposed, or the number of people affected are much smaller.

What are examples of projective separating?

This is demonstrated in a variant of the trolley problem where you can divert the trolley, but it will roll down a hill, cross a road and kill a sunbather relaxing in the park. We intuitively see this as wrong because the victim appears to be in a separate situation, and we view the problem posed by the trolley as a problem only for those on the track.

However, if a rogue missile is rerouted from the five to the sunbather, this action is clearly seen as acceptable because we do not view the missile as strongly associated with any one person.

The different judgements reflect how we assign people to situations, and problems to people. Unger argues that this assignment has no moral significance.

What are two other distortions?

Unger details numerous other morally empty principles which distort our thinking such as the notion that when serious loss will result, it is harder to justify moving a person into an object than it is to move the object into the person.

In the lever case there is a quite a fair amount of psychological distancing between pulling the lever and the other end of the causal chain where the one man gets killed. There is much less distancing when you push the fat man off the bridge.

What is there to say about norms?

When we make ethical judgements, we make an implicit reference to a set of norms. This is why we may say, for example, that a parent who deliberately fails to feed their children is guilty of killing their offspring, but one who merely fails to provide food to a stranger merely allows their death, as this action falls within the norm.

Conclusion on intuitions:

When we consider ethically intriguing examples, we are influenced by a highly complex combination of factors. Unfortunately, the factors which have the most psychological influence often also contain the least ethical significance. Thus, our ethical theories should not seek to try and capture our moral intuitions as this gives too much weight to potentially biased and unreliable intuitions.