Could it be that one morally ought to do something morally bad? Some people think the answer is obviously ‘No’. Indeed, these theorists may say, it is contradictory to suppose that one morally ought to do something morally bad. Others hold that it is not a contradiction but a sad fact of life that one may be morally required to do something morally bad. This latter position is the one I'll be supporting. If it's the right view, it really matters in practical affairs. For example, almost everyone would agree that it's morally bad to kill an innocent fetus. But this does not settle the question of whether one morally ought to have an abortion, if it can be morally required that one do something morally bad.

Here is a classic example.

In Sophie's Choice by William Styron, Sophie is required by a guard in the concentration camp in which she is interned with her two children to select one of them to be killed. If she refuses to choose, both will be killed. By choosing one child for death, Sophie saves the other. The same act is both saving a life and causing a death; it is both morally required and morally bad.

Let's call such cases moral dilemmas. They are cases in which there is a particular kind of moral conflict: every action available to the agent involves something morally bad, and at least one of the available options morally ought to be done. Thus defined, it is controversial whether there really are moral dilemmas. There certainly seem to be: the captain of the lifeboat might do nothing to save the people in his charge, which would be bad because then everyone would die, or he can lighten it, which would be good and so ought to be done, by sending innocent people to their death, which would be bad. Sophie might do nothing, and so fail to save a child, which would be bad, or she might save one child, which would be good and so ought to be done, and thereby ensure the death of the other, which would be bad. The question we'll discuss is whether moral dilemmas constitute paradoxes, and in particular whether a moral dilemma involves a contradiction. If it does, we would have to conclude that there cannot be moral dilemmas (assuming that moral reality is consistent). There may seem to be such cases, but closer inspection should reveal that they are not what they seem.

Here's an informal argument for contradiction. We adopt the principle

In a dilemma, all available options are morally bad and so, by (O), one ought not to do any one of them. But, by the definition, there is an option which I ought to take. So this option is one I both ought to take and ought not to take.

This might at first seem like a contradiction, but it is not. It is a case of conflicting obligations. A contradiction would take the form: I ought to do a certain action, and it is not the case that I ought to do it. If we could move from ‘I ought not to do A’ to ‘It is not the case that I ought to do A’ then we would have a contradiction; but it is far from obvious that this move is correct. It would certainly be incorrect in structurally similar cases. For example, I cannot move from ‘It's possible that it won't rain today’ to ‘It's not possible that it will rain today’. It's also incorrect in the case of desires and preferences, which one might suppose to have a similar logic to obligation. It is commonplace to have conflicting preferences: I'd prefer not to go to the dentist (it takes time and is disagreeable), but I'd prefer to go (to avoid worse problems from dental neglect). There could not be anything paradoxical or contradictory about such a set of preferences. The rational person checks to see which one matters most, and acts accordingly. But if we could move from ‘I'd prefer not to go’ to ‘It's not the case that I'd prefer to go’ this commonplace situation would be contradictory, and so impossible.

A little formalism will make it easier to follow the reasoning. By definition, if there are moral dilemmas there is an action, A, which one both ought and ought not to do:

(‘O’ abbreviates ‘ought’, ‘&’ abbreviates ‘and’, ‘¬’ abbreviates ‘not’.) To have a contradiction we would need something different:

So far, the only way we have contemplated reaching a contradiction is by the principle:

Footnote ¹

The examples of supposed dilemmas make this principle one which we should not accept without argument. That the captain is obliged not to throw any passengers overboard does not obviously entail that he is not obliged to throw any overboard: his duty to save as many people as possible seems precisely to oblige him to throw some overboard. However, there are other ways in which a contradiction might be reached.

The key feature of a dilemma is that O(A) & O(¬A). It is natural to think that obligations ‘agglomerate’, as expressed by the following principle:

Applying this to a dilemma yields O(A & ¬A). It has also often been held that one can be obliged to do only what one can do, that ‘ought implies can’:

(‘C’ abbreviates ‘can’.) Applying (CAN) to O(A & ¬A) yields:

But this is obviously false: one cannot do both of two inconsistent actions (we are assuming that the context is held constant). Indeed, its negation is an obvious truth:

Now we have a contradiction. And thus a paradox: dilemmas seem possible, but they are contradictory, and so impossible.

The principles which lead to this result are open to challenge. Indeed, they are in tension with one another, if the reality of dilemmas is granted. Given (CAN), a dilemma situation is precisely one in which obligations don't agglomerate. And given (AGG), a dilemma situation is precisely one in which one cannot do everything one ought. But are there any reasons independent of moral dilemmas for denying (CAN) or (AGG)?

Against (CAN), it could be suggested that it is possible to be under an obligation one cannot fulfill. For example, a drunken driver ought to negotiate the curve accurately, even if his condition makes it impossible for him to accomplish this. An addict ought to stop taking the substance, even if he cannot give it up. The attraction of ‘ought implies can’ (the argument might continue) is that in many cases in which the ability is lacking, we excuse failure to act in accord with the obligation. But this does not mean that the obligation is absent or somehow nullified. Maybe the agent is excused; but this would make no sense if there were no obligation for him to be excused from.

Against (AGG), it might be argued that there is a maximum total degree of obligation that can be imposed on a single agent. If actions A and B are each very demanding, it seems coherent to hold that each is obligatory whereas their conjunction is not. For example, perhaps I ought to give large amounts of money to charity. Perhaps I also ought to give large amounts of money to provide deluxe care for my aging mother. But I cannot do both.

Moral philosophers debate whether moral dilemmas are genuinely possible. My own view is that they are. They seem to me merely a reflection of the fact that sometimes we have to choose the lesser evil. The lesser evil is still an evil, even though chosen after the most careful and high-minded moral deliberation. The best thing to do may still be a bad thing.

In some cases of competing obligations, one obligation may abolish or annul another. The Dutch people who protected Anne Frank and her family from the Gestapo had to lie to do so. Perhaps they felt conflicting obligations: they ought not to lie, but they also ought not to betray the Jews in the attic, who would be sent to the death camps if their presence became known to the Gestapo. In this case, it seems that the obligation to protect the Jews simply annuls any general obligation to tell the truth. Indeed, the case could reasonably be taken as showing that there is no universal obligation to tell the truth: you ought not to tell it in Anne Frank type situations. It was not a dilemma, because the obligation to save lives simply annulled any obligation to tell the truth. This contrasts sharply with the situation of the boat's captain. We cannot say that the obligation not to kill the innocent was somehow annulled in his situation. All we can say, in my opinion, is that the obligation was one that had to be ignored if the captain was to act for the best. To do the best thing, he had to do something bad. There would be something wrong with the captain if he did not think that what he had to do was simply terrible.

One source of resistance to the view that there are genuine dilemmas is the belief that a person of good conscience can never be in a position in which she has to do something bad. If she morally has to do something, it cannot (in the circumstances) be bad. Otherwise people could fall from moral perfection for reasons entirely beyond their control; and this could not be so. The underlying thought is, I think, a kind of wishful thinking about the moral life, the belief that, given all one's faculties, being morally perfect is a goal that no outside circumstances could prevent one from attaining. Sadly, good and evil sometimes come inextricably entwined, and unless we are lucky we may find ourselves having to do something which is bad (bad even in those very circumstances) in order to act for the best.

If this is the right view, there are important practical consequences. Moralists are often happy to end their discussion with a conclusion of the form ‘Actions of type A are morally bad, and so ought not to be performed’. A conspicuous example is the debate over abortion. Suppose a pregnancy poses a genuine moral dilemma. Even the person who believes that abortion is morally bad can consistently believe that it is morally the best thing to do under the circumstances. The conclusion that it is bad can be accepted by all parties to the debate about how it is morally best to act. Mere belief in its badness is not enough to guarantee that those acting from the most carefully considered moral scruples will not properly regard an abortion as the morally best course in certain circumstances. If the pregnancy is a dilemma, it may be that something morally bad has to be done for the morally best thing to be done.

Moral dilemmas (as defined here) are cases in which obligations conflict: there is some action which both ought to be done and ought not to be done. If this entailed that there is some action of which it is true both that it ought to be done and that it's not the case that it ought to be done, moral dilemmas would be contradictory and so impossible. Our paradox would then be that they seem to be possible, while being impossible. My suggestion is that moral dilemmas are not really impossible, and so the paradoxical contradiction is avoided.