The case that we are considering might seem to present a simple choice: operate and we will save one baby; fail to do so, and both will die. So it might seem obvious that we both can and should operate. But this answer assumes that there is no significant difference between killing and letting die. If the difference between killing and letting die were significant, then the question would be cast in a very different light. For in operating we will, it seems, be intentionally killing one of the babies to save the other; and that, it might be argued, is worse than letting both die.

In this essay I argue that we can and should operate. I assess two different arguments for this conclusion. The first concedes that there is a distinction between killing and letting die, but holds that operating to separate the babies is not a case of intentionally killing the weaker. The argument employs what is known as the 'doctrine of double effect'. It may be true that the weaker baby will die as a result of what we do; but this does not amount to intentionally killing it, since this is not what we are aiming for. I contend, however, that this first argument is unsuccessful.

The second argument holds that no morally significant distinction can be made between killing and letting die. This argument, I contend, is a good one. Indeed, the very considerations that counted against the employment of the doctrine of double effect, count in its favour.

I take the two arguments one at time.

The Doctrine of Double Effect

The doctrine of double effect holds that it is sometimes permissible to bring about some outcome as an unintended but foreseen consequence of an action, whilst it would not be permissible to intentionally bring about that outcome. It the case at issue, it might be argued, we do not intentionally kill the weaker baby; this is shown by the fact that if we could operate in some way that saved them both, we certainly would. So the weaker baby's death is merely a foreseen but unintended consequence of acting to save the stronger baby. Since that action is a good one, the unintended consequence is acceptable.

The problem with this argument is that it makes the acceptability of an action far too dependent on a individual's state of mind. Suppose that in the course a robbery, a criminal shoots at a policeman and injures him. It might be true that the criminal didn't intend the policeman's injury, in the sense that if he could have stopped him without shooting him (by magically whisking him away to Mars, for instance), he would have done so. But we would not think of that as a defence; what the criminal would have done in this counter-factual state of affairs is irrelevant. Indeed, in the normal sense we would say that he intended the policeman's injury, since it was so bound up with shooting him as to be inseparable from it. Similarly in the baby case, the action of separating the twins is so bound up with killing the weaker as to be inseparable from it. In cutting the arteries that take blood from the functioning heart to the weaker twin, we are intentionally killing her. What this shows, I think, is that what is important in evaluating an action is what we knowingly bring about, not what we intend, in some highly theoretical sense of 'intend'.

Killing and Letting Die

I will now bring the conclusion of the last section to the distinction between killing and letting die. It seems to me that we knowingly bring about a situation just as much when we let it happen (given that we could intervene) as when we actively engineer it, since there is no clear difference between the two. To see this, I will consider a version of the 'trolley case'. Suppose a trolley is hurtling down the main line to a point at which two people have fallen onto the track. You could pull a lever and divert it onto a siding. But on that siding another person has fallen. Should you pull the lever? I say that you should: that way one person dies rather than two. The advocate of the distinction between killing and letting die will disagree. But to them I say: suppose that the reason the trolley will continue on the main line is because you are actively holding open the lever that makes it do so; releasing the lever and it will spring back on its own, resulting in the trolley going into the siding. Now is it impermissible to let go of the lever (is this a doing: 'a releasing'; or a letting happen: 'letting the lever go back to its natural place')? Or suppose that the lever is getting harder and harder to hold, so that you have to put in more and more effort: now is letting go of it a 'doing' or a 'letting happen'? I conclude that there is no real distinction between an act of doing and an act of letting happen, and so our moral evaluations cannot rest up it. What matters is what we knowingly bring about, whether by doing or by allowing. So, since it is clearly better to knowingly bring about the death of only one twin than of both, we should operate.