The Trolley Problem / Das Trolley-Problem (Englisch/Deutsch). Judith Jarvis Thomson
order to get something that threatens five to threaten the one, then he may proceed.
Where what is at work is a matter of degree, it should be no surprise that there are borderline cases, on which people disagree. I confess to having been greatly surprised, [74]however, at the fact of disagreement on the following variant on Bystander at the Switch:
The five on the straight track are regular track workmen. The right-hand track is a dead end, unused in ten years. The Mayor, representing the City, has set out picnic tables on it, and invited the convalescents at the nearby City Hospital to have their meals there, guaranteeing them that no trolleys will ever, for any reason, be turned onto that track. The one on the right-hand track is a convalescent having his lunch there; it would never have occurred to him [1412] to do so if the Mayor had not issued his invitation and guarantee. The Mayor was out for a walk; he now stands by the switch.14
For the Mayor to get the trolley to threaten the one instead of the five, he must turn the trolley onto the right-hand [76]track; but the one has a right against the Mayor that he not turn the trolley onto the right-hand track – a right generated by an official promise, which was then relied on by the one. (Contrast the original Bystander at the Switch, in which the one had no such right.) My own feeling is that it is plain the Mayor may not proceed. To my great surprise, I find that some people think he may. I conclude they think the right less stringent than I do.
In any case, that distributive exemption that I spoke of earlier is very conservative. It permits intervention into the world to get an object that already threatens death to those many to instead threaten death to these few, but only by acts that are not themselves gross impingements on the few. That is, the intervenor must not use means that infringe stringent rights of the few in order to get his distributive intention carried out.
It could of course be argued that the fact that the bystander of the original Bystander at the Switch makes threaten the one what already threatens the five, and does so by means that do not themselves constitute infringements of any right of the one’s (not even a trivial right of the one’s), shows that the bystander in that case infringes no right of the one’s at all. That is, it could be argued that we have here that independent ground for saying that the bystander does not infringe the one’s right to life which I said would be [78]needed by a friend of (ii).15 But I see nothing to be gained by taking this line, for I see nothing to be gained by supposing it never permissible to infringe a right; and something is lost by taking this line, namely the possibility of viewing the bystander as doing the one a wrong if he proceeds – albeit a wrong it is permissible to do him.
What counts as “an object which threatens death”? What marks one threat off from another? I have no doubt that ingenious people can construct cases in which we shall be unclear whether to say that if the agent proceeds, he makes threaten the one the very same thing as already threatens the five. [1413]
Moreover, which are the interventions in which the agent gets a thing that threatens five to instead threaten one by means that themselves constitute infringements of stringent rights of the one’s? I have no doubt that ingenious people can construct cases in which we shall all be unclear whether to say that the agent’s means do constitute infringements of stringent rights – and cases also in which [80]we shall be unclear whether to say the agent’s means constitute infringements of any rights at all.
But it is surely a mistake to look for precision in the concepts brought to bear to solve this problem: There isn’t any to be had. It would be enough if cases in which it seems to us unclear whether to say “same threat,” or unclear whether to say “non-right-infringing-means,” also seemed to us to be cases in which it is unclear whether the agent may or may not proceed; and if also coming to see a case as one to which these expressions do (or do not) apply involves coming to see the case as one in which the agent may (or may not) proceed.
If these ideas are correct, then we have a handle on anyway some of the troublesome cases in which people make threats. Suppose a villain says to us “I will cause a ceiling to fall on five unless you send lethal fumes into the room of one.” Most of us think it would not be permissible for us to accede to this threat. Why? We may think of the villain as part of the world around the people involved, a part which is going to drop a burden on the five if we do not act. On this way of thinking of him, nothing yet threatens the five (certainly no ceiling as yet threatens them) and a fortiori we cannot save the five by making what (already) threatens [82]them instead threaten the one. Alternatively, we may think of the villain as himself a threat to the five. But sending the fumes in is not making him be a threat to the one instead of to the five. The hypothesis I proposed, then, yields what it should: We may not accede.
That is because the hypothesis I proposed says nothing at all about the source of the threat to the five. Whether the threat to the five is, or is caused by, a human being or anything else, it is not permissible to do what will kill one to save the five except by making what threatens the five itself threaten the one.
By contrast, it seems to me very plausible to think that if a villain has started a trolley towards five, we may deflect the trolley towards one – other things being equal, of course. If a trolley is headed towards five, and we can deflect it towards one, we may, no matter who or what caused it to head towards the five.
I think that these considerations help us in dealing with a question I drew attention to earlier. Suppose a villain says to us “I will cause a [1414] ceiling to fall on five unless you send lethal fumes into the room of one.” If we refuse, so that he does what he threatens to do, then he surely does something very much worse than we would be doing if we acceded to his threat and sent the fumes in. If we accede, we [84]do something misguided and wrongful, but not nearly as bad as what he does if we refuse.
It should be stressed: The fact that he will do something worse if we do not send the fumes in does not entail that we ought to send them in, or even that it is permissible for us to do so.
How after all could that entail that we may send the fumes in? The fact that we would be saving five lives by sending the fumes in does not itself make it permissible for us to do so. (Rights trump utilities.) How could adding that the taker of those five lives would be doing what is worse than we would tip the balance? If we may not infringe a right of the one in order to save the five lives, it cannot possibly be thought that we may infringe the right of that one in order, not merely to save the five lives, but to make the villain’s moral record better than it otherwise would be.
For my own part, I think that considerations of motives apart, and other things being equal, it does no harm to say that
(II) Killing five is worse than killing one
is, after all, true. Of course we shall then have to say that assessments of which acts are worse than which do not by themselves settle the question of what is permissible for a person to do. For we shall have to say that, despite the truth of (II), it is not the case that we are required to kill one in [86]order that another person shall not kill five, or even that it is everywhere permissible for us to do this.
What is of interest is that what holds inter-personally also holds intra-personally. I said earlier that we might imagine the surgeon of Transplant to have caused the ailments of his five patients. Let us imagine the worst: He gave them chemical X precisely in order to cause their deaths, in order to inherit from them. Now he repents. But the fact that he would be saving five lives by operating on the one does not itself make it permissible for him to operate on the one. (Rights trump utilities.) And if he may not infringe a right of the one in order to save the five lives, it cannot possibly be thought that he may infringe the right of that one in order, not merely to save the five lives, but to make his own moral record better than it otherwise would be.
Another way to put the point is this: Assessments of which acts are worse than which have to be directly relevant to the agent’s circumstances if they are