The Trolley Problem as a Diagnostic
On this page(4)
A runaway trolley will kill five people on the track ahead. You are standing at a lever that would divert it onto a side track, where it will kill one. Most people say you may pull the lever.
Second case. The trolley will kill the same five, there is no side track, and you are on a footbridge above the line next to a large man whose body would stop it. Pushing him kills him and saves the five. Almost nobody says you may push.
Five saved and one killed in both. The interesting fact is not what people answer; it is that they answer differently and cannot readily say why.
What the problem is
Definition
Trolley problem
The problem of identifying a principle that permits diverting the trolley onto one person while forbidding pushing one person into its path, given that both save five at the cost of one.
The definition is worth reading twice, because the phrase is routinely misused. The trolley problem is not the question of what you would do. It is the demand for a principle that yields both verdicts, and it is a problem because the obvious candidates keep failing.
The cases originate with Philippa Foot, who used a version in 1967 while discussing the doctrine below. Judith Jarvis Thomson developed them, moving the agent from the trolley's cab to a bystander at a lever and setting that against the footbridge version. The pairing is designed rather than found: the two cases are built to differ in as few respects as possible, so that whatever explains the difference in verdict has to be one of the few things left.
Consequentialism has no problem here, and that is worth registering as a mark in its favour before the objections start. If only outcomes matter, the cases are identical and the answer is the same in both. A consequentialist who says push is not embarrassed; the theory predicted that most people's reaction to the footbridge is a squeamishness with no moral standing. The problem is a problem for those who think the reaction is tracking something.
Check your understanding
The first candidate fails
Definition
Doing and allowing
The distinction between bringing a harm about and failing to prevent one. It underlies the widely held view that killing is harder to justify than letting die, even when the outcome is the same.
The distinction is real and it does substantial work elsewhere. A surgeon who withdraws futile treatment and one who administers a lethal injection stand differently, and most moral thought treats the difference as more than a technicality about how the death was produced.
It does not solve the trolley problem. In both cases you kill the one. Pulling the lever is not letting anybody die; it is redirecting a lethal object onto a person who was in no danger a moment earlier. If doing were what mattered, the switch case should be as forbidden as the footbridge, and it is not.
This is what made Thomson's move from the driver to a bystander consequential. Foot's original driver is choosing between two things he does, and the case could be read as a choice between two killings. Once the agent is a bystander who has the option of doing nothing at all, the choice really is between killing one and allowing five to die, and the ordinary verdict is that killing the one is permitted. That verdict is hard to reconcile with the doing and allowing distinction being what drives the footbridge case.
Check your understanding
The second candidate
Definition
Doctrine of double effect
The principle that an act with a foreseen bad effect may be permissible when the bad effect is not intended, either as an end or as a means, provided the act is otherwise acceptable and the good achieved is proportionate to the harm.
The doctrine descends from Thomas Aquinas, who argued in the Summa Theologiae that an act of self-defence may have two effects, of which only one is intended, and that the killing of the attacker is beside the intention. Later writers built it into a set of conditions, and it has been central to Catholic moral theology, to the law of armed conflict, and to medical ethics.
Applied to the pair, it appears to give exactly the right answers.
At the lever, the death of the one is foreseen and not intended. If he were somehow to survive, nothing would have gone wrong with your plan; the trolley is already diverted and the five are already safe. His presence on the side track is an unwanted feature of the situation rather than a part of what you are doing.
On the footbridge, the man's being struck is the means. If he were to survive, your plan would have failed, because it is his body stopping the trolley that saves the five. You are not merely foreseeing his death; you are using him, and using a person as a mechanism is the thing deontological theories are most confident is forbidden.
The fit is impressive and it is not decisive. The standard pressure is a variant in which the side track loops back around and rejoins the main line before the five, so that diverting the trolley saves them only because the one person's body stops it on the loop. Here the one is used as a means exactly as on the footbridge, and yet many people's verdict stays where it was in the switch case. If that reaction is right, double effect is not what is doing the work.
Objection
This is a parlour game. No trolley has ever presented anyone with these options, the cases are engineered to have exactly two outcomes with certainty and no third alternative, and the whole apparatus of levers and footbridges has produced sixty years of commentary and no discernible improvement in anyone's conduct. Real moral difficulty involves uncertainty, competing obligations, and people whose interests you cannot enumerate. Building a method around cases stripped of all of that trains a skill with no application.
Reply
Some of this is right and the artificiality is not the part that is.
Cases are stripped down for the same reason experiments have controls. The point of the pair is that the two situations differ in as little as possible, so that a difference in verdict can be attributed to something specific. Adding uncertainty and competing obligations would make the cases realistic and useless, because any difference in reaction could then be put down to any of a dozen features.
Where the objection has teeth is the transfer question. It is a genuine and open issue whether a principle extracted from cases with stipulated certainty applies to a world in which nothing is certain, and philosophers who work on the ethics of war and of medicine have argued in both directions.
And the charge of unseriousness is not fair to the literature it targets. The distinctions the cases isolate, between what is intended and what is foreseen, between doing and allowing, and between treating a person as an obstacle and as an instrument, are the distinctions on which bombing campaigns, triage protocols, and the withdrawal of treatment are actually argued about. The trolley is where they get tested under controlled conditions.
Check your understanding
The question underneath
Every move in this article has treated a widely shared reaction to a case as evidence about what is true. That assumption has been doing all the work and has not been defended once. The next article asks whether it can be.
Further reading
- Philippa Foot, "The Problem of Abortion and the Doctrine of the Double Effect", Oxford Review, number 5 (1967); reprinted in Virtues and Vices and Other Essays in Moral Philosophy.
- Judith Jarvis Thomson, "Killing, Letting Die, and the Trolley Problem", The Monist, volume 59, number 2 (1976), pages 204 to 217.
- Judith Jarvis Thomson, "The Trolley Problem", The Yale Law Journal, volume 94, number 6 (1985), pages 1395 to 1415.
- Joshua D. Greene, R. Brian Sommerville, Leigh E. Nystrom, John M. Darley, and Jonathan D. Cohen, "An fMRI Investigation of Emotional Engagement in Moral Judgment", Science, volume 293 (2001), pages 2105 to 2108.
- Stanford Encyclopedia of Philosophy, "Doctrine of Double Effect", by Alison McIntyre.
- Stanford Encyclopedia of Philosophy, "Doing vs. Allowing Harm".
Quiz yourself on this article
8 questions. Miss one and it comes back until you get it right.
Reached the end?
Mark it, and the sidebar keeps track of where you have been.