The Trolley Problem

A runaway tram is heading toward five workers who cannot move out of the way. You stand beside a lever that can divert it onto a side track where one worker stands. Diverting kills that one; doing nothing lets five die. Philippa Foot used this kind of case in her 1967 discussion of the doctrine of double effect, and later writers modified it so that the chooser is a bystander rather than the driver.

The case is deliberately difficult, and that is the point. Both options are tragic, and the discomfort survives any amount of arithmetic. The case became a standard instrument because it holds consequences nearly constant while varying the agent's role in the harm.

Judith Jarvis Thomson sharpened the setup. In her footbridge case there is no lever, only a large stranger whose body would stop the tram if you pushed him from a bridge. The numbers are identical—five saved, one killed—yet a reader may accept the lever case and reject the push, suggesting that the refusal can involve more than the arithmetic.

Thomson also discussed loop variants in which the diverted tram could not stop without hitting the one person, so the death becomes part of the means rather than a side effect. Each variant removes one explanation while keeping the outcome constant: personal force, using someone as a means, intended versus merely foreseen harm. Whatever explains your pattern of judgements has to survive all of them.

The trolley problem is not really about trams. It is a probe for moral structure: when the totals are fixed, does what remains matter—how the harm is produced, what the agent intends, whether the victim is used, redirected, or simply allowed to suffer? Competing traditions give different answers. Consequentialists treat the numbers as close to decisive. Defenders of the doctrine of double effect distinguish harm that is intended from harm that is foreseen. Others emphasise the doing-versus-allowing distinction or the constraint against using persons as mere means.

No result from the thought experiment settles the issue by itself. A pattern of answers is useful when it makes the underlying distinction explicit enough to defend, revise, or reject.

The structure recurs wherever scarce harm must be allocated: medical triage under shortage, emergency routing, and the rules encoded in autonomous systems. Real cases add features the tidy story omits—uncertainty about outcomes, consent, responsibility for having built the situation, and accountability after the fact.

For that reason philosophers caution against treating trolley verdicts as ready-made policy. The experiment earns its keep by revealing which considerations carry weight for you, so that a rule written later can be argued for rather than assumed.