Trolley Problem

The trolley problem is a family of moral dilemmas, originally from Philippa Foot (1967) and sharpened by Judith Jarvis Thomson (1985), used to pressure-test ethical intuitions (see the SEP entry). The point isn’t to solve the dilemma but to expose the structure of the intuitions it triggers, and in particular to test the Doctrine of Double Effect.

Why do our verdicts flip?

If five-vs-one arithmetic stays constant but our verdicts flip, the moral work is being done by something other than the body count, and naming it matters.

The two canonical setups

Bystander (permissible to most): a runaway trolley will kill five on the main track. You can pull a switch to divert it onto a side track where it will kill one. Most people say: pull the switch.

Footbridge / Fat Man (impermissible to most): same five people. You’re on a footbridge above the track next to a large stranger. The only way to stop the trolley is to push him onto the tracks, where his body stops it. Five saved, one dead, same arithmetic. Most people say: don’t push.

Five-vs-one is identical. Intuitions flip. Why?

Doctrine of Double Effect (DDE)

The traditional scholastic principle (Mangan’s formulation). An action with both good and bad effects is permissible iff:

  1. The action itself is morally good or neutral
  2. The good effect is intended; the bad effect is not
  3. The good effect is not produced by means of the bad effect
  4. There’s a proportionate reason for tolerating the bad effect

Bystander passes DDE: the one death is a foreseen side effect of diverting the trolley, not your means. If the person stepped off the side track, your plan still works.

Footbridge fails DDE condition 3: the stranger’s body is instrumentally necessary. If he somehow survived the fall, the trolley doesn’t stop. You’re using him as a means.

Objections to DDE:

  • Closeness problem: the line between “intended” and “foreseen but regretted” is slippery. When a surgeon performs a life-saving hysterectomy on a pregnant woman, is the fetal death intended or merely foreseen? The doctrine can be manipulated by redescribing intentions
  • The Loop Variant (Thomson): a track loops back: divert the trolley and it hits one person, whose body then stops it from continuing around onto the five. The one’s death is now causally a means, not a side effect. But many still judge it permissible, undercutting DDE’s core distinction
  • Consequentialist objection: identical outcomes; why should an agent’s mental state determine permissibility?

Alternative explanations (per SEP):

  • Doing vs. allowing (Foot’s original view): pulling the switch redirects pre-existing danger; pushing introduces new danger
  • Kamm’s Principle of Permissible Harm: harm as a side effect of a greater good is different from harm as a causal means to that good
  • Using-as-means (Quinn, Kamm): direct agency (deliberately incorporating victims into your plan) is worse than indirect agency. This reframes DDE without leaning on the intention/foresight distinction

My take

I’ve read this as an engineer for years and my instinct is the consequentialist objection: five > one, pull every switch, push every man, stop moralizing about the switch/push distinction. The numbers are the numbers.

But I think that instinct is probably wrong, and the Trolley Problem is how I know. If I really held “five > one, always,” I should be indifferent between the Bystander and the Footbridge cases. I’m not. Nobody is. And “I’m not, but I should be” is a weaker position than taking the intuition seriously and asking what moral structure it’s tracking.

What it’s probably tracking, on the evidence: the using-as-means distinction, not DDE’s intention/foresight version. The Loop Variant kills DDE for me. But “don’t incorporate another person’s body into your plan as a load-bearing part of the machine” holds up across all the variants I’ve seen, and it squares with why Footbridge feels monstrous in a way Bystander doesn’t.

Relevant to my existing ethics notes and utilitarianism: this is the cleanest case I know where naive utilitarianism gives an answer that almost no one endorses on reflection. That’s not a refutation, but it’s a big data point. Utilitarianism has to either absorb the intuition (rule-utilitarian moves, indirect utilitarianism) or explain why it’s a bias to be overcome, and the latter gets harder the more the intuition survives scrutiny.