Philosophy · 6 min read
What the Trolley Problem Was Designed to Test
The original philosophical contrasts, what later variants test, and why survey responses do not settle moral theory.

What the Trolley Problem Was Designed to Test
The trolley problem is often presented as a forced choice: divert a runaway vehicle so that one person dies instead of five, or do nothing. That question can survey a judgment, but it is not the original philosophical problem by itself. Philippa Foot introduced the runaway tram in 1967 while examining the doctrine of double effect. Judith Jarvis Thomson later asked why diverting the trolley can seem permissible when killing one healthy person to save five patients does not.
The thought experiment was designed to test proposed distinctions among intention, foresight, action, omission, rights, and the use of a person as a means. It was not a personality test that sorts everyone into utilitarians and rule-followers, and agreement about one variant does not settle an ethical theory.
Foot's comparison concerned double effect
The doctrine of double effect attempts to distinguish harm intended as a means or end from harm foreseen as a side effect. Foot's 1967 essay, “The Problem of Abortion and the Doctrine of the Double Effect,” examined whether that distinction explained moral judgments in cases involving abortion, euthanasia, and killing. Her tram driver can steer from a track with five workers onto a track with one. The one person's death is foreseen, but redirecting the tram does not use that death to stop the vehicle.
Foot compared such cases with choices in which harm appears to be part of the means. She did not offer the tram case as a free-standing referendum on whether five lives outweigh one. Her broader analysis questioned whether intention alone could carry all the moral work assigned to it. She proposed that negative duties, such as not injuring, and positive duties, such as providing aid, had different stringency, while acknowledging complications.
This context matters because a bare count conceals the variable under examination. Both a consequentialist and a nonconsequentialist might redirect the tram, but for different reasons. One might minimize deaths; another might permit redirecting an existing threat while rejecting the deliberate creation of a new threat. The answer alone does not identify the principle.
Thomson turned a contrast into a problem
In “Killing, Letting Die, and the Trolley Problem” in 1976, Thomson gave the problem its durable name and sharpened the comparison. Why may a bystander turn a switch that redirects a trolley toward one person, while a surgeon may not kill one healthy patient and distribute the organs to five people who otherwise die? Both descriptions contain a choice associated with one death and five survivals.
Thomson's later variants tested candidate explanations. In the footbridge case, a bystander can stop the trolley only by pushing a large person into its path. Many readers who permit turning the switch reject the push. That contrast pressures an account based only on the number of deaths or only on acting versus allowing. Both the switch operator and the person on the bridge would act.
The cases also pressure double effect. In the switch case, the one person's presence on the side track is not what stops the trolley; in the footbridge case, the person's body is introduced as the obstacle. This supports a distinction between harming someone as a means and harming them as a side effect. Yet variants with loops, obstacles, prior threats, and self-sacrifice produce judgments that no short formula captures without counterexamples.
Variants test proposed distinctions
A philosophical thought experiment holds many features steady while changing one. The comparison can reveal that a proposed rule is too broad or that an intuition is sensitive to an unacknowledged fact. It cannot, by intuition alone, show whether the sensitivity is morally justified.
Consider proximity. Pushing a person involves physical contact, while turning a switch is remote. A variant can replace the push with a trapdoor operated from a distance. If judgments remain different, physical contact is unlikely to be the complete explanation. Consider intention: a rescuer may predict the one death without wanting it. But philosophers disagree about how intentions are individuated and whether “close intention” can be separated cleanly from accepted means.
Rights-based accounts ask whether the action redirects a threat or imposes a new one, whether the victim has a claim against the agent, and whether the victim is used. Consequentialist accounts can incorporate rules, uncertainty, institutional trust, and effects beyond the stipulated deaths. Virtue ethics may ask what practical wisdom requires rather than treating the choice as a calculation detached from character. The cases do not force each family of theories into its simplest classroom version.
Framing also changes the stipulated problem. Calling the five “workers” and the one a “bystander,” mentioning responsibility, or changing certainty can introduce morally relevant facts. A clean comparison requires equal probabilities and no rescue alternative, but real decisions rarely supply those conditions.
Experiments measure judgments rather than morality
Moral psychologists have used trolley cases to study how judgments vary with personal force, intention, causal structure, emotion, order, language, and culture. Joshua Greene and colleagues reported different patterns of brain activity for “personal” and “impersonal” dilemmas, helping launch dual-process accounts. Later work challenged simple mappings between brain regions, emotion, and ethical categories. Functional imaging measures blood-oxygen changes associated with tasks; it does not identify a moral truth detector.
Large surveys find both recurring contrasts and variation. Many participants permit switching and reject pushing, but proportions depend on wording and sample. A 2018 cross-cultural study led by Edmond Awad collected online responses from millions of participants to autonomous-vehicle scenarios. It found clusters of preferences across countries, but the volunteer internet sample, stylized alternatives, and multiple social attributes make the project unlike Foot's controlled contrast. Descriptive preference is not a public-policy mandate.
The cases have also been applied to self-driving vehicles. The analogy is limited. Vehicle engineering concerns risk distributions, sensing failures, braking, regulation, and behavior before a collision. A programmed system rarely faces a certain choice between known persons with fixed outcomes. Treating safety design as one dramatic last-second selection can distract from measures that reduce crash probability for all road users.
What the thought experiment can establish
The historical record establishes what the trolley cases were built to examine: first double effect and duties, then the difficulty of explaining contrasts among cases with similar totals. Variants can refute an explanation that predicts identical judgments where stable differences appear. Experiments can estimate how people respond under defined presentations.
What remains contested is whether the judgments track morally relevant principles, learned conventions, emotional reactions, framing artifacts, or combinations. No response to a single vignette demonstrates allegiance to a complete theory. The trolley problem endures because it isolates disagreements efficiently. Its proper conclusion is a question about the principle behind a judgment, not a verdict that the judgment has resolved ethics.
Sources
5 sources, 3 of them with a link you can follow.
- Philippa Foot, “The Problem of Abortion and the Doctrine of the Double Effect,” Oxford Review 5 (1967).
- Judith Jarvis Thomson, “Killing, Letting Die, and the Trolley Problem,” The Monist 59 (1976), doi:10.5840/monist197659224.
- Stanford Encyclopedia of Philosophy, “Doctrine of Double Effect,” substantive revision December 2022.
- Greene et al., “An fMRI Investigation of Emotional Engagement in Moral Judgment,” Science 293 (2001), doi:10.1126/science.1062872.
- Bruers and Braeckman, “A Review and Systematization of the Trolley Problem,” Philosophia 42 (2014), doi:10.1007/s11406-013-9507-5.
