psychology.games

Hall II · Psychological Phenomena, Tasks & Thought Experiments

Trolley problem

Station · Guided activity

Three classic hypothetical dilemmas, presented for reflection. There are no right answers and no score here; your choices stay on this page, are never recorded or sent anywhere, and say nothing diagnostic about you.

There are no right answers in this sequence, no verdict at the end, and nothing you choose is recorded.

Three short hypotheticals, each a forced choice between two acts with serious stakes. They are thought experiments: nobody in them is real, nothing you choose is recorded, and no correct answer is waiting at the end. Answer honestly, change your mind freely, and notice how your third answer sits beside your first — what this page cares about is the contrast between your answers, never any single one of them.

The experience

Three changes of scene, one shape of choice. A runaway trolley and a lever; a footbridge and a stranger; a surgeon and a healthy visitor. The sequence above walks you through all three — no score, and nothing you choose leaves this page. The interesting datum is not any single answer but the pattern across them, because for most people the pattern comes apart in a very particular way. The split is remarkably stable: in the questionnaire data behind one influential research programme, about nine in ten people judged turning the trolley permissible while only about one in ten said the same of pushing, and a study of 70,000 participants in 42 countries found the same ordering of cases — switch endorsed most, the loop variant next, the footbridge least — in every country sampled.Convergent / replicated If yours just came apart the same way, the rest of this page is about why.

What is happening

This is a thought experiment, so the philosophy is not background here; it is the object itself. The scenario began life in 1967, when the philosopher Philippa Foot imagined the driver of a runaway tram who can only steer from a track where five men are working onto a side track holding one, as one of a series of cases probing the doctrine of double effect. Judith Jarvis Thomson later moved the choice to a bystander at a switch, added the footbridge case, set the whole family beside a transplant surgeon who could save five patients by cutting up one healthy young man, and gave the puzzle its name: the trolley problem.Historical / descriptive Foot's own verdict: the driver faces two duties not to harm and should do the least harm — while harming one to bring aid to five is another matter. On the standard analysis the two cases differ in structure, not arithmetic: turning the trolley redirects an existing threat, so the one death is a side effect of saving five, whereas pushing uses the man himself as the brake — and Thomson pressed the same point with her surgeon, who may not cut up one healthy visitor to save five patients even though the exchange is identically one life for five.Interpretive, not empirical Whether that structure should carry moral weight is exactly what philosophers dispute; the cases make the disagreement concrete.

What this shows

As a demonstration, the trolley problem exposes something odd about moral judgment itself: the verdicts arrive fast and feel certain, but the reasons arrive late, if at all. When people are asked to justify their own pattern of answers, most cannot: in a large web study the majority failed to produce justifications that could account for their judgments, and the judgments themselves are rapid, confident, and barely predictable from gender, age, education, religion or nationality.Supported, with limits That gap is what turned a pair of fanciful cases into a serious research instrument. Philosophers use the contrast to test which principles actually fit our intuitions; psychologists use it to study machinery that introspection cannot reach. Either way, the finding worth keeping is that ordinary moral judgment is patterned — it honours distinctions its owners cannot name.

What this does not show

Your answers here are not a moral X-ray. Diverting the trolley does not make you a cold calculator, refusing the footbridge does not make you a Kantian, and no combination of answers diagnoses character, empathy or anything else — the summary this page offers describes your pattern of answers this session, nothing more. There is also a live argument about what any trolley answer is worth. An answer to a cartoonish hypothetical is not a forecast of behaviour. Critics of trolleyology point out that the dilemmas were built as philosophical intuition pumps rather than realistic stimuli, that people often find them funny — in one survey 63% laughed at least a little at the footbridge case — and that responses may not generalize to real moral situations; the same critics add that the dilemmas remain a legitimate source of data when those limits are respected.Supported, with limits Foot herself wrote that the levity of her examples was not meant to offend; the levity is real, and worth keeping in mind.

Research and limitations

What has half a century of trolleyology established? Brain-imaging and behavioural experiments give the felt difference some mechanics: Greene and colleagues found that up-close 'personal' dilemmas like the footbridge engage emotion-associated brain areas more than switch-like cases and slow down those who approve, and later experiments sharpened the factor to an interaction — harm delivered by the agent's own muscular force is judged less acceptable, but only when the harm is intended as a means rather than foreseen as a side effect.Supported, with limits Even that is not the last word: results on subtler variants such as the loop track disagree between laboratories, and Greene's group described its own personal-versus-impersonal cut as preliminary.

Culture enters more quietly than you might expect. The 42-country study also maps where culture enters: every country showed the same qualitative ordering of the three dilemmas, but countries differed considerably in how acceptable the sacrifices were overall, with endorsement lower where relational mobility is low — and the authors caution that their self-selected web sample skewed young, male and university-educated.Supported, with limits So treat the sequence above as the field treats the dilemma: a sharp probe of how moral judgment is structured, and a poor oracle for what anyone would actually do beside a real track.

Research drawer — claims, sources and limits

Every highlighted claim above is recorded here with its evidence grade and sources. Grades are per claim; this site never computes an overall truth score.

↩ The split is remarkably stable: in the questionnaire data behind one influential research programme, about nine in ten people judged turning the trolley permissible while only about one in ten said the same of pushing, and a study of 70,000 participants in 42 countries found the same ordering of cases — switch endorsed most, the loop variant next, the footbridge least — in every country sampled.

Convergent / replicated

Mikhail, J. (2007). Universal moral grammar: theory, evidence and the future. Trends in Cognitive Sciences, 11(4), 143-152. https://doi.org/10.1016/j.tics.2006.12.007 · accessed 2026-08-24
Support note: Table I (data attributed to Mikhail's Rawls' Linguistic Analogy) reports 90% judging the bystander switch case (Hank) permissible versus 10% for the footbridge case (Ian); Box 3 additionally reports 94% for the trolley-driver case and 8% for Transplant. The text calls the judgments widely shared across demographically diverse populations.

Awad, E., Dsouza, S., Shariff, A., Rahwan, I., & Bonnefon, J.-F. (2020). Universals and variations in moral decisions made in 42 countries by 70,000 participants. Proceedings of the National Academy of Sciences, 117(5), 2332-2337. https://doi.org/10.1073/pnas.1911517117 · accessed 2026-08-24
Support note: Full text (read via Europe PMC) reports 70,000 participants in 42 countries; overall endorsement 81% (Switch), 72% (Loop), 51% (Footbridge), and states that every country in the dataset showed the same qualitative Switch > Loop > Footbridge ordering.

↩ The scenario began life in 1967, when the philosopher Philippa Foot imagined the driver of a runaway tram who can only steer from a track where five men are working onto a side track holding one, as one of a series of cases probing the doctrine of double effect. Judith Jarvis Thomson later moved the choice to a bystander at a switch, added the footbridge case, set the whole family beside a transplant surgeon who could save five patients by cutting up one healthy young man, and gave the puzzle its name: the trolley problem.

Historical / descriptive

Foot, P. (1967). The Problem of Abortion and the Doctrine of the Double Effect. Oxford Review, 5, 5-15. Reprinted in Foot, Virtues and Vices and Other Essays in Moral Philosophy (1978). https://sites.pitt.edu/~mthompso/readings/foot.pdf · accessed 2026-08-24
Support note: Full text read. The paper (Oxford Review no. 5, 1967) poses 'the driver of a runaway tram which he can only steer from one narrow track on to another; five men are working on one track and one man on the other', inside an examination of the doctrine of double effect and the abortion debate.

Thomson, J. J. (1985). The Trolley Problem. The Yale Law Journal, 94(6), 1395-1415. https://openyls.law.yale.edu/handle/20.500.13051/16338 · accessed 2026-08-24
Support note: Pages 1395-1409 read. Thomson presents Bystander at the Switch, the Fat Man footbridge case and Transplant (five patients each needing an organ; a healthy young man in for a check-up), and writes that she is 'inclined to call this The Trolley Problem' — the question why the bystander may turn the trolley though the surgeon may not operate.

↩ On the standard analysis the two cases differ in structure, not arithmetic: turning the trolley redirects an existing threat, so the one death is a side effect of saving five, whereas pushing uses the man himself as the brake — and Thomson pressed the same point with her surgeon, who may not cut up one healthy visitor to save five patients even though the exchange is identically one life for five.

Interpretive, not empirical

Foot, P. (1967). The Problem of Abortion and the Doctrine of the Double Effect. Oxford Review, 5, 5-15. Reprinted in Foot, Virtues and Vices and Other Essays in Moral Philosophy (1978). https://sites.pitt.edu/~mthompso/readings/foot.pdf · accessed 2026-08-24
Support note: Defines the doctrine of double effect (direct versus oblique intention, harm aimed at versus foreseen) and contrasts the steering driver with cases of killing one person to obtain 'spare parts' or serum for others; Foot's own resolution appeals to negative versus positive duties rather than double effect.

Thomson, J. J. (1985). The Trolley Problem. The Yale Law Journal, 94(6), 1395-1415. https://openyls.law.yale.edu/handle/20.500.13051/16338 · accessed 2026-08-24
Support note: States both cases yield 'a net saving of four lives'; analyses the bystander as one who 'saves his five by making something that threatens them instead threaten one' without means that themselves infringe rights, while shoving a person off a footbridge, like the surgeon's operation, is itself an infringement of the victim's rights.

↩ When people are asked to justify their own pattern of answers, most cannot: in a large web study the majority failed to produce justifications that could account for their judgments, and the judgments themselves are rapid, confident, and barely predictable from gender, age, education, religion or nationality.

Supported, with limits

Hauser, M., Cushman, F., Young, L., Jin, R. K., & Mikhail, J. (2007). A dissociation between moral judgments and justifications. Mind & Language, 22(1), 1-21. https://doi.org/10.1111/j.1468-0017.2006.00297.x · accessed 2026-08-24
Support note: Official abstract (read at Wiley) states the web-collected judgment patterns showed 'little variation across differences in gender, age, educational level, ethnicity, religion or national affiliation' and that 'a majority of subjects failed to provide justifications that could account for their judgments'.

Mikhail, J. (2007). Universal moral grammar: theory, evidence and the future. Trends in Cognitive Sciences, 11(4), 143-152. https://doi.org/10.1016/j.tics.2006.12.007 · accessed 2026-08-24
Support note: Describes trolley-problem judgments as 'rapid, intuitive and made with a high degree of certitude', not predictable from age, sex, race, religion or education even in large cross-cultural samples, and as exhibiting 'a dissociation between judgments and justifications'.

↩ An answer to a cartoonish hypothetical is not a forecast of behaviour. Critics of trolleyology point out that the dilemmas were built as philosophical intuition pumps rather than realistic stimuli, that people often find them funny — in one survey 63% laughed at least a little at the footbridge case — and that responses may not generalize to real moral situations; the same critics add that the dilemmas remain a legitimate source of data when those limits are respected.

Supported, with limits

Bauman, C. W., McGraw, A. P., Bartels, D. M., & Warren, C. (2014). Revisiting External Validity: Concerns about Trolley Problems and Other Sacrificial Dilemmas in Moral Psychology. Social and Personality Psychology Compass, 8(9), 536-554. https://doi.org/10.1111/spc3.12131 · accessed 2026-08-24
Support note: Pages 536-541 read. The paper argues sacrificial dilemmas may lack experimental, mundane and psychological realism; reports the authors' survey in which 63% laughed at least a little at the footbridge scenario and 33% at the bystander scenario; describes the philosophers' cases as rhetorical devices/intuition pumps; and states explicitly that the authors are not proposing abandonment — dilemmas 'can be a legitimate source of data' if their limitations are recognized and they are not relied on exclusively.

↩ Brain-imaging and behavioural experiments give the felt difference some mechanics: Greene and colleagues found that up-close 'personal' dilemmas like the footbridge engage emotion-associated brain areas more than switch-like cases and slow down those who approve, and later experiments sharpened the factor to an interaction — harm delivered by the agent's own muscular force is judged less acceptable, but only when the harm is intended as a means rather than foreseen as a side effect.

Supported, with limits

Greene, J. D., Sommerville, R. B., Nystrom, L. E., Darley, J. M., & Cohen, J. D. (2001). An fMRI investigation of emotional engagement in moral judgment. Science, 293(5537), 2105-2108. https://doi.org/10.1126/science.1062872 · accessed 2026-08-24
Support note: Full text read. Emotion-associated areas (medial frontal gyrus, posterior cingulate, bilateral angular gyrus) were significantly more active for 'moral-personal' dilemmas (footbridge-like) than 'moral-impersonal' ones (trolley-like); 'appropriate' responses in the personal condition were significantly slower. Two experiments of nine subjects each; the authors call their conclusion descriptive, not prescriptive, and the personal/impersonal cut a preliminary 'first cut'.

Greene, J. D., Cushman, F. A., Stewart, L. E., Lowenberg, K., Nystrom, L. E., & Cohen, J. D. (2009). Pushing moral buttons: The interaction between personal force and intention in moral judgment. Cognition, 111(3), 364-371. https://doi.org/10.1016/j.cognition.2009.02.001 · accessed 2026-08-24
Support note: Full text read (proof version). Defines personal force (force generated by the agent's muscles); Experiments 1a-b found no effect of spatial proximity or physical contact but a significant personal-force effect; Experiments 2a-b found personal force lowers acceptability only when harm is intended as a means (interaction p = .006), and found loop and loop-weight cases equivalent, contradicting some earlier reports.

↩ The 42-country study also maps where culture enters: every country showed the same qualitative ordering of the three dilemmas, but countries differed considerably in how acceptable the sacrifices were overall, with endorsement lower where relational mobility is low — and the authors caution that their self-selected web sample skewed young, male and university-educated.

Supported, with limits

Awad, E., Dsouza, S., Shariff, A., Rahwan, I., & Bonnefon, J.-F. (2020). Universals and variations in moral decisions made in 42 countries by 70,000 participants. Proceedings of the National Academy of Sciences, 117(5), 2332-2337. https://doi.org/10.1073/pnas.1911517117 · accessed 2026-08-24
Support note: Full text (read via Europe PMC) reports the universal Switch > Loop > Footbridge ordering across all 42 countries with substantial country-level variation in endorsement levels; low relational mobility strongly predicted rejecting the sacrifices; the demographic subsample was about 75% male, 75% under 32 and 73% college-educated, and the authors state the viral web sample 'shows clear signs of self-selection'.