Where the Trolley Problem Comes From
10 Claims, Checked

The trolley problem is a thought experiment in ethics: a runaway vehicle will kill five people unless you divert it onto a track where it will kill one. Here are ten common claims about where it came from and what it shows, checked against the papers themselves. Whether you should pull the lever is a question of values, and this page does not answer it.

1. “Philippa Foot invented the trolley problem”

Partly true She wrote the case everything descends from, in “The Problem of Abortion and the Doctrine of the Double Effect” (Oxford Review, 1967). In the text as later reprinted in her collection Virtues and Vices, the chooser “is 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”. But she did not give it a name, and she built it from an example already in circulation. After a judge case that, she says, “the controversy has raged around”, she writes: “Beside this example is placed another in which a pilot whose airplane is about to crash is deciding whether to steer from a more to a less inhabited area. To make the parallel as close as possible it may rather be supposed that he is the driver of a runaway tram”. The tram is her adjustment of the pilot case. The name came later (see 4).

2. “The original paper was about trolleys”

Myth It was about abortion. Foot’s stated method was “setting up parallels involving adults or children once born”, and her target was the doctrine of double effect, which she defines as “the thesis that it is sometimes permissible to bring about by oblique intention what one may not directly intend”. The tram is one of those parallels. She pairs it with a judge, faced with rioters, who can prevent the bloodshed “only by framing some innocent person and having him executed”, and asks why we would say “without hesitation, that the driver should steer for the less occupied track, while most of us would be appalled at the idea that the innocent man could be framed”. Her answer is that the driver faces “a conflict of negative duties”, while the judge is “weighing the duty of not inflicting injury against the duty of bringing aid”. Then she writes: “Let us now return to the problem of abortion”. And she wrote “tram”, not “trolley”.

3. “The dilemma began in 1967”

Myth A version was put to students at the University of Wisconsin in December 1905, and it was not new then: Sharp writes that “one hundred and forty-five persons … answered IX of Series I in an inquiry made in 1896 and 1897”, and that “the returns of 1905 were practically identical.” Frank Chapman Sharp published the results in 1908 in A Study of the Influence of Custom on the Moral Judgment. His Question IX: a man sees that “a careless switchman had left a switch open which would mean death or injury to the several hundred people on the” train, while “he saw his only child playing upon the track in front of the approaching engine. He had time only to turn the switch and save the train or else to save the child.” It is not the modern case: the one person is the chooser’s own child, and Sharp was studying custom and moral judgment.

German criminal law has its own “switchman” (Weichensteller) case, discussed by Karl Engisch in 1930 and Hans Welzel in 1951, according to a 2019 article by Liane Wörner. In Welzel’s version a railway official throws the switch at the last minute and, as he foresaw, three workers are killed. German criminal law treats it differently from most survey respondents: according to Wörner, under the overwhelmingly prevailing view today, throwing the switch “kann … nicht gerechtfertigt werden” (cannot be justified; our translation), though it may be excused on other grounds. Neither Foot’s 1967 paper nor Thomson’s 1985 paper cites any of these.

4. “Judith Jarvis Thomson gave it its name”

True Her 1976 paper in The Monist is titled “Killing, Letting Die, and the Trolley Problem”. In her 1985 Yale Law Journal paper, “The Trolley Problem”, she opens by crediting Foot (“Some years ago, Philippa Foot drew attention to an extraordinarily interesting problem”), turns Foot’s tram into a trolley, and changes the chooser from the driver to “a case I will call Bystander at the Switch”. Then: “Since I find it particularly puzzling that the bystander may turn his trolley, I am inclined to call this The Trolley Problem.” She adds: “Those who find it particularly puzzling that the surgeon may not operate are cordially invited to call it The Transplant Problem instead.”

5. “The footbridge version is called ‘the fat man’”

True, and the label is older than the footbridge. Thomson’s 1985 paper sets out “a case … which I shall call Fat Man”, in which “you are standing on a footbridge over the trolley track”. But Foot had already used a fat man in 1967, in a different story she calls “well known to philosophers”: “the fat man stuck in the mouth of the cave”, whom the people trapped behind him could blast out with dynamite. Foot does not say who first told it. The 2020 study in 6 calls it “Footbridge”.

6. “Most people would pull the lever”

Partly true, for the switch, and much less so for the footbridge. A large study, by Awad and colleagues in PNAS (2020), analysed “three sacrificial dilemmas by 70,000 participants in 10 languages and 42 countries”. Participants endorsed the sacrifice in the switch case at a “country-level average: 81%”, in a loop variant at 72%, and in the footbridge case at 51%. The ordering was the same everywhere, but “the quantitative acceptability of each sacrifice, however, showed substantial country-level variations”. Two cautions come from the authors themselves: “our sample shows clear signs of self-selection”, and “we estimate that a third of our participants were young, college-educated men”.

Thomson’s own “Everybody to whom I have put this hypothetical case says, Yes, it is”, said of Foot’s driver case, was a remark about the people she had asked, not a survey.

7. “What people answer in a trolley dilemma predicts what they would do”

Disputed Psychologists took the dilemmas up. Greene and colleagues used them in brain scans in 2001, arguing that “moral dilemmas vary systematically in the extent to which they engage emotional processing”. Critics question what the answers show. Bauman and colleagues (2014) worried that such studies “may lack experimental, mundane, and psychological realism”, partly because “they are amusing rather than sobering”. FeldmanHall and colleagues (2012) found that “real moral decisions can dramatically contradict moral choices made in hypothetical scenarios”, though their task was not a trolley case. In a 2018 study in which participants chose whether to “administer an electroshock (that they did not know was bogus) to a single mouse or allow five other mice to receive the shock”, Bostyn and colleagues reported that “responses to hypothetical dilemmas are not predictive of real-life dilemma behavior, but they are predictive of affective and cognitive aspects of the real-life decision”. Five later pieces in the same journal discuss it, and it is one study. On the other side, in a virtual-reality simulation of the switch case, Navarrete and colleagues (2012) found that “the majority of research participants behaved as ‘moral utilitarians’”, the choice most people also make on paper. The fair summary: the dilemmas reliably show how people judge; how well they show what people do is argued over.

8. “Thomson always held that you should pull the lever”

Myth She changed her mind. According to the Stanford Encyclopedia of Philosophy, “Thomson (2008) changed her mind and argued that the consensus that it is permissible for the bystander to turn the trolley was mistaken.” Her argument adds a third option: “turning the trolley onto and killing oneself”. If you would not do that, the entry summarises, “we are not entitled to turn the trolley onto a stranger.”

9. “It was invented to study self-driving cars”

Myth It is decades older (see 1 to 3). The self-driving framing arrived with work such as Bonnefon and colleagues in Science (2016), and the Moral Machine experiment in Nature (2018), “an online experimental platform designed to explore the moral dilemmas faced by autonomous vehicles”, which gathered “40 million decisions in ten languages from millions of people in 233 countries and territories”. The 2020 figures in 6 came from that site’s “Classic” mode, “where visitors made decisions about the traditional Switch, Loop, and Footbridge variants”.

10. “Self-driving cars will face real trolley problems”

Disputed There are three positions, each with published support.

• The question is real. Bonnefon’s team wrote that such cars “will sometimes have to choose between two evils”, and found that participants “approved of utilitarian AVs (that is, AVs that sacrifice their passengers for the greater good) and would like others to buy them, but they would themselves prefer to ride in AVs that protect their passengers at all costs”.
• The format distorts the answer. Yochanan Bigman and Kurt Gray argued in Nature (2020) that the Moral Machine’s apparent preferences for some lives over others were “driven by the specific ‘trolley-type’ paradigm”, and that “people overwhelmingly want autonomous vehicles to treat different human lives equally”. The Moral Machine authors replied that Bigman and Gray “make the important point that the way we measure these preferences can affect the results we obtain”.
• The analogy is weak. Sven Nyholm and Jilles Smids (2016) identified “three important ways” the two differ: the decision situation, “moral and legal responsibility”, and “decision-making in the face of risks and uncertainty”.

The evidence does not settle which of the three is right.

Germany’s Ethics Commission on Automated and Connected Driving (June 2017) recommended that in unavoidable accidents “any distinction based on personal features (age, gender, physical or mental constitution) is strictly prohibited”, that it is “also prohibited to offset victims against one another”, and that “general programming to reduce the number of personal injuries may be justifiable”. It added: “Those parties involved in the generation of mobility risks must not sacrifice non-involved parties.” It also noted a difference from the classic case: in a real crash “the identity of the injured or killed parties is not yet known (unlike in the trolley dilemma)”. Those are a report’s recommendations; what a car ought to do is a question of values, and we leave it there.

Our reading

The trolley problem started as a side example in an argument about abortion, used to test the doctrine of double effect rather than to be answered for its own sake. Foot used it to ask why two cases with the same numbers feel so different. That question has lasted longer than any answer to it.

Sources

• Philippa Foot, “The Problem of Abortion and the Doctrine of the Double Effect”, Oxford Review 5 (1967), as reprinted in Virtues and Vices (PhilArchive copy via the Internet Archive).
• Judith Jarvis Thomson, “Killing, Letting Die, and the Trolley Problem”, The Monist 59(2) (1976), bibliographic record only; “The Trolley Problem”, Yale Law Journal 94 (1985), Yale Law School open repository.
• Frank Chapman Sharp, A Study of the Influence of Custom on the Moral Judgment, Bulletin of the University of Wisconsin (1908), Internet Archive.
• Liane Wörner, “Der Weichensteller 4.0”, Zeitschrift für Internationale Strafrechtsdogmatik 1/2019.
• Alison McIntyre, “Doctrine of Double Effect”; Fiona Woollard and Frances Howard-Snyder, “Doing vs. Allowing Harm”, Stanford Encyclopedia of Philosophy.
• Awad et al., “Universals and variations in moral decisions made in 42 countries by 70,000 participants”, PNAS 117 (2020); Awad et al., “The Moral Machine experiment”, Nature 563 (2018); Bonnefon, Shariff and Rahwan, “The social dilemma of autonomous vehicles”, Science 352 (2016).
• Greene et al., Science 293 (2001); Navarrete et al., Emotion 12 (2012); Bauman et al., Social and Personality Psychology Compass 8 (2014); FeldmanHall et al., Cognition 123 (2012); Bostyn, Sevenhant and Roets, Psychological Science 29 (2018).
• Bigman and Gray, “Life and death decisions of autonomous vehicles”, and the authors’ reply, Nature 579 (2020); Nyholm and Smids, Ethical Theory and Moral Practice 19 (2016).
• Ethics Commission on Automated and Connected Driving, Report (June 2017), German Federal Ministry of Transport.

Checked September 2026.

Related: Logical fallacies cheat sheet · Cognitive biases, checked · How to read a study