Trolley problem game: what would you do, and what did people really choose?
Four dilemmas, one at a time. After each choice you see what people in published studies chose. No score, nothing sent.
See all 4 dilemmas and the numbersFour dilemmas. Pick what you would do, then see what people in real studies chose.
The game needs JavaScript. Without it, every dilemma and its numbers are in the list below.
Not a test. No score, no type, and nothing you tap leaves your browser.
Trolley problem questions, and what people really chose
-
1 The switch
A runaway trolley is heading for five workers. You stand by a lever. Pull it, and the trolley turns onto a side track, where it will kill one worker instead.
Do you pull the lever?
Said yes to turning the trolley - About 70,000 people online in 42 countries, 2020 (country average, asked what to do)81%
- Hauser and colleagues, 2007 (asked if it was allowed, when a person on the trolley turns it)89%
- Many Labs 2 replication, 2018 (same question as Hauser)71%
Every study finds a clear majority for the switch. The exact share moves with the wording and with who is asked. Awad et al. 2020, Hauser et al. 2007, Many Labs 2, 2018
-
2 The loop
Same trolley, same five workers. This time the side track loops back to the main track. One worker stands on the loop, and his body would stop the trolley before it reaches the five.
Do you pull the lever?
Said yes to turning the trolley - About 70,000 people online in 42 countries, 2020 (country average)72%
Here the one worker's death is what stops the trolley, not a side effect. In the 2020 data, fewer chose it than the plain switch, in every country. Awad et al. 2020
-
3 The footbridge
The trolley is heading for the five again. You are on a footbridge above the track, next to a large man. Push him off, and his body stops the trolley. He dies, and the five live.
Do you push him?
Said yes to pushing - About 70,000 people online in 42 countries, 2020 (country average, asked what to do)51%
- Hauser and colleagues, 2007 (asked if it was allowed)11%
- Many Labs 2 replication, 2018 (same question as Hauser)17%
Why so far apart? Asking what you would do is not the same as asking if it is allowed, and the samples differ: in the 2020 data, 75% of players were men and 75% were under 32. Awad et al. 2020, Hauser et al. 2007, Many Labs 2, 2018
-
4 The mice (this one was real)
In a lab in Belgium, students sat in front of two cages: five mice in one, one mouse in the other. When a 20 second timer ran out, a very painful shock would hit the five, unless the student pressed a button that sent it all to the one.
Do you press the button?
Pressed the button - 83 students asked to imagine it: said they would press66%
- 192 students in the lab for real: pressed84%
No mouse was shocked. The shocks were fake, and every student was told straight after. Their answers to ten classic paper dilemmas did not predict what they did at the button. Bostyn et al. 2018
What a trolley answer can and cannot tell you
In 2001 Joshua Greene's team scanned people's brains while they judged dilemmas like these. They argued that up close dilemmas, like the footbridge, engage emotion more than ones like the switch, and that this shapes the answer.
In 2018 the Many Labs 2 project ran Hauser's switch and push pair again with 6,842 people in many countries. The gap held: 71% said the switch was allowed and 17% the push. It was about half as large as in the first study.
The largest study so far collected about 70,000 answers in 42 countries. Every country showed the same order: the switch first, then the loop, then the footbridge. How far people would go changed a lot between countries, most of all on the footbridge.
Then the twist. When Dries Bostyn and colleagues made the choice real, with live mice and a shock the students believed in, more of them pressed than said they would on paper. Their paper answers did not predict their real choice.
Not a test. Your choices do not put you in a type, and there is no right answer. The numbers are what people chose in these studies, in their wording and on their samples. They describe those groups, not you.
Questions
What is the trolley problem?
A thought experiment. A runaway trolley will kill five people unless someone turns it or stops it in a way that kills one. Philosophers use it to ask when it is right to harm one person to spare more. Psychologists use it to study how people decide.
What do most people choose in the trolley problem?
Most pull the lever. In a 2020 study of about 70,000 people in 42 countries, the country average was 81% for the switch, 72% for the loop and 51% for the footbridge.
Why will people pull the lever but not push the man?
One dies and five live either way. But pushing means using your own hands, and his death is what stops the trolley, not a side effect. Greene's 2001 brain scans suggested up close dilemmas like this engage emotion more.
Do trolley problem answers predict what people really do?
Not in the 2018 mouse study. Students' answers to paper dilemmas did not predict whether they shocked one mouse to spare five, and more pressed for real (84%) than said they would when asked to imagine it (66%).
How is this trolley problem game different from others online?
Many trolley games show how other players voted. This one shows numbers from published studies after each choice, and the paper behind every number is in the sources below. It stores nothing and sends nothing.
Does my answer say what kind of person I am?
No. There is no score, no type and no right answer. The page only puts your choice next to what groups of study participants chose.
Sources (5)
- Awad, E., Dsouza, S., Shariff, A., Rahwan, I., and 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 to 2337. doi.org/10.1073/pnas.1911517117
- Hauser, M., Cushman, F., Young, L., Jin, R. K. X., and Mikhail, J. (2007). A dissociation between moral judgments and justifications. Mind and Language, 22(1), 1 to 21. doi.org/10.1111/j.1468-0017.2006.00297.x
- Klein, R. A., Vianello, M., Hasselman, F., and many others (2018). Many Labs 2: Investigating variation in replicability across samples and settings. Advances in Methods and Practices in Psychological Science, 1(4), 443 to 490. doi.org/10.1177/2515245918810225
- Bostyn, D. H., Sevenhant, S., and Roets, A. (2018). Of mice, men, and trolleys: Hypothetical judgment versus real life behavior in trolley style moral dilemmas. Psychological Science, 29(7), 1084 to 1093. doi.org/10.1177/0956797617752640
- Greene, J. D., Sommerville, R. B., Nystrom, L. E., Darley, J. M., and Cohen, J. D. (2001). An fMRI investigation of emotional engagement in moral judgment. Science, 293(5537), 2105 to 2108. doi.org/10.1126/science.1062872
The 2007 Hauser figures are as reported by Many Labs 2. The mouse study figures are from the authors' accepted manuscript at Ghent University. The dilemma wording here is ours.
Updated