The Trolley Problem, Explained

A runaway trolley is heading toward five people tied to the tracks. You can pull a lever to divert it — but there is one person on the side track. Do you pull the lever?

This thought experiment was introduced by philosopher Philippa Foot in 1967. It became one of the most studied cases in moral philosophy — and one of the most argued about at dinner tables. The Stanford Encyclopedia of Philosophy treats it as the textbook example of the doctrine of double effect.

The original question

Foot's version asked whether it is ever permissible to do harm in order to prevent greater harm. Pulling the lever kills one person to save five — is that acceptable? Most people say yes. Saving five at the cost of one feels like straightforward arithmetic.

The footbridge variant

Philosopher Judith Jarvis Thomson later introduced a twist: instead of a lever, imagine you are on a bridge above the tracks. You could push a large man off the bridge — his body would stop the trolley, saving the five. The math is identical: one dies, five survive. But almost everyone refuses.

This asymmetry is the interesting part. Same outcome, dramatically different moral intuitions. Why?

The loop case and other variants

Thomson and her successors built dozens of variants to isolate what exactly people are tracking. In the loop case the side track curves back to the main track, so the one person on the loop is the only thing stopping the trolley from continuing on and killing the five — meaning their death is required, not merely foreseen. People hesitate more here than in the original lever case, which suggests the moral weight depends on whether the death is used as a means or simply happens alongside the rescue. The doing vs. allowing harm distinction, the doctrine of double effect, and the agent-relative vs. agent-neutral debate all live inside these variants.

What the research shows

• –In large surveys, roughly 80-90% of people pull the lever — but only ~30% push the man
• –The split maps onto deontological thinking (do not use a person as a means) vs. consequentialist thinking (minimize harm)
• –Joshua Greene's fMRI work (Princeton, 2001) showed the footbridge case activates emotional brain regions; the lever case activates regions associated with cool cost-benefit reasoning
• –Culture, age, and framing all shift the numbers

At population scale, MIT's Moral Machine collected roughly 40 million decisions on trolley-like dilemmas adapted for autonomous vehicles. The dataset showed clear cross-cultural differences in how people weigh age, social status, and pedestrians vs. passengers — variation the researchers grouped into three rough "moral clusters" (Western, Eastern, Southern). The results have been cited in real autonomous-vehicle policy discussions across the EU, Singapore, and the US.

What would you choose?

SplitVote collects real votes on the trolley problem from people around the world. See where you land — and where the global split sits today.

The two ways of thinking it reveals

The split between pulling the lever and pushing the man tracks a real divide in moral philosophy. People who pull but refuse to push are usually mixing consequentialism and deontology — minimising harm when they can stay at arm's length, but refusing to use a person as a tool when the harm becomes personal. Virtue ethics offers a third angle: ask not "which rule wins" but "what would a person of good character do?"

For more on the canonical thought experiments and the research behind them, the Moral Dilemmas hub collects everything in one place. The statistical breakdown covers what real vote data has shown across thousands of respondents.

Educational content, not professional advice.