The trolley problem

In 1967, the Oxford philosopher Philippa Foot published an essay on abortion and the doctrine of double effect in the Oxford Review. To make a point about moral reasoning, she introduced a driver of a runaway tram. The brakes have failed. Ahead, five workers are on the track, and the driver can steer onto a side track where one worker is. Most people say he may steer. She then compares cases in which a surgeon could kill one healthy patient to save five with transplants, or a judge could frame an innocent man to stop a riot. These seem wrong, though they also trade one life for several.¹³

Foot was probing the doctrine of double effect, rooted in Aquinas, which says it can be permissible to bring about a bad effect as a foreseen side effect but not as a means or aim. She also tested the difference between doing harm and allowing it, and between negative and positive duties. Judith Jarvis Thomson took the puzzle further in 1976 and 1985. She gave it the name trolley problem and added new versions: a bystander at a switch, and a fat man on a footbridge who could be pushed to stop the trolley. Most people would flip the switch but not push the man. Why? The utilitarian arithmetic is the same, one life for five, so the difference seems to reflect something about using a person as a means, as Kant forbids, or about directly causing harm.¹²³

The problem has been studied with brain scans, led by Joshua Greene, and with surveys across the world. It became a point of discussion for the ethics of self-driving cars, and MIT's Moral Machine experiment collected about 40 million decisions from people in 233 countries and territories, published in 2018.²⁴