The Trolley Problem Is Not About Trolleys

The Setup Everyone Knows

Philippa Foot introduced the trolley problem in 1967, and Judith Jarvis Thomson refined it into its now-classic form. A runaway trolley is heading toward five people tied to the tracks. You are standing next to a lever. If you pull it, the trolley will be diverted to a side track where only one person is tied. Do you pull the lever?

Most people say yes. The utilitarian calculus is straightforward: five lives outweigh one. Pull the lever, save the five, accept the moral cost of the one death you caused. The answer feels uncomfortable but defensible.

Now consider a variant. The same runaway trolley, the same five people. But this time you are on a bridge above the tracks, and standing next to you is a large man whose body, if pushed onto the tracks, would stop the trolley and save the five. Do you push him?

Most people say no. Emphatically no. And yet the numbers are identical. One death to save five. The utilitarian logic is the same. Why does it feel so different?

The Doctrine of Double Effect

Philosophers have proposed various explanations for this asymmetry. The most influential is the doctrine of double effect, which holds that there is a morally relevant difference between harm that is a foreseen side effect of achieving a good outcome and harm that is the means by which the good outcome is achieved. In the lever case, the one person's death is a side effect of diverting the trolley. In the bridge case, the large man's death is the mechanism — you are using his body as a trolley-stopper.

This distinction has a long history in moral philosophy and Catholic theology. It is not obviously wrong. There does seem to be something morally significant about the difference between killing someone as a means and killing someone as a side effect. But it is also not obviously right. If the outcome is the same — one person dead, five people alive — it is not clear why the causal structure of the killing should matter morally.

What the Neuroscience Found

Joshua Greene and his colleagues at Harvard ran a series of neuroimaging studies in the early 2000s that produced a striking finding. When people consider the lever case, the brain regions most active are those associated with deliberative reasoning. When people consider the bridge case, the regions most active are those associated with emotional processing — particularly the regions involved in disgust and social pain.

Greene's interpretation was provocative: our moral intuitions are not a reliable guide to moral truth. They are the output of emotional systems that evolved for small-group social life, not for abstract [ethical reasoning](/blog/moral-philosophy-thought-experiments-everyday-decisions). The revulsion we feel at pushing the large man is not a moral insight — it is an evolved response to the act of physically harming a person with our own hands. It is, in Greene's phrase, a "moral illusion."

This is a deeply uncomfortable conclusion. It suggests that our strongest moral convictions — the ones that feel most certain, most non-negotiable — may be the least trustworthy precisely because of their emotional intensity.

What We Reveal Under Pressure

The trolley problem matters not because trolleys matter, but because it reveals the gap between what we say we believe and what we actually do when the stakes are real. In the abstract, most people are at least partly utilitarian — they believe outcomes matter, that saving more lives is better than saving fewer. In the concrete, most people are deontologists — they believe there are things you simply cannot do to a person, regardless of the consequences.

This gap is not hypocrisy. It is the structure of human moral psychology. We carry multiple, partially inconsistent moral frameworks simultaneously, and different situations activate different frameworks. The trolley problem is valuable precisely because it creates a controlled situation in which these frameworks come into direct conflict and we can observe which one wins.

What wins, in most people, most of the time, is the emotional response. The feeling of wrongness overrides the calculation. And that feeling, whatever its evolutionary origins, is not nothing. It may be tracking something real about the moral significance of persons — something that pure utilitarian calculus misses.

Or it may be a bias. The honest answer is that we do not know. And living with that uncertainty — rather than resolving it prematurely in either direction — is what serious moral thinking requires.

Beyond the Thought Experiment: Real-World Parallels

Thought experiments are not ornaments; they are diagnostic tools. The trolley problem is a minimalist probe into perennial ethical tensions, and those tensions crop up in messy real life all the time. Consider some concrete situations where the abstract structure of the trolley problem is present — not identical, never sterile, but structurally similar in the trade-offs they demand.

• Medicine and triage. In wartime or during pandemics, clinicians and administrators make agonizing choices about who receives scarce ventilators, ICU beds, or organ transplants. The classic transplant case (also discussed by Philippa Foot in the same intellectual neighborhood as the trolley) asks whether it is permissible to kill one healthy person to harvest organs to save five dying patients. Most ethicists protest — and rightly so — that medical practice rests on consent and trust; turning patients into involuntary donors would corrode the very system that allows organ transplantation to work. Yet faced with mass casualty incidents, triage protocols intentionally prioritize some lives over others based on prognosis and social function. The moral calculus is deliberate, regulated, and ritualized to minimize the corrosive effects of treating people as means rather than ends.

• Wartime strategy and collateral damage. Military decisions often trade lives in a way that seems utilitarian. The Allied bombing campaigns during World War II, for instance, entailed strategic calculations: bombing factories, railways, and infrastructure to degrade the enemy’s capacity even when civilian casualties were expected. A particularly wrenching historical debate concerned Allied responses to the Holocaust — whether and when to bomb Auschwitz or the rail lines leading to it. Proponents of bombing argued a utilitarian logic: interrupt the trains, save many lives. Opponents pointed to the uncertainty of success, the risk of killing prisoners, and the diversion of resources from other military targets. These were not abstract dilemmas; they were real, with specific causal networks and dreadful uncertainty.

• Autonomous vehicles and algorithmic choice. Perhaps the most public, tangible contemporary instantiation of the trolley intuition has been the ethics of self-driving cars. When a crash is unavoidable, should a vehicle be programmed to minimize total harm even if it must swerve and kill pedestrians? The MIT Moral Machine project (2016) collected millions of judgments worldwide and revealed cultural variation — some societies favored utilitarian outcomes more than others, and judgments varied with who the potential victims were (elderly vs. young, pedestrians vs. passengers). But there is a crucial difference: once programmers hardwire moral priorities into software, those priorities become enforceable, public policy. That converts private moral intuitions into societal rules, and that is precisely when the doctrine of double effect, consent, and the law come charging in.

• Leadership and organizational choice. CEOs and public officials who lay off employees to save an organization face trolley-like trade-offs. The choice to close a plant to prevent broader bankruptcy may doom some to immediate hardship while sparing many others. These decisions are not hypothetical; they have institutional frameworks, legal constraints, and reputational consequences. The moral friction here is the same: treating people as instruments of a larger good corrodes trust in different ways than a tragic, unavoidable side-effect would.

In all these examples we see why philosophers worry about context. The ‘one versus five’ formula misleads us when it abstracts away agency, intention, institutional embedding, uncertainty, and long-term consequences — the very variables that make real moral choices both painful and intelligible.

Philosophical Responses and Alternatives

The trolley problem has catalyzed a small library of philosophical reactions. Below I sketch four families of response, with a few historical anchors.

1. Deontology and Kantian dignity. Immanuel Kant offers the canonical rejoinder to utilitarian instrumentalism: persons are ends in themselves. Kant’s categorical imperative forbids using a person merely as a means to an end. The bridge case outrages Kantian intuitions because you would be treating the large man as a mere instrument for saving others. That stress on dignity has been moral philosophy’s bulwark against modern utilitarianism since the 18th century.

2. Rule utilitarianism and institutional heuristics. If act-utilitarianism suggests we should always maximize aggregate welfare (and therefore push the fat man), rule utilitarianism responds by recommending rules that generally produce better outcomes — rules like “do not kill innocents” because adherence to such rules tends to produce more utility in the long run (trust, stability). This is a pragmatic synthesis: sometimes acts that look sub-optimal in the moment produce worse consequences if they become general practice.

3. Virtue ethics and moral character. Aristotle and the Aristotelian tradition ask a different question: what kind of person does this action mark me to be? Would pushing the man be an action of courage, practical wisdom, or of vicious disregard? Virtue theorists focus on the moral psychology and formation of habits rather than on the atomized weighing of lives. Philippa Foot herself, influenced by Aristotelian thought, saw recognizable moral patterns in our intuitions and resisted reductive utilitarianism.

4. Particularism and moral context. Jonathan Dancy and other particularists argue that moral reasoning is not rule-bound and that particulars matter. From this angle, the trolley problem’s force is limited because the thought experiment strips away particulars that would morally determine the right action in a real scenario (the identity of the people, foreknowledge of outcomes, social roles, etc.). Particularism pushes us to resist the notion of neat moral formulas. It’s an argument for nuance, not nihilism.

There are also hybrid responses — for instance, philosophers who blend deontological constraints with consequentialist sensitivity (Sidgwick wrestled with pluralist intuitions in the 19th century, and contemporary pluralists still try to reconcile incommensurables).

None of these responses completely dissolves the embarrassment. They do, however, move us from the sterile tug-of-war between “calculators” and “gut moralists” toward a richer picture in which rules, emotions, institutions, and character all play roles.

Practical Ethics: Policy, Law, and Machines

If the trolley problem were merely a philosophical parlor trick, we could dismiss it after a few clever rejoinders. But it is not. Its form shows up in policy-making, law, and technological design — areas where decisions get codified.

• Law. The legal distinction between intended harm and foreseen but unintended harm mirrors the doctrine of double effect. In many legal systems, mens rea (the mental state) is critical: intentionally killing is punished far more severely than recklessness or negligence leading to death. The trolley thought experiment helps students see why law distinguishes between killing and letting die, between purpose and side-effect. But legal doctrine also recognizes necessity defenses, duress, and the exigencies of combat — always the messy compromise between moral ideal and lived contingency.

• Public policy. Policymakers routinely face choices with distributive consequences: how to allocate scarce vaccines, which infrastructures to prioritize, or how to contain an epidemic. Such decisions are governed not only by abstract welfare calculations but also by principles of fairness, administrative feasibility, and political legitimacy. During COVID-19, for example, debates about triage and vaccine prioritization invoked both utilitarian calculations (save most lives) and deontological claims (protect the vulnerable or essential workers as a matter of justice).

• Technology and design. Returning to the autonomous car: programmers cannot safely ask each car passenger “what should I do?” in the split-second before impact. The choices made by designers become moral laws enacted at scale. That raises the question of democratic oversight. Should engineers be permitted to encode utilitarian trade-offs into vehicles, or should the law prohibit deliberate programming that would sacrifice one group to save another? Here the trolley problem is not an island; it forces us to talk about regulation, transparency, and public values. If you want to explore how these debate fragments translate into civic engagement and products, see my related essays in the /blog and consider tangible resources in the /shop for further reading.

Practical ethics thus demands not only moral philosophy but public deliberation. We can translate thought experiments into policy — but only if we accept the responsibility of choosing which ethical intuitions to institutionalize.

A Short History of the Thought Experiment

It is worth briefly tracing how this vignette moved from footnote to cultural touchstone. Philippa Foot first formulated the modern trolley-style dilemma in a 1967 paper, situating it within debates about abortion and the doctrine of double effect. Judith Jarvis Thomson (1976) sharpened and popularized variations — including the “fat man” variant — to probe moral permissibility. The elegance of the setup encouraged many philosophers, psychologists, and neuroscientists to test and exploit it as a diagnostic device.

The doctrine of double effect itself traces to medieval moralists — Thomas Aquinas and his commentators — who debated whether foreseen harm that accompanies a good act is morally comparable to directly intended harm. Over centuries, this distinction threaded through Catholic moral theology and into secular moral philosophy. The modern analytic tradition, with figures such as G. E. Moore, Henry Sidgwick, and later modern ethicists like Bernard Williams and Philippa Foot, has used the trolley as a way to make stubborn theoretical distinctions palpably obvious.

In the early 2000s, Joshua Greene and colleagues translated the trolley problem into neuroscience, using fMRI to show differential engagement of emotional and cognitive brain systems. That move was controversial — many scholars worried about the leap from neural activation patterns to moral epistemology — but it opened new interdisciplinary dialogues between philosophers, psychologists, and neuroscientists. In the last decade those dialogues expanded to include computer scientists, legal scholars, and policymakers as the problem leapt from the classroom into the design of machines and institutions.

Conclusion — Why We Should Care

The trolley problem is not about the trolleys. It is about our moral architecture: the competing frameworks we hold, the institutions we design to manage them, and the ways emotion and reason collaborate and conflict. It is about what happens when a private intuition becomes public policy. It is about the humility required to steward moral knowledge responsibly.

If you like these kinds of interrogations — history folded into philosophy folded into human psychology — you might enjoy the novels and essays I offer on the subject. See the /shop for books and essays that take the ethical imagination into historical settings and thrilling plots. Or browse more reflections like this in the /blog.

FAQ

Q: Is the trolley problem a realistic way to study moral decision-making?
A: It’s deliberately unrealistic in many respects — a feature, not a bug. The value of the trolley is that it isolates variables: intention versus side-effect, numbers versus proximity, active versus passive. Those isolations let us see how people’s judgments shift when one parameter changes. Real-life moral decisions include additional factors (uncertainty, identities, institutions) that the trolley omits; that is precisely why philosophers use it as a probe. But to make policy or law, we must translate the insights from such probes back into the messy particulars of lived contexts.

Q: Does neuroscience show that our moral intuitions are unreliable?
A: Neuroscience reveals mechanisms — which brain regions light up under particular circumstances — but does not settle normative questions. Joshua Greene’s work suggests that emotional responses play a large role in so-called “personal” moral dilemmas, whereas controlled reasoning appears in “impersonal” ones. That is descriptively valuable. But whether those emotional responses are reliable indicators of moral truth is a philosophical question, not a neuroanatomical one. Brain data can inform debates about origins and vulnerability to bias, but cannot do the normative lifting by themselves.

Q: Are there cultural differences in trolley judgments?
A: Yes. Cross-cultural studies, including projects like MIT’s Moral Machine, show variation in how societies weight different lives (young vs. old, human vs. animal, law-abiding vs. jaywalking). These differences reflect social norms, institutions, and lived experiences — for instance, societies with stronger communitarian values may prioritize family ties differently than more individualistic cultures. Cultural variation reminds us that moral intuitions are partly shaped by social environments, which strengthens the argument for democratic deliberation when we must convert intuition into policy.

Q: How does law treat distinctions like killing versus letting die?
A: Legal systems typically differentiate based on intention, causation, and duty. Intentional killing (murder) is punished more severely than reckless or negligent killing (manslaughter). Omissions (letting die) are usually treated differently from actions unless there is a legal duty to act. The doctrine of double effect has analogues in law — courts often consider whether harm was intended or merely foreseeable. Still, law introduces practical constraints (burden of proof, institutional precedent, social deterrence) that moral philosophers might not emphasize. The intersection between legal doctrine and moral philosophy is a rich field for applied ethics.

Q: If my intuitions disagree with utilitarianism, does that make me irrational?
A: Not necessarily. Moral reasoning is not a unitary cognitive faculty; it’s an interlocked system of principles, emotions, social rules, and habits. Intuitions that resist utilitarian aggregation may be tracking values that utilitarianism overlooks (rights, personal integrity, relational duties). That does not make one side irrational; it makes moral reasoning plural and pluralistically structured. The better response is to examine why intuitions exist, where they lead when generalized, and whether institutions should embody them. Serious moral thinking tolerates paradox and remains open to revision in light of evidence, reflection, and public deliberation.

If you enjoyed this investigation into moral psychology and the examined life, you might like the historical thrillers and philosophical essays I sell in the /shop, or you can read more essays like this in my /blog.