The Runaway Trolley and the Surgeon's Dilemma

The Problem Nobody Expected to Matter

Here is a thought experiment that seems almost childishly simple. You are standing next to a lever. A runaway trolley is hurtling toward five workers on the track, and they cannot move in time. If you pull the lever, you divert the trolley to a side track — where there is one worker who also cannot move. Pull the lever: one person dies. Do nothing: five people die.

What do you do?

Most people, when asked, say they would pull the lever. The utilitarian math is obvious: one death is better than five. If you agree — pull the lever — then you have implicitly endorsed a consequentialist principle: the morally right action is the one that produces the best outcome, measured in lives saved.

Now the scenario shifts. You are standing on a footbridge above the track. The same trolley is bearing down on the same five workers. Beside you on the bridge stands a very large man. You realize, in a flash of horrible physics, that if you push him off the bridge, his body will stop the trolley. He will die, but the five workers will be saved.

Do you push him?

Almost nobody says yes. The number is strikingly low — typically around 10 to 15 percent in experimental settings, compared to 85 to 90 percent who would pull the lever. But the arithmetic is precisely the same: one death to prevent five.

Something has changed in the second scenario. The question is: what, and why does it matter morally?

Michael Sandel has made this thought experiment the opening move in his Justice course at Harvard for decades. He doesn't use it to reach a conclusion. He uses it to reveal a problem. Our moral intuitions — those immediate, pre-theoretical reactions that feel like they must be telling us something real — are in direct conflict with themselves. And the conflict doesn't go away when you notice it. In fact, noticing it makes it worse. You cannot reason your way to comfortable resolution. The discomfort is the lesson.

This is not a parlor game. The trolley problem has generated more academic philosophy papers than any other thought experiment in the history of ethics. It shows up in neuroscience labs, in military ethics training, in discussions of autonomous vehicle programming, in debates about triage protocols during pandemics. It matters because it reveals the architecture of moral reasoning — the hidden structure of how we actually think about right and wrong — in a context just simple enough to see the structure clearly.

Before we can argue about justice, rights, punishment, or the ethics of wealth, we need to understand the conceptual tools we are arguing with. The trolley problem is where that understanding begins. It is Sandel's pedagogical masterstroke: a case so simple that you cannot hide behind complexity, so uncomfortable that you cannot retreat to platitudes.

Philippa Foot's Original Insight

19671967

The trolley problem was invented by the British philosopher Philippa Foot in a 1967 paper on abortion and the doctrine of double effect. This is crucial context: Foot was not constructing a puzzle for amusement. She was trying to understand why certain moral intuitions that seem deeply reasonable — like the intuition that you should not kill an innocent person even to save others — resist the clean logic of utilitarian calculation. She was, in other words, trying to explain why most people are not thoroughgoing consequentialists even when they think they are.

Foot's original case was not quite the trolley version we know today. She asked readers to compare a runaway tram that could be steered to kill one or allowed to kill five, against a surgeon who could kill one healthy patient to harvest organs for five dying patients. Her point was that most people feel these cases are morally different — and yet pure consequentialism treats them identically. Five lives saved, one lost: the calculation is the same. But the moral response is not.

❝

"We are not inclined to think that we must apply the same rule in both cases... The distinction we want is between something done to one man and something done to five, where one gets hurt as a by-product of helping the others."

Foot's key distinction was between negative and positive duties. We have a strong duty not to harm people — this is a negative duty, a duty of non-interference. We have a weaker, more qualified duty to help people — this is a positive duty, a duty of assistance. In the lever case, pulling the lever redirects an existing threat. You are not initiating an attack on the single worker; you are allowing harm to be redirected from five to one. In the surgeon case, you are directly killing an innocent person as the means to your good end. The surgeon's hands become the instrument of death in a way the lever-puller's hands do not.

This is the doing/allowing distinction: there is a morally significant difference between actively causing harm and failing to prevent harm, even when the consequences are identical.

Is this distinction philosophically defensible? That is one of the deepest questions in moral philosophy. But notice that the distinction is already doing significant moral work in our actual moral practices. We treat a soldier who fires on civilians differently from a commander who fails to prevent those firings, even when the body count is the same. We hold a drunk driver who kills a pedestrian to a higher standard than a person who fails to install a guardrail on their property where someone later falls. We charge active killers with murder and passive bystanders with nothing, even when both could have prevented the same death.

Whether these distinctions are morally correct is a separate question. The first observation is that they are genuine and widespread — that the doing/allowing distinction is deeply embedded in our moral practices, our legal systems, and our pre-theoretical moral reactions.

Foot herself did not claim the distinction was the final word on trolley problems. She was raising a problem that pure consequentialism cannot account for. Her achievement was to make the problem visible and to give it a name.

The Surgeon's Dilemma in Full

Let us stay with the surgeon case, because it is more disturbing than it first appears — and because the moral structure it reveals is one you will encounter again and again in this course.

You are a transplant surgeon. Five patients are dying from organ failure. Each needs a different organ. In your waiting room sits a healthy person who has come in for a routine checkup. You notice, reviewing their file, that this person is a tissue match for all five of your dying patients. If you were to anesthetize and kill this person, you could distribute their organs and save five lives. By the utilitarian calculus — five lives saved, one lost — this is clearly the right thing to do.

You should not do it. Everyone knows you should not do it. But why not?

This is not a rhetorical question. It is a genuine philosophical puzzle. The consequentialist answer must be something like: the expected consequences of a world in which surgeons harvest healthy patients are far worse than the expected consequences of a world in which they do not. People would stop going to doctors. The overall health system would collapse from loss of trust. The terror inflicted on potential patients would be enormous. This is the indirect utilitarian or rule utilitarian response: the rule "do not harvest healthy patients" is justified by its better long-run consequences.

This response saves utilitarianism from the most horrifying case, but only by making the framework far more complex. Act-utilitarianism — evaluate this specific action, now — says you should harvest. Rule-utilitarianism — evaluate which rules, reliably followed, produce the best outcomes — says you should not. But if the only reason not to harvest is consequentialist, then the framework in principle permits harvesting in a case where there would be no long-run consequences — if the surgeon could be certain no one would ever find out, and the patient were an anonymous stranger passing through. Most people find this even more disturbing than the original case. Something has gone wrong in the reasoning.

The deontological answer is simpler and, to many people, more satisfying: it is simply wrong to kill an innocent person and use them as a tool for other people's benefit, regardless of consequences. The healthy patient is not a bag of spare parts. They are a person. Using them as a means — literally, as an instrument for producing the good outcomes we want — is a fundamental violation of their dignity. This is Kant's second formulation of the categorical imperative, and we will examine it in depth in Unit 3.

The virtue ethics answer asks a different question entirely: what kind of person could look at a healthy patient and see a reservoir of harvestable organs? What has happened to the character of someone for whom this calculus is natural? The virtue ethicist argues that we should be deeply worried about moral frameworks that make such thinking coherent, let alone obligatory. The surgeon who could perform this operation without experiencing it as monstrous has not become more rational — they have become less human, in a morally significant sense.

All three responses point toward the same conclusion: harvesting the patient is wrong. But they reach that conclusion by different routes, and the routes matter — because they will diverge sharply in other cases.

Thomson's Variations: The Problem Gets Worse

Judith Jarvis Thomson, working from Foot's original case, made the problem considerably more uncomfortable in her 1985 paper "The Trolley Problem." Thomson was one of the most rigorous analytical philosophers of the 20th century, and her contribution to the trolley debate is a masterwork of philosophical dialectic: she generated variant after variant, each designed to test whether any single moral principle could explain the full pattern of our intuitive responses.

Her key introduction was the footbridge variant — the large man scenario — which she used to argue against Foot's own solution. If the doing/allowing distinction explains why we should pull the lever, it should also explain our reluctance to push the man off the bridge. After all, pushing him is a clear case of doing harm rather than merely allowing it. The distinction seems to work — so far.

But Thomson pushed further. She constructed what is called the loop variant: the side track curves back and rejoins the main track further along. If the single worker were not there, the trolley would loop back and hit the five. But the single worker's body would stop the trolley and prevent this. In the loop variant, the single worker's death is not merely a foreseen side effect of diverting the trolley — it is the mechanism by which the five are saved. His body stops the trolley, exactly as the large man's body does on the footbridge.

Most people still say pull the lever in the loop variant. But the doing/allowing distinction no longer cleanly explains why. You are doing something (diverting the trolley), and the death of the single worker is the mechanism by which the five are saved — not a side effect at all.

❝

"The question which faces us is this: why is it that Edward may turn the trolley to save five, but David may not drop the man? In both cases one will die if the agent acts, five will die if he does not. If the agent does not act, five will die. What is the morally relevant difference?"

Thomson's conclusion was unsettling: our intuitions are not generated by a single coherent principle. They are messy, context-dependent, and often contradictory. The trolley problem is not revealing a truth we hadn't stated clearly. It is revealing that we don't have a coherent moral theory at all — just a bundle of intuitions that pull in different directions depending on details that we struggle to articulate as morally relevant.

What makes Thomson's contribution to the debate so significant is not just the variants she introduced, but the intellectual honesty of her conclusion. She generated all these cases to find a principle that would explain them — and found that no single principle does the job. The trolley problem resists resolution. It was not designed to have a clean answer. It was designed to show that moral reasoning is more complex than any single framework can capture.

The Doctrine of Double Effect

The traditional philosophical response to cases like the trolley problem is the doctrine of double effect, a principle with roots in Thomas Aquinas and the broader Catholic philosophical tradition. It holds that an action which causes harm may be morally permissible if four conditions are all met:

1. The action itself is not intrinsically wrong
2. The agent intends only the good effect, not the harmful one
3. The harmful effect is not the means by which the good effect is achieved
4. The good achieved must be proportionate to the harm caused

This is why, in the classic analysis, pulling the lever is permissible and pushing the large man is not. When you pull the lever, the death of the single worker is a foreseen but unintended side effect of diverting the trolley — the worker happens to be on the track, but their death is not what you are aiming at; it is a tragic byproduct of saving five. When you push the man off the bridge, his death is not a side effect — it is the means by which you stop the trolley. You are deliberately deploying his body as an obstacle. The doctrine of double effect prohibits using someone's death as a means, even for a genuinely good end.

The doctrine has substantial real-world application, and its influence extends far beyond philosophy seminars.

Just war theory has used the doctrine for centuries to permit bombing military targets while foreseeing civilian casualties, provided those civilian deaths are unintended (not the aim of the operation), unavoidable (not preventable by other means), and proportionate to the military gain. This framework is embedded in contemporary international humanitarian law. The NATO bombing of Serbian military installations during the Kosovo War was considered permissible under this framework even though it produced civilian casualties. Deliberately targeting the civilian population as a means of pressuring the Serbian government would have been an unambiguous war crime. The distinction between the two is exactly the distinction the doctrine of double effect draws.

Medical ethics invokes the doctrine in terminal care. Giving a dying patient high-dose morphine relieves suffering but may hasten death. If the intent is pain relief — and the dose is calibrated for that purpose rather than for killing — the resulting death is a foreseen but unintended side effect. This permits palliative care that would otherwise be indistinguishable from euthanasia. The legal and moral distinction between adequate pain relief and intentional killing tracks the doctrine's third condition: is the patient's death the means, or a foreseen side effect?

Interrogation ethics draws a similar line. Causing pain as a side effect of necessary medical treatment is different from causing pain as the means of extracting information. Torture uses pain instrumentally — the suffering is the mechanism of compliance. The doctrine of double effect prohibits this even when the goal of the interrogation is genuinely important, because the victim's suffering is being used as a means.

The objection that matters most to the doctrine is this: Is the distinction between intended and merely foreseen harm really morally decisive? If I know with absolute certainty that my action will kill someone, does my intention really change the moral weight of their death from their perspective? The person pushed off the bridge and the person diverted into are equally dead. Their families are equally bereaved. The doctrine asks us to believe that the difference in the agent's mental state changes the moral character of the event — which seems to make morality oddly agent-centered in a way that ignores the victim's experience entirely.

The most pointed version of this objection comes from imagining a bomber pilot who destroys a civilian neighborhood but insists he only intended to hit military targets — the civilian deaths are merely foreseen. Has the doctrine genuinely absolved him? Or has it provided a formula that can be used to rationalize almost any harm by carefully managing one's stated intentions? The concern is that "I didn't intend it" can become a moral escape hatch for anyone sufficiently practiced at self-description.

What the Trolley Problem Previews

Sandel's insight — the one that makes this thought experiment pedagogically indispensable — is that the trolley problem doesn't have a correct answer. It has correct observations. It observes that we are implicitly committed to conflicting moral principles, and that any complete ethical theory must account for this conflict or explain it away.

The trolley problem previews the three great ethical traditions that this course will examine in depth:

Consequentialism says: what matters is outcomes. Pull the lever, push the man — five is more than one. The consequentialist is baffled by our reluctance in the footbridge case. We are allowing squeamishness about personal contact to override the obvious calculus. Jeremy Bentham and John Stuart Mill are the canonical consequentialists, and their framework is enormously powerful for thinking about policy, welfare, and collective decision-making. The great strength of consequentialism is its clarity: it tells you exactly what to aim for and provides a metric for evaluating whether you have achieved it. The great weakness is that it can seem to justify almost anything if the numbers are arranged correctly — harvesting organs from healthy patients, sacrificing the few for the many, framing innocent people if it prevents riots. These conclusions strike most people as clearly wrong. That is the central problem with consequentialism as a complete account of morality.

Deontology says: there are absolute constraints on what you may do, regardless of consequences. Kant argued that persons must be treated as ends in themselves, never merely as means — never as instruments for producing outcomes, however good those outcomes may be. Pushing the large man uses him as a means, a trolley-stopper, and violates his dignity as a person. The death of the single worker in the lever case is different because he is not being used; he is simply in the wrong place, and his death is a foreseen but unintended consequence of a permitted action. Deontological thinking undergirds human rights, informed consent, and the rules of war — some of the most important moral achievements of modern civilization. The great strength of deontology is its resistance to rationalization: you cannot argue your way into torture, slavery, or murder by running the numbers. The great weakness is rigidity: when following an absolute rule produces outcomes that seem clearly catastrophic or unjust, deontology struggles to explain why the rule should yield.

Virtue ethics says: the question is not "what rule should I follow?" but "what kind of person should I be?" A person of virtuous character would feel the pull of both lives in the trolley situation — and would carry the weight of whichever choice they made, not dismissing the death with a utilitarian shrug. Aristotle's framework focuses not on the act but on the agent: the dispositions, perceptions, and emotional responses that constitute good character. The virtuous person is not looking for a formula; they are cultivating the practical wisdom to respond appropriately to each situation's particularity. The great strength of virtue ethics is its psychological realism — it attends to who we are becoming through our choices, not just what we do. The great weakness is that it struggles to give action guidance in emergency cases where there is no time for character cultivation, which is precisely the kind of case the trolley problem presents.

None of these frameworks gives us a clean answer to the trolley problem. That is not a failure of the frameworks — it is the data the trolley problem is providing us. Our moral world is genuinely complex, and the frameworks illuminate different features of it.

The Variants That Reveal the Most

Thomson's original loop variant is illuminating, but philosophers have since generated dozens of other trolley variants specifically designed to isolate different moral variables. Each variant is a kind of moral experiment: hold everything constant except one feature, and see how intuitions shift. The pattern of shifts reveals what people are implicitly treating as morally relevant.

The transplant variant (Foot's surgeon): One healthy patient, five dying patients, a tissue match. The math is identical to pulling the lever. The intuitive verdict is radically different. This variant isolates the means/side effect distinction: using someone's death as the means to an end versus allowing a death as a side effect. The lever-puller does not use the single worker's death as the mechanism of rescue; the surgeon does use the patient's death as the mechanism.

The bystander variant with a child: The trolley is heading toward five. A very small child is nearby and, through tragic physics, would stop the trolley if pushed. Most people refuse, even more strongly than in the footbridge case. What if the child were terminally ill and had a week to live? Some people's intuitions shift slightly. What exactly has changed? This variant tests whether the moral weight of a life is affected by its expected remaining duration. Most people feel it should not be — that the terminally ill child's life matters as much as a healthy child's. But then what is the slight intuitive shift about?

The fat villain variant: The large man on the footbridge is the person who intentionally set the trolley in motion, intending to kill the five workers. Is it now permissible to push him? Most people's intuitions shift strongly toward yes. He is just as innocent of doing anything to you. He has the same mass and the same physics. The only thing that has changed is his moral responsibility for the situation and the deaths that will result. This variant suggests that our intuitions about what we may do to someone are deeply affected by judgments about their culpability — which is neither purely consequentialist (his culpability doesn't change the arithmetic) nor purely deontological (he hasn't consented to being pushed).

The double effect variant: You can divert the trolley, and you know that the one worker will almost certainly die, but you do not intend their death — you are targeting only the track diversion. Now consider: you enjoy the fact that the one worker will die, even as you pull the lever for the right reasons. Your action is identical. Your outcome is identical. But most people feel this changes the moral quality of what you are doing, even if they would not say you have acted wrongly. This tests whether intentions and emotional responses are morally relevant even when they do not change outcomes.

The loop with a friend: The loop variant, but the person on the side track is your close friend, and the five on the main track are strangers. Your friend begs you not to pull the lever. Most people's intuitions shift — some toward not pulling, some feel deeply conflicted. This variant tests whether special relationships generate genuine moral weight inside what seemed like a pure utilitarian calculation. It brings care ethics, which we will encounter in Unit 5, directly into the trolley problem.

Joshua Greene's Neuroscience: Two Systems Fighting

In the early 2000s, philosopher and neuroscientist Joshua Greene brought brain imaging into the trolley debate and produced findings that generated intense controversy — because they seemed to suggest that the moral intuitions most people find most trustworthy might be the least reliable guides to moral truth.

When people consider the lever case — diverting the trolley by pulling a lever — brain regions associated with rational deliberation and cognitive control are most active: the dorsolateral prefrontal cortex, the regions associated with logical reasoning, planning, and cost-benefit calculation. When people consider the footbridge case — physically pushing someone to their death — regions associated with emotion and social cognition light up: the medial prefrontal cortex, the amygdala, the regions associated with empathy, disgust, and the processing of social relationships.

The cases feel different because they are differently processed by distinct neural systems. One case activates the calculating part of the brain. The other activates the feeling part.

Greene's interpretation, which remains contested, is that our emotional reactions in the footbridge case represent an evolved response to personal violence — a response calibrated for small-group, face-to-face social life, not for the abstract, large-scale moral problems that modernity creates. The emotional alarm that fires when we contemplate pushing someone off a bridge evolved to prevent the kind of interpersonal violence that destroyed early human communities. It is real, but it may not be tracking anything morally significant in a modern context where we routinely make decisions that affect people we will never see or touch.

❝

"The tension between utilitarian and deontological thinking is a manifestation of a more general conflict between automatic emotional responses and controlled cognitive processing... Our emotional brains were not designed to handle the moral problems of a modern, global society."

This is a profoundly uncomfortable claim. It suggests that the very intuitions that feel most certain — of course I should not push an innocent person to their death — are not reliable guides to moral truth. They are evolutionary artifacts that may actively mislead us when the morally relevant features of a situation are statistical and abstract rather than immediate and personal.

If Greene is right, then the revulsion most people feel at the idea of pushing someone off a bridge is morally irrelevant noise — the same kind of emotional static that led our ancestors to favor their tribe over strangers, to be more disturbed by one visible death than by ten thousand statistical deaths far away. The rational response, Greene argues, is to override these emotions with deliberate calculation when we have good reason to believe they are misfiring.

Deontologists push back hard, and the pushback is philosophically serious. The fact that my emotional aversion to pushing the man has a neurological cause does not mean it has no moral validity. After all, my emotional response to my child's suffering also has a neurological cause. We do not conclude, from the existence of that cause, that my child's suffering doesn't matter. The genetic fallacy — dismissing an argument or belief because of its causal origin — is itself a logical error. The evolutionary origin of a moral intuition does not determine whether the intuition is correct, any more than the evolutionary origin of the capacity for mathematics determines whether mathematics is valid.

There is also a practical argument against Greene's position that has grown more pressing over time. If we train ourselves to override emotional aversion to harm with rational calculation whenever we think the numbers favor it, we are not merely getting better at trolley problems. We are becoming the kind of people who find it easier to use other people as means to our own ends whenever the expected value seems to justify it. And the expected value can always be arranged. The utilitarian calculations that justified the Tuskegee syphilis experiments, forced sterilization programs, colonial resource extraction, and factory farming were internally coherent — the numbers seemed to work. What they lacked was the moral brake that says: there are things you do not do to people, regardless of the arithmetic.

We arrive at a standoff that defines the field: Are our moral intuitions data that ethical theories must account for? Or are they noise that reason should override? This question does not resolve cleanly. It is the central tension in moral philosophy.

Sandel's Method: Intuition, Then Pressure

What makes Michael Sandel's approach distinctive — and what makes it one of the most effective models of philosophical education ever developed — is his insistence that moral intuitions are not things to be explained away. They are things to be questioned. And questioning them is different from dismissing them.

His Socratic method works like this: begin with what students already believe. Then find a case that creates pressure on that belief. Then find a case that creates pressure on the opposite belief. Keep going until students discover that they cannot consistently hold both intuitions — and that this inconsistency is itself a philosophical problem worth taking seriously, not a sign of personal intellectual failure.

The trolley problem is Sandel's opening move because it demonstrates this method concisely and powerfully. You arrive in the classroom with a ready-made intuition: of course it's better to save five than one. Then the footbridge case arrives and destabilizes that intuition. You are now in productive discomfort — holding two commitments that cannot be simultaneously satisfied by any simple principle.

This discomfort is not a failure of philosophy. It is philosophy beginning.

❝

"Sometimes our moral convictions carry more weight than the arguments made against them... If an argument leads to a conclusion that seems clearly wrong, this gives us reason to doubt one of the argument's premises. Sometimes a result is so clearly unjust that the right response is to reject the premises rather than accept the result."

This methodological point is crucial and often overlooked. In mathematics, a valid argument from true premises forces us to accept the conclusion — the logic is binding. In ethics, it is less clear whether this holds. When a seemingly valid argument produces a conclusion that strikes virtually everyone as monstrous, we have two options: (1) bite the bullet and accept the monstrous conclusion, revising our moral views accordingly; or (2) reject one of the premises that led there, because a premise that generates such a conclusion must be false or at least not obviously true.

Philosophers call option 2 tollensing the ponens — running the argument backwards from a rejected conclusion to a rejected premise. If the conclusion is clearly false, at least one premise must be false. The trolley problem doesn't refute consequentialism — but it does put pressure on it at a critical point, forcing consequentialists to either accept conclusions most people find repugnant (harvest the patient, push the man) or develop a more sophisticated version of their view that adds rules, or rights, or threshold constraints.

Sandel's technique is also a demonstration of intellectual respect. He does not tell students what to think. He constructs cases that reveal to students what they already think — and then shows them where those thoughts conflict. The moral work is theirs to do. He just creates the conditions for it.

Real-World Applications: Autonomous Vehicles and Pandemic Triage

The trolley problem is not merely academic. It became a live engineering question when autonomous vehicles became possible and a live clinical question during the COVID-19 pandemic.

A self-driving car must be programmed with decision rules for accidents it cannot avoid. If a brake failure makes a collision inevitable, should the car: (a) continue straight and hit the pedestrians in front of it, or (b) swerve onto the sidewalk and hit a smaller number of pedestrians? Should it protect its passengers at the expense of pedestrians? Should it protect children preferentially? Should it take into account the age, gender, or social status of potential victims?

These are trolley problems, and they must be answered before the car ships. Engineers who have never taken a philosophy course must make decisions that embody consequentialist or deontological frameworks, because there is no option of not choosing. Encoding "protect the passenger at all costs" is a moral choice. Encoding "minimize total deaths" is a different moral choice. Encoding "treat all potential victims equally" is a third moral choice. All three can be technically implemented. None of them is morally neutral.

The MIT Media Lab ran a project called the "Moral Machine" that surveyed people in 233 countries about autonomous vehicle dilemmas. The results were revealing:

Cross-cultural universals: save more lives over fewer; protect children over the elderly; spare pedestrians over jaywalkers.

Cross-cultural variation: collectivist societies showed less preference for saving higher-status individuals; individualist societies showed more preference for protecting passengers over pedestrians; countries with strong rule-of-law traditions showed stronger tendencies to punish jaywalkers (the person who violates the rules bears more responsibility for the consequences).

During the COVID-19 pandemic, hospitals in overwhelmed regions faced genuine triage decisions: when there are more patients in need of ventilators than ventilators available, who gets them? This is a trolley problem at scale, and it was not a thought experiment. Hospitals developed crisis standards of care that involved exactly the utilitarian calculations we have been examining — prioritize patients with better prognosis, exclude patients with conditions that reduce expected survival — and these decisions were controversial precisely because they felt like sacrificing identifiable individuals for statistical benefit.

The trolley problem is preparation for these decisions. Understanding the moral frameworks — consequentialist, deontological, virtue-based — is not academic luxury. It is the conceptual equipment that allows these decisions to be made thoughtfully rather than arbitrarily.

❓

Concept Check

What is the doing/allowing distinction, and how does it apply to the lever and footbridge variants of the trolley problem?

▸

The doing/allowing distinction holds that there is a moral difference between actively causing harm (doing) and failing to prevent harm (allowing), even when consequences are identical. In the lever case, pulling the lever redirects an existing threat rather than directly attacking the single worker — you are not initiating harm against him. In the footbridge case, you are physically acting on the large man's body, using him as the direct means of stopping the trolley. The distinction explains why many people who would pull the lever would not push the man: they feel that actively causing a death is worse than allowing one, even with the same outcome.

Why This Unit Matters

We are not studying the trolley problem because runaway trolleys are a pressing social problem. We are studying it because it forces every student to do something almost no one does voluntarily: examine the structure of their own moral thinking before committing to a position.

Most of us go through our moral lives acting on intuition, occasionally appealing to principles, and rarely noticing when our principles conflict. The trolley problem makes the conflict visible in a context low-stakes enough to think clearly — the stakes are entirely hypothetical, which frees us to follow the reasoning wherever it leads without the emotional pressure of a real decision with real consequences for real people we know.

The reasoning leads somewhere important. It leads to the recognition that the three major ethical traditions are not just different answers to the same question. They are different questions: What are the consequences? What constraints does reason impose? What kind of person am I becoming? These questions cannot easily be combined into a single master theory. But they can be used together, as lenses that illuminate different aspects of moral situations. The task of ethical reasoning is not to pick one framework and apply it mechanically. It is to develop the judgment — Aristotle called it practical wisdom — to know which considerations are most important in which situations.

The trolley problem is the beginning of that development. And it leaves you with something more important than an answer: the recognition that moral thinking is a serious discipline, not a matter of following your feelings or applying a simple algorithm.

The answer-reaching comes in the units ahead.

❓

Concept Check

What is the doctrine of double effect, and how does it distinguish the lever case from the surgeon's dilemma?

▸

The doctrine of double effect holds that causing harm may be permissible if: the action itself is not intrinsically wrong; the agent intends only the good effect; the harmful effect is not the means to the good effect; and the benefit is proportionate to the harm. In the lever case, the single worker's death is a foreseen but unintended side effect of diverting the trolley — not the means by which five are saved. In the surgeon's dilemma, killing the healthy patient is the direct means of obtaining organs to save five — the death is not a side effect but the mechanism. The doctrine permits the first and forbids the second, which is why many people find it captures their intuitions more accurately than pure consequentialism.

The Unresolved Question

Here is where we leave the trolley problem — not with a resolution, but with a sharper question.

Philippa Foot gave us the problem to argue against utilitarianism: our intuitions resist the clean calculation, and that resistance should make us skeptical of any theory that requires us to override it systematically. Judith Jarvis Thomson showed that our intuitions are not generated by any single coherent principle either — which means neither the consequentialist nor the deontologist can simply invoke "our intuitions" in their favor without more work. Joshua Greene suggested our intuitions are evolutionary artifacts rather than reliable moral guides — which might mean we should trust the calculation over the intuition when they conflict. And Michael Sandel insisted that the discomfort of holding conflicting intuitions is not something to escape — it is the beginning of genuine moral inquiry.

You have now been introduced to the key distinction of the course: between positions your intuitions endorse and positions that philosophical argument supports. These often align. When they don't, you face a genuine choice: accept the argument and revise the intuition, or reject a premise of the argument to preserve the intuition. Neither choice is comfortable. Both require a kind of intellectual courage.

The question we carry forward from this unit is not: should you pull the lever or push the man? The question is: what kind of moral reasoning are you willing to trust? And what are the consequences of trusting it?