The Trolley Problem

Introduction

You have probably heard of it before, as it remains the most famous thought experiment in the history of philosophy. But if you haven't, the scenario goes like this...

The Origins of the Dilemma

This dilemma is a common retelling of the thought experiment originally developed by British philosopher Phillipa Foot in 1967. While today it is often used as a pop-culture reference or a psychological personality test, its original purpose was far more academically rigorous. She introduced the scenario to explore the doctrine of double effect and to ask what we should do when the fundamental interests of human beings directly conflict with one another.

Specifically, she was attempting to navigate the complex moral intuitions surrounding debates on abortion at the time. She wanted to isolate the variables that make us feel comfortable sacrificing one life for another in some contexts, but horrified by it in others. By stripping away the messy emotional details of real-world scenarios, the trolley problem distils moral decision-making down to its rawest mathematical and ethical components.

The Consequentialist Calculation

For many people, the answer to the first question seems incredibly obvious. You pull the lever, you kill one person, and you save five. This aligns perfectly with a philosophical approach known as Consequentialism. Consequentialist frameworks, most notably utilitarianism—championed by philosophers like Jeremy Bentham—argue that the morality of an action is determined solely by its outcomes.

Under this view, the mathematically correct choice is the only ethical choice. You want to choose actions that avoid the most pain and create the most overall pleasure or wellbeing. Five lives hold more utility than one life, so flipping the switch is not just permissible; it is a moral obligation. However, taking this purely calculated approach can quickly lead us down a slippery slope where individual human rights are completely ignored in favour of the greater good.

To pressure-test this consequentialist logic, philosopher Judith Jarvis Thomson posed several brilliant variations of the original problem. If we truly believe that saving five lives always justifies taking one, we have to see if that logic holds up when the context changes.

In this variation, you are forced to take into consideration the accountability and guilt of each person on the track. If you justify killing the five mafia members because they are guilty, where does the calculation end? Would you kill a regular person over a doctor because the doctor has the potential to save more lives in the future? Would you kill an elderly person over a child because the child has more years left to live?

Physical Force and Violations

Things become even more complicated when we introduce physical force and direct bodily violations into the scenario. If you are a strict utilitarian, the mechanism of death shouldn't matter, only the final headcount. But human intuition strongly disagrees.

While most people say they would flip the switch in the original trolley problem, almost no one thinks it is justified for a doctor to murder a healthy patient to harvest their organs, even if it saves five lives. But mathematically, the outcome is identical: one person dies so that five may live. So why does this feel so deeply wrong?

Some philosophers argue that the difference lies in consent. The healthy patient in the waiting room did not consent to the surgery, so the doctor cannot operate. However, as Thomson rightfully points out...

To further highlight this inconsistency, Thomson introduced another famous variation.

In surveys, the vast majority of people say they would absolutely not push the large man. In the original problem, you are merely diverting a threat that already exists. But here, you are actively introducing a new, uninvolved bystander into the threat. You are using his body as a physical tool to stop the train, which feels like a profound violation of his bodily autonomy.

Deontology and Moral Rules

Does the fact that you actively pushed someone make you more responsible for their death than if you simply flipped a switch? This introduces Deontology, an ethical framework that argues some moral rules are unbreakable, regardless of the consequences. Deontological theories, most famously developed by Immanuel Kant with his concept of the Categorical Imperative, establish that certain ethical bounds can never be crossed. He argued that human beings must be treated as ends in themselves, and never merely as a means to an end. Pushing the large man violates this supreme moral principle because it uses him entirely as a tool—a literal meat shield—to achieve your goal. You are stripping him of his autonomy and humanity, reducing him to an object with a calculated physical mass, which Kant would argue is a universal moral failure.

Similarly, the Doctrine of Doing and Allowing states that it is fundamentally worse to actively do harm (pushing a man) than it is to passively allow harm to occur (letting the trolley run its course). This explains the dissonance many feel: flipping a switch is seen as an indirect allowance of harm, while pushing a man is a direct, violent action.

Furthermore, the Doctrine of Double Effect—a concept originating with St Thomas Aquinas—states that it is permissible to perform an action with negative side effects, provided the negative effect is not intended. When you flip the switch, your intention is to save five workers; the death of the one worker on the side track is an unfortunate, foreseen side effect, but not your goal. If that worker miraculously escaped, you would be thrilled! But when you push the large man, his death is inherently part of your plan. If he manages to roll off the tracks and survive, your plan has failed. His death is required for your success.

Real-World Applications

It is incredibly easy at this point to dismiss 'trolleyology' as an abstract, overly sterile academic game with no bearing on reality. Critics often argue that no one will ever find themselves standing next to a lever with a runaway train approaching, and that people's real-world panic responses would differ wildly from their survey answers.

However, these dilemmas are far from fictional. In our modern world, we sit down and program these exact ethical calculations into algorithms every single day. The most prominent example is the development of autonomous vehicles. If a self-driving car's brakes fail, should it swerve into a crowd of pedestrians, or crash into a concrete barrier, killing its own passenger?

Programmers and engineers must explicitly encode a moral philosophy into the vehicle's software long before an accident occurs. MIT's Moral Machine project tested exactly this by asking millions of people worldwide how self-driving cars should prioritize lives in identical scenarios. The results revealed massive cultural differences: some societies overwhelmingly chose to save the young over the elderly, while others valued pedestrians over passengers, or women over men. This proves that there is no universal consensus on the mathematics of human life.

These same ethical friction points appear in medical triaging, such as during the peak of the COVID-19 pandemic when overwhelmed hospitals had to decide who received limited ventilators and ICU beds. Do you give the ventilator to the patient who arrived first, abiding by a rigid, rule-based deontological approach? Or do you give it to the younger patient with the highest statistical chance of survival, following a strict consequentialist logic? Doctors and ethics boards were forced to play out the trolley problem in real time, making life-or-death choices with incomplete information.

The purpose of these extreme edge cases is not to predict behavior, but to expose the inconsistencies in our moral intuitions. They force us to define what we truly value before a crisis hits. By stripping away the noise, we can test the structural integrity of our ethical beliefs and keep ourselves accountable. But given the endless permutations of harm, consequence, and duty, is this dilemma ultimately solvable, or is the 'right' choice inherently unknowable?