A runaway trolley is about to kill five people. You can pull a lever and send it onto a side track where it will kill one. Most people say pull. Now the five are in the same danger, but the only way to stop the trolley is to push a stranger off a bridge into its path. Most people say no. The numbers are identical: one for five. The answers differ. Something is deciding, and it isn’t the arithmetic.
That “something” is what ethics, as a philosophy course, is about. It is not a list of dos and don’ts, and it is not a debate about whose opinions are nicest. It is a toolkit of principles, each with a long history, each with an elegant core and a case that makes it squirm. A graduate keeps the toolkit: in a hospital, a boardroom or a family argument they can say, “you are reasoning as a consequentialist here, and the cost of that is…”.
The library already has an easier start in How Should I Live? (05/01), The Trolley Problem (05/05) and Moral Philosophy, Seriously (05/15). This piece goes a step further: it lets you trace which principle your own answers follow, and adds the two families that introductions often skip, care and contractualism.
The idea is easy to state and hard to escape: the right act is the one with the best results. Jeremy Bentham (Introduction to the Principles of Morals and Legislation, 1789) said to count everyone’s pleasure and pain equally; Mill reports his rule as “everybody to count for one, nobody for more than one”. John Stuart Mill (Utilitarianism, 1861) added that some pleasures are better in kind. Henry Sidgwick, Derek Parfit and Peter Singer carried it into the present. Singer’s 1972 essay “Famine, Affluence, and Morality” asks what separates a child drowning in a pond in front of you (saving her costs you some muddy clothes) from a child far away whom a donation could save. If nothing, he argues, we owe far more than we give.
Its strengths: it takes suffering seriously wherever it happens, it explains why we judge a policy by its effects, and it gives an answer when principles collide. Its hardest cases: the surgeon who could kill one healthy visitor to save five patients, the framing of an innocent to calm a riot. If only the total matters, why not? Rule consequentialists (Brad Hooker, Ideal Code, Real World, 2000) reply that we should follow the rules whose general acceptance would do the most good, and a world where doctors might harvest visitors would be a worse one. Act consequentialists reply that our intuitions here are tuned to a world of small groups, not to the logic of the case.
Deontology (from the Greek deon, duty) says that actions have a moral character before you count results. There are duties and there are rights, and they are not just heavy items on the scale. The best-known version is Immanuel Kant’s (Groundwork of the Metaphysics of Morals, 1785). One form of his test: act only on a maxim you could will to be a universal law. Another: treat humanity, in yourself and others, never merely as a means but always also as an end. W. D. Ross (The Right and the Good, 1930) offered a more flexible family of “prima facie duties” (keep promises, repay kindness, do not harm) that you weigh case by case.
Kant’s universalising test is easier to feel than to describe. Take the maxim “when I’m short of money I’ll promise to repay a loan, knowing I won’t.” Suppose everyone did that. Then no one would believe a promise, and the practice that makes the maxim work would vanish. That is a contradiction in conception. Or take “I’ll never help anyone”: a world could run on it, but you could not will it, since you will sometimes need help (a contradiction in will). Move the slider to see how the first one eats itself.
A toy: Kant gave no formula. Here “a promise is believed” and “a stranger helps” each fall as the share of people using the maxim rises. The point is the endpoint.
Without JavaScript: if everyone made false promises, nobody would believe a promise, so the maxim destroys the practice it needs. If nobody ever helped, a person in need would get no help, which the maxim’s own follower could not will.
The critics: Hegel said the test is empty, since you can describe almost any maxim so that it passes; Kant’s own insistence that you may not lie even to a murderer at the door (“On a Supposed Right to Lie from Philanthropy”, 1797) strikes most people as a reductio. Defenders (Christine Korsgaard, Onora O’Neill) reply that the test is meant to expose a kind of practical self-defeat, not to generate a rulebook, and that Kant’s own harsh example is not forced by the theory.
Here is where the theories earn their keep. Take three standard cases: the switch (divert the trolley, one dies, n are saved), the footbridge (push a stranger, one dies, n are saved), and the transplant (a surgeon takes one healthy visitor’s organs to save n patients). Set the number saved, decide what you would do, and see which principles give your pattern. The verdicts are stylised, since each theory comes in versions, and nothing here scores your morality. It shows what each answer would commit you to.
Tap “act” or “refrain” for each case. Rule consequentialism here assumes the best rule forbids killing a bystander even when it saves many; versions with a “prevent disaster” clause would differ.
Without JavaScript: act-consequentialism says act in all three cases once more than one person is saved. A “means principle” deontologist (never use a person as a tool) diverts the trolley but won’t push or harvest. An absolute prohibition on killing refuses all three. A threshold deontologist refuses until the numbers pass a breaking point. A common answer, “switch yes, footbridge no, transplant no”, fits the means principle and the stylised rule-consequentialist.
Two facts frame it. In the 2020 PhilPapers Survey of professional philosophers (about 1,800 respondents in the main group), 63% chose to switch and 13% not to, and when asked about normative ethics, people who could tick more than one answer gave deontology 32%, consequentialism 31% and virtue ethics 37%, so the field has not picked a winner. And the pattern “switch yes, footbridge no” is not just a Western habit: the Moral Machine experiment (Awad and colleagues, Nature, 2018) collected tens of millions of decisions from people in over two hundred countries and found some shared preferences (save more lives, save the young) and real regional differences in others.
Philosophers have proposed several differences between the switch and the footbridge: the doctrine of double effect (Philippa Foot, 1967; the death is a side effect in one and a means in the other), the distinction between redirecting a threat and creating a new one (Judith Jarvis Thomson, 1976 and 1985), and the claim that some of our reaction is gut revulsion rather than principle (Joshua Greene’s imaging work from 2001, itself contested). The honest summary: the pattern is common, the principle behind it is argued over.
Consequentialists ask “what should I do?” and Kantians “what is my duty?”. Virtue ethics asks a different question: what kind of person should I be? Aristotle (Nicomachean Ethics, 4th century BCE) argued that the aim of life is eudaimonia, flourishing, and that flourishing needs the virtues: courage, temperance, justice, practical wisdom. A virtue is a settled trait that gets the response right, in the right amount, to the right things, at the right time. Courage sits between cowardice and rashness, and Aristotle insists the “mean” is relative to us and the situation (NE II.6), not a fixed number.
The mean moves with the case. Running into a fire is rash when the danger is high and little is at stake, and cowardly to hold back when the child you could save is inside and the fire is small. The tool below uses a made-up formula to show the shape of it.
Invented formula for illustration (fitting level = 10 × stakes ÷ (stakes + danger), with a margin of 1.5). Aristotle gave no formula; his point is that practical wisdom, not a rule, finds the mean.
Without JavaScript: with danger 6 and stakes 8, the toy’s fitting level is about 5.7 out of 10; going in at 9 would be rash, staying out at 2 would be cowardly.
Virtue thinking is also a worldwide family. In the Analects, Confucius (551–479 BCE) puts ren, humaneness, at the centre, practised through li, the ritual forms of respect, and by shu, reciprocity: asked for a single word to live by, he answers it is shu, not doing to others what you would not want done to you (15.24). Mencius argued that everyone has the “sprouts” of virtue, and offers the sight of a child about to fall into a well, and the pang we feel, as his proof (2A6); Xunzi replied that human nature is raw and needs shaping by ritual and teaching. Southern African ubuntu thought, in the saying that a person is a person through other persons, grounds virtue in community. Modern Western revival began with Elizabeth Anscombe’s “Modern Moral Philosophy” (1958) and Philippa Foot, and Rosalind Hursthouse’s On Virtue Ethics (1999).
The usual objections: “be virtuous” is thin guidance in a hard case (replies: so is “maximise good”, and wise people do give guidance), and social psychology’s “situationist” challenge that behaviour depends more on situations than on stable traits (see Social Psychology, 35/25, for how much of that evidence held up).
Most moral theories start from a detached chooser weighing options for strangers. Care ethics starts where most of life starts: with dependence. Everyone was once a baby, and many will be old; a morality that forgets this leaves caregivers out. Carol Gilligan’s In a Different Voice (1982) argued that psychology’s picture of moral development had measured everyone against a justice-and-rights ideal and under-rated the ethic of responsibility and responsiveness. Nel Noddings (Caring, 1984), Virginia Held and Eva Feder Kittay (Love’s Labor, 1999) built it into a moral theory, and Joan Tronto’s Moral Boundaries (1993) tied it to politics.
On this view a duty to your mother is not a rule that happens to cover her; it comes from who she is to you. The tension with impartial theories can be put as a dial: how much more do those close to you count? Bentham and Mozi (the 5th-century BCE Chinese philosopher of “impartial care”, jian ai) say not at all. Confucians, including Mencius, who criticised Mozi for it, say love begins with the family and extends outward with differences. Bernard Williams (1976) added that a husband who, rescuing his wife, pauses to check that it is morally permitted has “one thought too many”.
The slider runs from 1× (impartial: Bentham, Mozi) to 1000×. Care ethicists would say no number captures it: the point is the shape of the choice.
Without JavaScript: with weight 1 (strict impartiality) one loved person is outweighed by any two or more strangers. With weight 10 you save your loved one unless more than ten strangers are at risk.
The critiques: care can slide into favouritism and clannishness, and it can be used to tell women they should do the caring; Gilligan’s claim about gender differences in moral reasoning has met mixed evidence, with meta-analyses finding small differences. Care ethicists answer that the remedy is to make care a public and shared task, and that the ethics of everyday dependence is not a minor corner of morality but most of it.
The contract tradition asks what we could agree to live by. Thomas Hobbes (Leviathan, 1651) argued from self-interest; John Rawls (A Theory of Justice, 1971; see The Veil of Ignorance, 05/09) from fairness behind a veil. T. M. Scanlon’s What We Owe to Each Other (1998) turned it into a theory of right and wrong: an act is wrong if it would be disallowed by principles that no one could reasonably reject as the basis for informed, unforced general agreement. The test looks at each person’s complaint separately.
That is why contractualism has a different attitude to numbers. Scanlon’s example: Jones is injured in the transmitter room of a TV station, and rescuing him means cutting the broadcast of a World Cup match for fifteen minutes. A utilitarian can add up the mild annoyance of a vast audience until it exceeds Jones’s agony. Scanlon says no viewer’s complaint is anywhere near as strong as Jones’s, and a principle that left him in pain would be one he could reasonably reject, however many viewers there are. Slide the numbers.
Illustrative units, not measured. Critics (including some who admire Scanlon) ask where a principle that refuses to add can stop: surely enough tiny harms matter. This is a live debate, not a solved one.
Without JavaScript: if each viewer loses 0.01 against Jones’s 100, the sum passes Jones’s harm at 10,000 viewers. Summing says let Jones suffer once the audience is larger than that; comparing individual complaints says rescue him whatever the audience.
So far we have asked what is right. Metaethics asks what kind of thing a moral claim is. Is “cruelty is wrong” true in the way “water is H₂O” is true (moral realism)? Is it an expression of attitude (expressivism, from A. J. Ayer in 1936 to Simon Blackburn and Allan Gibbard)? Is it always false because there are no such facts, as J. L. Mackie argued in 1977 (error theory)? Or is it the result of a procedure rational agents would follow (constructivism, Korsgaard)? David Hume (Treatise, 1739–40) noticed writers sliding from “is” to “ought” without saying why; G. E. Moore’s “open question” argument (Principia Ethica, 1903) says that for any natural property you can still ask “but is it good?”.
The field has an opinion: in the 2020 PhilPapers Survey, 62% of target philosophers accepted or leaned towards moral realism and 26% towards anti-realism. That is a majority, not a consensus, and each position has respected defenders. Cultural variety in moral codes is a fact (the anthropologists’ point); whether it implies that morality is relative is a separate argument, and a weaker one than it first sounds. Ethical disagreement is a fact too; so is wide agreement on a thin core, such as that gratuitous cruelty is bad.
Behind every moral answer is a principle: results, duties, character, care or what could be justified to each person. Each gets many cases right and a few awkwardly wrong. A graduate does not pick one and defend it to the death; they can say which one is speaking, what it costs, and what the others would add.
Try this: the next time you hear “that’s just wrong” or “it’s for the greater good”, ask which of the five principles it follows, and what that principle would say about a similar case you’d answer differently.
Sources. Bentham 1789; Mill 1861; Singer, “Famine, Affluence, and Morality”, Philosophy & Public Affairs 1972; Hooker 2000; Kant, Groundwork 1785 and “On a Supposed Right to Lie from Philanthropy” 1797; Ross 1930; Foot, “The Problem of Abortion and the Doctrine of the Double Effect”, 1967; Thomson, “The Trolley Problem”, Yale Law Journal 1985; Greene et al., Science 2001; Awad et al., “The Moral Machine experiment”, Nature 2018; PhilPapers Survey 2020 (Bourget and Chalmers; trolley switch 63% vs 13%, moral realism 62% vs 26%, normative ethics 32/31/37 among respondents who could select more than one); Aristotle, Nicomachean Ethics II.6; Analects 15.24 and Mencius 2A6; Anscombe 1958; Hursthouse 1999; Gilligan 1982; Noddings 1984; Kittay 1999; Williams, “Persons, Character and Morality”, 1976; Hobbes 1651; Rawls 1971; Scanlon 1998; Hume 1739–40; Moore 1903; Mackie 1977. The interactives’ numbers are invented for illustration.