“Trolley problems” are philosophical thought experiments where we make an imaginary choice that usually ends in someone getting, well, run over by a trolley.
Here we will use trolley problems to introduce Kantian Ethics, which is the ethical theory developed by Immanuel Kant (1724-1804), and introduce deontological ethical theories in general.
1. Trolley Problems
Five people are tied to a trolley track. An out-of-control trolley is coming down the track. The trolley can’t be stopped, but you can pull a lever to divert the trolley to a side track, saving five but killing one.
What should you do? Most people respond: Switch! We should try to save as many lives as possible.
But consider a modification to this experiment:
The story is the same except now there is no switch and no side track. However, on a footbridge, farther up the track, before the five tied-up individuals, is a precariously perched giant man. If you give the giant a push, he will fatally fall on the track, stopping the trolley and sparing the five tied-up souls.
What should you do? Most people insist: Don’t Push.
But in both cases, it’s one life versus five lives. If we were solely concerned with total lives saved, then Don’t Switch and Don’t Push should be morally equivalent and Switch and Push should be morally equivalent. Are they?
2. Kantian Deontology & the Categorical Imperative
To many people, these actions don’t seem to be equivalent, despite their consequences’ being the same.
Deontology is a type of moral theory that denies that morality is solely about consequences. The most famous deontological theory was developed by Immanuel Kant. Kant’s ethics, and the overall philosophical system in which it is embedded, is vast and incredibly difficult, but we can see his ethics as grounded in a view about what we are, namely persons.
According to Kant, persons are, essentially, rational creatures who are deserving of respect. This rationality grounds what Kant calls The Categorical Imperative, the fundamental ethical rule from which all particular ethical rules derive. This imperative is categorical in that we must follow it, even if we don’t want to.
Kant argues that there is one Categorical Imperative that can be expressed in three different formulations, although the first two are, by far, the most important.
The “first formulation” is based on the idea of a maxim: a principle for acting in a certain way to achieve a certain goal. If a maxim were a universal law, then everyone would act on that maxim. In turn, Kant’s first formulation goes like this:
Act only in accordance with that maxim through which you can at the same time will that it become a universal law.
The basic idea of this formulation is that it’s wrong to make a special exception to some rule for yourself. If your maxim is, ‘I will sneak into a movie without paying, so I can see the movie for free,’ then you are trying to make a special exception for yourself: the maxim wouldn’t work if it were a universal law, because almost no one would show movies, because it wouldn’t be profitable. But of course, ‘I will buy a ticket to a movie so I can see the movie’ is perfectly universalizable: if everyone followed that maxim, movies could still exist.
Its “second formulation” is this:
Act in such a way that you treat humanity, whether in your own person or in the person of any other, never merely as a means to an end, but always at the same time as an end.
This means that we are never allowed to merely use people: people must be respected as ends in themselves. Using people as “mere means” involves treating them as mere objects or in ways to which they could not or would not consent, which is to disrespect them.
3. Applying the Categorical Imperatives
The “second formulation” of the Categorical Imperative helps justify the common trolley-problem intuitions above. In the first experiment, while flipping the switch would kill someone, we are not thereby using the person we kill. His death is an unfortunate and unintended consequence of our attempt to save five lives. In the giant-man variant, pushing him in front of the train does use him—as a mere sack of sinew and bones, no better than a rock. This difference, says the Kantian, accounts for our moral intuitions and the differing moral status of the two cases.
To apply the “first formulation” here we must determine which rules we would be following with different courses of action, such as the rule ‘I will kill to save lives,’ which may not be universalizable.
4. An Objection
A number of objections have been raised to Kantian Deontology, but we’ll review just one.
Kant argued that, given the Categorical Imperative’s “second formulation,” lying always involves using someone as mere means, since that manipulation disrespects their rationality. So, Kant maintained that even if a known murderer asks you where someone is (presumably, to murder them), it would be wrong to lie to the murderer about where that person is. But most people say it’s not morally required to tell the truth to the murderer.
At this point, the Kantian deontologist has two options. They can bite the bullet and insist that lying to the murderer is wrong. Alternatively, they can back off from Kant’s strict deontology to a more moderate deontology, according to which it’s okay, in extreme cases, to break the rules. One might take the latter approach in a way that tries to be compatible with Kant’s other views, by arguing, for example, that helping the murderer would be treating the potential victim as a mere means.
To many, there seems to be something essentially right with a Kantian brand of ethics, i.e., that respect and rationality are fundamentally important for our ethical lives. Nevertheless, Kant’s theory, like all ethical theories, does come with its theoretical and practical challenges.
 These thought experiments are work of hypothetical fantasy meant to stoke your ethical intuitions. Leave aside any implausibilities and notice that these situations are at least possible. Also, assume that all of the tied-up persons are morally equivalent, e.g., it’s not the case that one is a murderer and another is about to cure cancer. Finally, assume that the choices presented are the only possible choices. For the original presentation of the trolley problem, see Foot (1978). For extensive analysis, see Thomson (1976).
 Consequentialism is a moral theory that evaluates actions solely in terms of their consequences. A consequentialist should argue that since the consequences in terms of lives saved are the same in Don’t Switch and Don’t Push and in Switch and Push, we have morally equivalent pairs. See Shane Gronholz’s Consequentialism for an overview of this popular moral theory.
 The Greek root, deont—, means a certain sort of necessity and has been interpreted as duty or obligation. A safe characterization of deontological ethics is that these types of theories alway deny that only consequences matter to determining the morality of actions: deontology can also be understood as non-consequentialism.
 See especially the Groundwork of the Metaphysics of Morals, the Metaphysics of Morals, and the Critique of Practical Reason, all in Kant (1999). While here we will only look at the most famous deontological ethical theory, there are many more deontological theories. Some of them are modifications of Kant’s account, and some of them explicitly non-Kantian. For an overview and taxonomy of such theories, see Alexander & Moore (2012).
 It is a mistake to read “persons” as Homo sapiens. In Kant’s ethical sense, any self-conscious, reasons-responsive, autonomous creature will count as a person in the relevant sense. Accordingly, some non-human animals, e.g., apes, might count as persons. Similarly, potential future artificially intelligent computer systems could count as persons. Finally, some Homo sapiens, e.g., the very young and the severely mentally enfeebled, will not count as persons in the relevant moral sense. For related discussion, see Speciesism by Dan Lowe. Relatedly, you might wonder what it is about persons that makes us so deserving of value. The basic answer, for Kant, is that rational beings are the only creatures that can act with a good will: recognize the existence of moral reasons and act such that one’s reason for acting is that the action is morally required.
 Categorical imperatives, which we must follow no matter what we want, contrast with hypothetical imperatives, rules that we need to follow only if we have certain desires. Consider the rule, “Work on difficult math problems every night.” Someone would have to follow that rule only if they want to learn to learn challenging math: if they don’t want that, they don’t have to follow that rule and study math. A rule like “Don’t lie to other people and manipulate them for your own personal gain,” however is a categorical rule, at least on Kant’s view: you must follow that rule, even if you don’t want to.
 The third formulation is very similar to the first but emphasizes one’s role as choosing or willing rather than merely being bound by rules or the existence of those rules.
 Here we follow Russ Shafer-Landau (2012: 157-9 ff.) in our specific definition of a maxim.
 For example, if someone is going to steal (for fun), their maxim or rule might be “I will steal whatever I want, so I can obtain it.” Can this be willed, as a universal law, that everyone follows?
No. Kant’s explanation for why it cannot be willed, however, is not that bad consequences would follow from everyone doing this, but that everyone doing this would undo the whole idea of stealing in the first place: if everyone could steal whatever they wanted, then nothing really belongs to anyone (nobody really owns anything), and so the whole idea of ownership, which stealing requires, is undercut. On the other hand, the rule “Help people in dire need, when you can, so as to benefit them” can be willed as a universal law: we can will that everyone do that.
 All persons are just like you in their rational capacities, and so are deserving of respect. To treat them differently, all evidence to the contrary, is both to treat them incorrectly by denying them the very thing that makes them moral persons and to lie to yourself, to mis-use your own rationality. Thank you to R. Jenkins and D. Purves for pressing me on this point.
 For example, because everyone would have been killed, so everyone would be dead anyway, so there would be no one to save. Kant’s discussion of why killing violates the first formula is somewhat more obscure, but he says that there couldn’t be a world in which the same principle both preserves and ends life.
Consequentialism by Shane Gronholz
Virtue Ethics by David Merry
John Rawls’ ‘A Theory of Justice’ by Ben Davies
Responding to Morally Flawed Historical Philosophers and Philosophies by Victor Fabian Abundez-Guerra and Nathan Nobis
Indoctrination: What is it to Indoctrinate Someone? by Chris Ranalli
Thank you to Addison Ellis, Ryan Jenkins, and Duncan Purves for very helpful comments on an earlier version of this essay.
This 6/29/2020 revision is of an article originally posted on 6/9/2014. The main update involves including the “first formulation” of the Categorical Imperative. That earlier version is here.
Download this essay in PDF.
About the Author
Andrew Chapman is a lecturer in philosophy at the University of Colorado, Boulder. He holds a PhD in philosophy from the University of Colorado, Boulder, an MA in philosophy from Northern Illinois University and a BA in philosophy and a BM in bassoon and sound recording technology from Ithaca College. He specializes in epistemology, metaethics, and the history of philosophy (especially Kant and the 20th Century Anglophone and Phenomenological traditions). When not philosophizing, Andrew is skiing, hiking, listening to great music, or playing the bassoon.