Saturday, March 08, 2008

Complaints on Kant's Categorical Imperative

When I was a freshman, someone once tried to explain the difference between a consequentialist and deontological morality by claiming that the former was concerned only with the results of actions (an action with good consequences is moral) while the latter was concerned with the action itself -- the instructor may have even said the intention, although that is so clearly false that I'm probably purposefully remembering wrong. Yet consequences have everything to do with both of them. Consequentialist is an artificial, modern term which has been imposed on broad groups of historical philosophies. (According to Wikipedia, Elizabeth Anscombe introduced the word -- no citation.) It is misleading.

Both Kant's categorical imperactive and utilitarianism, to use the two most common (or only?) examples of "deontological" and "consequentialist" philosophies, are concerned with one's intentions. Without good intentions, one cannot be moral. The difference lies in the action, and, in particular, the consequences of the action. However, for Kant the abstract, universal consequences are the most important, while for someone like Mill the real (human) consequences take priority. Kant's argument is similar to the slippery slope: if you let one person do it, then everyone is allowed to do it, and then society descends into madness. What he ignores, however, is the laws that prevent everyone from doing it, and that, at least ideally, impose substantial risk upon those who do act immorally.

Kant tries to dress up his moral philosophy in flowery metaphysics: acting in a way that cannot be universalized to all men is logically contradictory, he says. Even if he's right, his categorical imperative is not as strong as he thinks, because an action should be universalized with all relevant factors included. That includes one's position and the circumstances surrounding one's position. Abstraction sheds all irrelevant information, yet in this case nearly all information is relevant. He cites an example of the suicidal man, and then claims that if we let him kill himself, then we would be saying its OK for all of us to kill ourselves, as we all have all suffered misfortune. Yet some of us have suffered far more misfortune than others, and not all of us have suffered enough misfortune to consider killing ourselves. If we allowed the terminally ill to kill themselves, for example (euthanasia), that doesn't lead to no people -- because not all humans will become terminally ill at the same time. Similarly, if all starving men who are able to snatch a bit of bread do snatch that bread, it's not as if there will be no bread left (unless there's a very, very serious famine -- in which case the rich will live and the poor will die). Instead, poor people will generally be barred from grocery stores and arrested fairly often. These examples can be applied in many cases -- even to murder, robbery, drug-dealing, ect. Further, one's position is especially relevant: a ruler of a nation is obviously allowed to do things that the average peasant cannot -- even, arguably, declare war.

Lastly, treating everyone as a end is impossible, but even worse (for Kant), insisting on the categorical imperative as you deal with other people leads to you treating others as a means to an end -- in this case, the end is the categorical imperative, and your friend who you cannot lie to save is simply a means to that end. This contradiction lies on top of the empirical impossibility of the entire idea. All individuals must treat all others as means to ends. Friends are a means to laughter and enjoyment, businessmen are a means to certain business ends, ect. This is no minor problem.

Like most historical philosophy, Kant's imperative is full of holes. But ultimately neither Kant or Mill's conception of the good is truly satisfactory, because the good is a subjective, personal decision. Everyone's conception of what is good will differ to some degree. For a philosopher knowledge may be the highest good; for your average American, pleasure; for an environmentalist, environmental beauty.

5 comments:

ADHR said...

Well, leaving aside the "goodness is subjective" malarkey... ;)

First, generally, you don't find non-utilitarian consequentialists, but they are at least theoretically possible. Hence the difference in terminology.

Second, Kant isn't the only deontologist. Rawls is a deontologist. So is WD Ross. And Ronald Dworkin. Contractarian/contractualist ethics sometimes gets dropped in with deontology, so Locke and Hobbes could count as well.

Third, you're misreading Kant, I think. The idea that the first formulation of the CI relates to logical contradiction isn't "flowery metaphysics" -- it's not metaphysics at all, really. The issue is that, for Kant, some actions if universalized will contradict the possibility of those actions. So, for example, lying. You shouldn't lie, says Kant, because if everyone lied, there would be no sense in lying. Lying, he says, to be coherent requires a general practice of truth-telling. Similarly for suicide. If suicide were universalized, the point of suicide would be negated. There wouldn't be any lives, misfortunate or not, to escape from.

Moreover, according to the second formulation of the CI, you have to treat all rational beings such that you respect their autonomy. And suicide destroys autonomy, it's alleged, because it prevents autonomy from being exercised in future.

You're right, though, that there's a real problem with how we formulate the duty we're trying to universalize.

Fourth, when it comes to treating others as ends, you're not supposed to be acting for the sake of the CI, but in accordance with the CI. (Although I don't honestly recall exactly Kant's language on the point.) The idea is that you shouldn't slavishly follow the (moral) law for the sake of following the moral law, but should instead act from respect for the moral law. This amounts to requiring that you act as the law requires because, in essence, you have internalized the law as the standard for your own conduct. Furthermore, Kant is clear here: the problem is treating people merely as means. Treating people as a means and also as an end is okay; as you note, it's pretty essential.

Fifth, Kant isn't giving a theory of the good as such. He's giving a theory of the right. That's the fundamental difference between consequentialist and deontological moral theories.

undergroundman said...

First, generally, you don't find non-utilitarian consequentialists, but they are at least theoretically possible. Hence the difference in terminology.

Well, sure. I hear that a lot, but I've yet to find an example. So what is one?

Second, Kant isn't the only deontologist. Rawls is a deontologist. So is WD Ross. And Ronald Dworkin. Contractarian/contractualist ethics sometimes gets dropped in with deontology, so Locke and Hobbes could count as well.

What does it mean to have a deontological system of ethics? Deon, according to Wikipedia, means "obligation or duty". Yet under that definition, all ethical systems are duty-based: utilitarians simply believe that one has a duty to maximize the good. What I've read about John Rawls doesn't seem duty-based at all; it's more political and focused on distribute justice. How is contract ethics similar to deontological ethics? One has a duty to hold up their side of the bargain?

Similarly for suicide. If suicide were universalized, the point of suicide would be negated. There wouldn't be any lives, misfortunate or not, to escape from.

I had two points. The first is that Kant believes there is a (metaphysical, in that it is so fundamental that it must be followed regardless of practical concerns -- his universalization is really a slippery slope argument) contradiction in making an exception out of ourselves. The second is that, even assuming he's right, you can't universalize while abstracting away the particular circumstances. Thus, allowing terminally ill people to commit suicide doesn't lead to everyone being dead.

Fourth, when it comes to treating others as ends, you're not supposed to be acting for the sake of the CI, but in accordance with the CI.

How is that not a semantic difference?

Furthermore, Kant is clear here: the problem is treating people merely as means. Treating people as a means and also as an end is okay; as you note, it's pretty essential.

Are you sure those two are compatible? I think the proper distinction is that you can treat people as a means in ways that are harmless (as friends, for humor ect.) or in ways that harm the person. In both cases you are treating people as a means. I suppose you might say that in the former case you are treating the person as an end as well, but I don't see how.

Fifth, Kant isn't giving a theory of the good as such. He's giving a theory of the right. That's the fundamental difference between consequentialist and deontological moral theories.

Another semantic difference. Kant (and especially modern scholars) may say that he's making a theory of the "right" (though I don't recall him using that word), but there's no real difference between that and "the good". His entire theory is based around the consequences of actions universalized.

Universalizing does make some sense, if you take into account circumstances. We're seeing the results of bad actions universalized right now in the real world, actually, because people in the modern world (both you and I) do things that, if everyone in the world did, would not work. The natural resources simply are not there to support the lifestyles of people in developed nations' universalized.

When it comes down to it, ultimately, it may be that most of my aversion to these ethical systems is that I just plain don't like restrictions on my behavior. So I may not be purely logical in my arguments -- but I still think there's a lot of flaws in Kant's argument. In particular, the fact that he discards relevant information while universalizing is wrong.

ADHR said...

Well, sure. I hear that a lot, but I've yet to find an example. So what is one?

Why do you need one? If it's possible to be a consequentialist and not a utilitarian, then they are distinct ethical theories. And hence anyone critiquing one may or may not be critiquing the other.

What does it mean to have a deontological system of ethics? Deon, according to Wikipedia, means "obligation or duty". Yet under that definition, all ethical systems are duty-based: utilitarians simply believe that one has a duty to maximize the good. What I've read about John Rawls doesn't seem duty-based at all; it's more political and focused on distribute justice. How is contract ethics similar to deontological ethics? One has a duty to hold up their side of the bargain?

Appeals to etymology are often misleading. A deontological ethical system is one which focuses on qualities of the action itself rather than consequences. Hence why Rawls counts as a deontologist: some things, for Rawls, are just plain wrong even if they would have good consequences (e.g., trade-offs of basic liberties for improvement in the lot of the worst-off). The same point applies to contractarian/contractualist theories: the idea is that the contract makes some actions just plain wrong, regardless of consequences.

I had two points. The first is that Kant believes there is a (metaphysical, in that it is so fundamental that it must be followed regardless of practical concerns -- his universalization is really a slippery slope argument) contradiction in making an exception out of ourselves. The second is that, even assuming he's right, you can't universalize while abstracting away the particular circumstances. Thus, allowing terminally ill people to commit suicide doesn't lead to everyone being dead.

How to formulate the maxim is a problem with Kant's system. The maxim "terminally ill people can commit suicide" seems universalizable. However, you're still misreading Kant's point: this is not a slippery slope argument. Slippery slope arguments trace bad consequences of accepting some claim or norm. That is, "if A, then B; but B is bad; therefore, not-A". That's not Kant's argument. Kant's argument is, approximately, "if A, then not-B; but if not-B, then not-A." That is, what follows from universalizing some maxim is the negation of the conditions which make the action the maxim is a maxim of make any sense. So, from universalizing the maxim it follows that the action which expresses the maxim is impossible. Not that it's bad. That it can't happen.

Fourth, when it comes to treating others as ends, you're not supposed to be acting for the sake of the CI, but in accordance with the CI.

How is that not a semantic difference?


Because, for Kant, acting for the sake of the CI is a kind of moral fetish. Doing the right thing, whatever the right thing is. Instead, you should act out of respect for the moral law, which -- consequently -- means you do what the CI says.

Are you sure those two are compatible? I think the proper distinction is that you can treat people as a means in ways that are harmless (as friends, for humor ect.) or in ways that harm the person. In both cases you are treating people as a means. I suppose you might say that in the former case you are treating the person as an end as well, but I don't see how.

You're allowed to harm people, in Kant's system. Remember, it's not utilitarianism: harm isn't at issue. What you're not allowed to do is violate someone's autonomy, i.e., make decisions for them as if they were things. That's actually the best way to capture the second formulation: never treat people as if they are just things.

Another semantic difference. Kant (and especially modern scholars) may say that he's making a theory of the "right" (though I don't recall him using that word), but there's no real difference between that and "the good". His entire theory is based around the consequences of actions universalized.

Kant doesn't say that, no, but that's how it gets classified by later commentators. As I pointed out above, though, it's got nothing to do with consequences of actions. It's got to do with conditions for actions, and what universalizing some actions does to those consequences.

When it comes down to it, ultimately, it may be that most of my aversion to these ethical systems is that I just plain don't like restrictions on my behavior. So I may not be purely logical in my arguments -- but I still think there's a lot of flaws in Kant's argument. In particular, the fact that he discards relevant information while universalizing is wrong.

That is a reasonable complaint. As said, how to formulate the maxims is not made totally clear by Kant, and it's an issue that would need to be resolved before his system could work.

undergroundman said...

If it's possible to be a consequentialist and not a utilitarian, then they are distinct ethical theories. And hence anyone critiquing one may or may not be critiquing the other.

Unless you can find an example, how can I know that it is possible?

So, from universalizing the maxim it follows that the action which expresses the maxim is impossible.

That's not how it sounded when I read it. Rather, he seemed to say that it would rather be bad. Even if he didn't say it, it is, in effect, the problem of immoral actions. Besides, it's not impossible for, say, everyone to lie. It would simply cease to be useful -- that doesn't mean it's impossible; that it can't happen. Certainly we could have a world in which everyone lied. It would just be a damn confusing world. Ditto for the others.

Let's look at another example. Say we universalize our developed world lifestyle -- but let's say that such a lifestyle wouldn't be possible in that case. Thus, it is unacceptable under CI. What do we do? Become, basically, peasants?

How about universalizing the life of a university lecturer? If everyone was to be a teacher, no farming could be done. Thus, that's impossible. In fact, any profession is impossible under the CI, at least in the modern world, because we have this necessary division of labor.

Also, if "terminally ill people can commit suicide" is universalizable, then why is "depressed people can commit suicide" not? In fact, as long as you generalize it for a certain subset of people and not the whole population, it seems as if they are all universalizable. Deciding which group can and which group cannot is a value judgment, not some sort of mathematical judgment (as Kant would like it to be -- Kant would like to say nobody, including the terminally ill/paralyzed/ect., can commit suicide). Kant wants it to be this nice, neat thing, but it's all just individual value judgments.

ADHR said...

If it's possible to be a consequentialist and not a utilitarian, then they are distinct ethical theories. And hence anyone critiquing one may or may not be critiquing the other.

Unless you can find an example, how can I know that it is possible?


In your previous comment you accepted it was. Are you retracting that? ;)

The key distinction rests on how broadly you want to read utility. Since it's a term of art, you could, in principle, just define it as consequences and call the distinction moot. Of course, this would mean overlooking the differences between, say, Jeremy Bentham and Thomas Hurka. Here is a good SEP article on the varieties of consequentialism and utilitarianism. FWIW, depending on how it was formulated, ethical egoism could be seen as a non-utilitarian consequentialism, insofar as maximizing what benefits oneself is seen as a moral rule.

That's not how it sounded when I read it. Rather, he seemed to say that it would rather be bad. Even if he didn't say it, it is, in effect, the problem of immoral actions. Besides, it's not impossible for, say, everyone to lie. It would simply cease to be useful -- that doesn't mean it's impossible; that it can't happen. Certainly we could have a world in which everyone lied. It would just be a damn confusing world. Ditto for the others.

It's often hard to figure out how the first formulation of the CI works, especially the way the Groundwork sometimes gets excerpted. Here is another SEP article, specifically on the first formulation. The issue with lying is that lying just doesn't exist without a general practice of truth-telling. It's not lying, as such, unless there's a distinction between lying and telling the truth. If there's no distinction, because truth-telling doesn't exist, then there's really no lying any more.

Let's look at another example. Say we universalize our developed world lifestyle -- but let's say that such a lifestyle wouldn't be possible in that case. Thus, it is unacceptable under CI. What do we do? Become, basically, peasants?

How about universalizing the life of a university lecturer? If everyone was to be a teacher, no farming could be done. Thus, that's impossible. In fact, any profession is impossible under the CI, at least in the modern world, because we have this necessary division of labor.


Sure, but a lifestyle and the life of a certain profession aren't actions. I can't will, at all, "live a certain lifestyle" or "be a university lecturer". The first formulation of the CI -- indeed, all of them -- only applies to particular actions.

Also, if "terminally ill people can commit suicide" is universalizable, then why is "depressed people can commit suicide" not? In fact, as long as you generalize it for a certain subset of people and not the whole population, it seems as if they are all universalizable. Deciding which group can and which group cannot is a value judgment, not some sort of mathematical judgment (as Kant would like it to be -- Kant would like to say nobody, including the terminally ill/paralyzed/ect., can commit suicide). Kant wants it to be this nice, neat thing, but it's all just individual value judgments.

I'm not sure how you got value judgements in there. There's certainly a hole, in that Kant needs to tell us how to figure out what the maxim of our action is. He thinks it's obvious, but it's clearly not. Why that's a "value judgement" is not clear to me, though.