John asked:
How ‘universalizable’ does the categorical imperative need to be, i.e. what counts as a relevantly similar circumstance?
Let’s imagine a woman who is in the middle of her Ph.D. and waist-deep in student debt. She discovers she’s pregnant and wants an abortion. The first categorical imperative seems to forbid this, since if every rational agent had an abortion every time, the human species would go extinct. But I’ve heard and read the first categorical imperative articulated such that it takes into account ‘relevantly similar circumstances,’ entailing, I assume, that we only need to universalize this rule to all rational agents who are in such-and-such a financial/ educational situation.
But how fine-grained should we get about this, given that no two situations are 100% identical? The extreme of this would mean that ‘rules’ would be ‘universalized’ only to particular people in particular places at particular times, e.g. ‘All rational agents with X DNA code at Y location at Z time should commit X action.’ Since this would entail cutthroat self-interest, we’d obviously want to disregard circumstances like these, but on what grounds?
So I guess my question really amounts to: how can one identify relevant circumstances, and thereby determine how general or specific the categorical imperative must be, and can one do this within the categorical imperative (i.e. without resorting to consequentialism, etc.)?
Answer by Geoffrey Klempner
I’ll stay with your example. Kant’s Categorical Imperative in its first formulation states, ‘Act only in accordance with that maxim through which you can at the same time will that it become a universal law.’ In this form it appears identical to the Universalizability Principle (see the previous post The ethics of immigration raids).
However, Kant interprets the Categorical Imperative in a much stronger way, as one can see from later formulations leading up to, ‘Act as a lawmaking member of the Kingdom of Ends’ which makes explicit a strong teleological element that is missing from the first formulation. As this is more controversial, let’s put this on one side for the moment and concentrate on the first formulation.
An error often made with the Categorical Imperative is thinking that it can be applied in the same way as a litmus test (pink for acid, blue for alkali). Just plug in your proposed course of action and see whether it is ‘universalizable’. As you point out, everything turns on what words you use to describe what you intend to do.
Kant talks about the ‘maxim’ of one’s action. It is important that your maxim includes the motive for your proposed action, and not simply a description of the action itself. So, for example, ‘I need an abortion because I don’t want to ruin my figure’, is a different motive from, ‘I need an abortion because I am waist-deep in debt and want to finish my PhD.’
Maybe both of these actions are right, or maybe both wrong. Or maybe one is right and one is wrong — how do we decide?
It is possible to make further distinctions within these two examples, but the point is that each increase in ‘fineness of grain’ has to be linked to a plausible reason. ‘I need an abortion because I want one and its Tuesday,’ will not do, nor will, ‘I need an abortion because I want one and my social security number is 12344321.’ Kant doesn’t state this explicitly, but there is a question of onus involved. You need to give a reason for undertaking the proposed action that is recognizable as a reason, and not just some arbitrary fact like the day of the week or your social security number.
Now we’ve got something to work with. I have no comment to make regarding whether ‘preserving one’s figure’ or ‘finishing my PhD’ are good reasons or not. I don’t recall any place where Kant explicitly discusses abortion but I believe it is quite likely that Kant would have viewed it in the same way as he views telling a lie: it is unjustified in any circumstances whatsoever.
The general point, however, is that rather than applying a litmus test we are challenging the agent, putting the onus on them to give a reason for the proposed action that is defensible as such. Kant was a supporter of the death penalty. So ‘I propose to take a life,’ is an action that is defensible if you are an executioner and the life belongs to a convicted murderer, because it’s your job. On the other hand, ‘I propose to take a life,’ is an action that is indefensible if you are pregnant and the life belongs to your unborn child.
Is that a reasonable view? On the face of it, a ‘defensible reason’ is in the eye of the beholder. What is the purpose of human life? Why are we here? Kant has an answer: we are rational beings and rationality is an end in itself. This leads to the final, teleological formulation of the Categorical Imperative in terms of the ‘Kingdom of Ends’. Without this further element, the Categorical Imperative is toothless on the abortion question.
Ask a Philosopher 
I’m confused about how or why an agent’s reasons for action could be decisive in Kantian ethics, since such reasons, insofar as they delineate what’s more important and what’s less important, pre-suppose ethical judgments that (I assume, perhaps mistakenly) are the subject of ethics. That is, it’s ethics’ job to give such reasons rather than accept them. For example, this PhD student could argue that she and any family she may eventually want would be happier if she finished her PhD and had a better job with which to support her post-doc kid and a greater sense of personal accomplishment with which to inspire him or her; but such an appeal to consequences wouldn’t fly with Kant, for whom consequences are (I understand, perhaps mistakenly) irrelevant to the moral value of an action. Isn’t this why Kant forbids lying for any reason whatsoever, even if one has an incredibly compelling reason to do so? There would seem to be absolutely zero slack cut for what any agent considers to be a plausible reason for action, since the whole purpose of the categorical imperative (and ethics generally) is to inform and evaluate those reasons.
Plus, it seems to me that relegating a reason’s defensibility to “the eye of the beholder” just sets up an infinite regress, since that same beholder could ask herself “What would happen if everyone in a position to judge this action accepted such a reason?” Then she’d have to come up with reasons for accepting the reason, which themselves would have to be validated or invalidated by some further beholder’s gaze. (And I’m assuming that the beholders aren’t allowed to accept reasons on a consequentialist basis, since if this is OK why should the PhD student have bothered with the categorical imperative in the first place? I’m not saying that consequentialism is wrong or anything, just trying to think hard about how the categorical imperative could work entirely on its own terms without being subsumed into some other system.)
The ‘reason for action’ is part of the maxim which is tested by the Categorical Imperative. That’s because the reason identifies the kind of action that it is. In my example, ‘taking a life’ is ethically neutral until you give the reason. So we need to distinguish between ‘reason’ in the sense of ‘in order to’ and ‘reason’ in the sense of justifying an action, e.g. the ‘just’ or ‘kind’ thing to do in the circumstances, or the one that would promote the most happiness. You’re right that the categorical imperative is intended to supply the complete justification, with no further justificatory reason needed.
About ‘consequences’. J.S. Mill claims at the beginning of ‘Utilitarianism’ that he is fleshing out the real meaning of the Categorical Imperative. (I’m not saying he’s right.) R.M. Hare the British moral philosopher who developed a theory of ethics based on the Universalizability Principle argues that the only ‘non-fanatical’ way to apply the Universalizability Principle is via preference utilitarianism. If I’m universalizing, then I have to consider all possible values that agents might have, not just my own values.
So there are two directions in which one might take Universalizability: towards teleology, as Kant does, or towards Hare’s theory. The brilliance of Kant’s idea, is that the whole of ethics ultimately balances on the bootstrapping idea of rationality being an end in itself. In the Kingdom of Ends, all laws, all moral imperatives are designed with that one end in view.
There remains an unresolvable difference between private “situational” principles and larger concerns. From a personal perspective, any abortion is a killing of an unborn human, thus murder. From a more general perspective, if there were no abortions at all, on a finite planet we would see far too many people and so humanity would be doomed (“the Malthus argument”). Thus what may be good and advised “locally” could be the end of all humanity “globally”. When Truman dropped the bomb on Hiroshima, his intention was not “to murder many humans” but to send a strong message. He had consulted many experts on this decision. So “what is a maxim?” It all depends on the perspective. In the end, it is all about “giving reasons as a rational and responsible being”. Which implies that you cannot delegate such decisions to logically arguing robots.