A Question for Kantians

Recently I gave some lectures on Kant’s moral philosophy in my introductory ethics class. After explaining the “first formulation” of the Categorical Imperative — act only according that maxim which you can will to be a universal law — and working through a few toy examples, I pointed out the common objection that the CI seems often to be indeterminate. In most cases, it seems, a single action will conform to many different maxims, some of which pass the “universalisation” test and some of which fail. If I give the would-be murderer false information regarding the location of his intended victim, am I to reflect on the maxim “tell lies whenever the fancy takes you” or instead on the maxim “tell lies when doing so will prevent a murder”?

My students seemed to catch on quickly, prompting one to ask “doesn’t that make the Categorical Imperative completely useless?” To which, I must confess, I replied “well, yes, I think it does.” So far as I can tell (and perhaps this only reveals my gross ignorance of the literature), the indeterminacy objection is quite devastating.

My question for Kantians, then, is this: what is the current view among Kantians on this issue? Are there those who think the CI, in its first formulation, can be salvaged by proposing some principled way in which to single out one maxim for each action? Or are Kantians more apt to abandon the first formulation and rest their case on the second instead?

40 Replies to “A Question for Kantians

  1. I used to think that the first formulation of the Categorical Imperative just is hopeless for these kind of reasons. Then I met a german Kant scholar who was very devoted defender of Kantian ethics in its purest and original form. He had ingenious ways of dealing with this kind of problems, and as the debates went on I became more convinced that the first formulation of the CI is not as easily disposed of.
    I guess the first thing we need to get clear of is what constitutes a maxim. Kant defines a maxim as ‘the subjective principle of volition’. From the four examples he gives, one can read that the maxim must consist of the situation one is in, the act one is considering, and the end or purpose of that act. That is a maxim must be a sentence of the sort:
    In _____, I_______ in order to ____________.
    Already this hints that strictly speaking neither of the example’s ‘maxims’ actually qualify as maxims.
    On this construal, in the murder case, the natural maxim to test would be the one in which the purpose of the act is to save the friend’s life, because it seems clear that that would be one’s end in the situation (well, of course one’s end could also just be to do what one fancies). And, similarly, in general the maxims would be in part defined by the ends of acts our desires and other natural inclinations suggest to us. The idea of choosing a maxim for an action does not seem to rise in this view as we already seem to be endowed with the potential ends of our acts, conception about the means to pursue those ends and the situation one is in (if it does, this seems to imply that one’s end in addition to whatever end one already is pursuing is the end of acting morally). One thus *has* maxims the testing of which can reveal whether one’s actions have moral worth. If the maxim is universalisable, then there is a corresponding law the respecting of which enables autonomy and morality.
    Of course, the interesting question then is whether the maxim:
    ‘When the murderer is at the door, I lie in order to save my friend’s life’
    is universalizable. Korsgaard has interesting discussion of this in her ‘Right to Lie’. She argues that on Kant’s own criteria it is universalisable, and thus morally ok like our intuitions tell us. This clashes with Kant’s own treatment of the case, and Korsgaard goes in length to explain what the moral ideals were that lead Kant to his unintuitive treatment of the case.

  2. Three Questions to Jussi (and others): does Kant presuppose that whenever I act, there is some maxim that is *my* maxim? On this view, whenever a person acts, there is some fact about that person’s psychology that determines which injunction is in fact that person’s maxim.
    A second question: does Kant need to presuppose this?
    (A third question: is it true that whenever a person acts, that person is following some maxim? I guess the answer to this question is “no”?)
    I thought the answer to the first question is “yes”; this seems to be in line with what Jussi is saying? And this seems to help with the indeterminancy problem, for Kant could always say, “Although there are many maxims that you *could* be following, there in fact is only one maxim that you *are* following, and your action is right only if that maxim could be universalized.”
    I also thought that the second question gets a “yes” answer as well. K says an action is morally right only if the maxim he is following can be universalized. If there is no maxim, there is no maxim to be universalized. Do we want to say that in all such cases the action is impermissible?

  3. Basically I want to agree with the above comments. Onora Nell (now O’Neill) called this problem the problem of “relevant act descriptions” in her Acting on Principle. She claimed that:
    A person cannot simply claim a highly specific maxim…. He must, in fact, intend his act to be contingent on those restrictions and not merely pursued by these means if he is to hold [that] his maxim is [really so] specific….
    Nell’s formulation makes use of what the agent’s intentions really are–does she really plan to not steal from the bank if the bank changes its name, etc. One could, I suppose, instead say that what is relevant is what the agent would in fact do in counterfactual circumstances. I suspect it is better to go with the former as this tests the agent’s plan rather than her dispositions.

  4. I’m not a Kantian, but on the interpretive issue I agree with David and Jussi. I remember reading one of the leading female Kant scholars on this issue in an article where she proposed such a reading. But since I there are quite a few such leading Kant scholars my memory is vague on which one. David’s relating of Onora O’Niell’s book fits my memory.
    What I want to add to what has already been said is an answer to Kris’s question:
    ” I also thought that the second question gets a “yes” answer as well. K says an action is morally right only if the maxim he is following can be universalized. If there is no maxim, there is no maxim to be universalized. Do we want to say that in all such cases the action is impermissible?”
    Since every action involves an intention, every actual action will involve a maxim. For actions that are not done we need a bit more complicated. Sometimes we can know what the maxim would be if we did a contemplated action. In that case, we can use the permissibility of the maxim to evaluate the act.
    For act-types we can think of the proposal as a two stage proposal. In the first instance we divide maxims into the permissible and the impermissible. Then to evaluate an act-type we say that an act-type is permitted if there are permissible maxims that could motivate a person to do acts of that type. An act-type is impermissible if there are no permissible maxims that could motivate one to do acts of that type. More controversially, act-types are required if no maxims that would get you to refrain from doing them are permissible. I think this last bit is needed to get the imperfect duties out. The idea is that a person who never gave to help others must have some reason for not doing so and this reason would be capturable in a maxim that could not pass the test.
    That anyway is how it seems to me that it has to work. I still think that you don’t get a plausible moral theory out of it. But you get further towards that goal than you do without the added suggestion that comes from O’Niell.

  5. My impression of the literature is that the jury is still out on whether there’s a solution to the problem of relevant descriptions, but I’m a bit more hopeful that there’s a solution, which we’re working towards in patchwork fashion in each successive article. The trick seems to be to find a principled reason for singling out one description of an action as the uniquely relevant one. Presumably there is a uniquely relevant description, but Campbell’s question is about the usefulness of the principle, which requires something further, namely a decision procedure for implementing the principle in our actions and assessments.
    So, in Jussi’s example, the end could be either “to save a life” or “to do what one wants” or whatever. I’m not so sure, though, that the fact that “we’re already endowed” with the proper maxim is a principled reason for selecting that maxim as the one to test, particularly since we can deceive ourselves about such things.
    One principled reason for selecting M1 over M2 as the uniquely relevant maxim is offered in Mark Timmons’ article in Jahrbuch fuer Recht und Ethik 5, 1997. There Timmons argues that Kant’s second formulation of the CI, the Formula of Humanity, provides Kant’s answer to the question, “What makes right acts right?”, and that it also provides a principle for relevant descriptions, namely, the relevant information to include in one’s maxim is any and all information that bears on the humanity of those affected by the proposed action.
    Barbara Herman suggests something close to this in ch. 10 of her The Practice of Moral Judgment. In ch. 7 (if memory serves), she also makes a proposal that I’m fond of, namely that the Formula of Universal Law is only supposed to be used to test general forms of maxims (thereby barring the overly specific ones that generate false positives and negatives). That is, FUL is used to get what she calls “deliberative presumptions,” which I’d take to be roughly equivalent to Rossian prima facie duties (though Herman doesn’t call them that). On this view, FUL is only supposed to test general maxims, such as “I will rob a bank when I need money to satisfy my base desires,” or “I will make false promises to secure loans that I know I won’t pay back in order to buy luxurious items.” Assuming we get the correct results on such general maxims, we’ll have general duties, like “don’t rob banks” and “don’t make false promises.” So FUL won’t be used to test our particular maxims for particular actions; instead it will test general maxims. At that point, we’re supposed to engage in something like casuistical reasoning and use our judgment to identify which of the general duties applies in particular cases. Algorigthmic principles are out on this view, but I take that to be a good thing for the theory, as it preserves ordinary intuitions about the need for judgment in moral deliberation.

  6. Small response to Mark:
    Mark wrote, “Since every action involves an intention, every actual action will involve a maxim.”
    This seems too fast to me. Surely every action involves intention. But a maxim is something of the form, “Whenever I am in circumstances x, I will do an act of type y”, right? An intention need not take the form of a maxim, or so it seems to me.
    I might intend to lift the teapot because the water is boiling without forming, endorsing, or following any general injunction about teapots or boling water, or so it seems to me. So I am a little skeptical about the claim that because every action involves intention it involves a maxim.

  7. Kris,
    I think we can say that while we do not consciously formulate all of our intentions in the act-circumstances-end format, all of our intentions can be so formulated. And thus while some of our maxims might be tacit, or even unconscious, every action has a maxim.

  8. Josh – what reason is there for thinking that? It seems to me I can intentionally eat food without acting on any maxim, explicit or tacit, governing all similar situations.
    The assertion that there’s always a tacit maxim seems no more plausible than the assertion by psychological hedonists that, despite all appearances, everyone really does always act to maximize his or her own expected hedonic utility. It’s an empirical, psychological hypothesis for which its backer will accept no disconfirming evidence.

  9. A lot in the comments seem right to me.
    Josh,
    you brought up the option that we might deceive ourselves about the maxims we think we are acting on. That would be true, and fit what Kant had in mind. After all he thought we can never be certain whether someone’s acts, including our own, had moral value just for this very reason.
    On the issue of whether we always act on a maxim I think my intuitions are on your side. It seems to be that whenever we act intentionally we need to have some conception of the situation we are in, the act we are doing and an end in mind to do it for. These needn’t be in mind in explicit conceptual form though.
    One question to everyone. If you were Kantians, what would you say: Can an act have value if the maxim its done out of is just universalisable or need the agent herself realise that the maxim is universalisable before she acts in order to the act to be moral right or good? Kant seems to suggest the latter. In some situation that seems right, but it seems to produce an awful burden of reflection for all acts. Herman’s idea of middle principles seem to help here to some extend. Yet, even with those, one can ask whether the agent may have to remind herself at the time of action that those are *moral* principles and the particular reasons there are for following them.
    To be honest, now that we are discussing the first formulation of the CI, I got to say that my problem with it is more to do with the contradiction part in the universalisation. You read pretty much any commentator and they have their own conception about it. This suggests to me that there is no shared, clear and precise idea about it that would exist before we begin to test the maxims. Everything seems to work the other way round. People have their idea of what kinds of acts they regard as right and wrong, and on the basis of that they form an idea of what kind of a contradiction is in question.
    This seems to make the view slightly circular. Moral intuitions shape the contradiction in the universalisation test to get the right results, and then the test justifies the moral intuitions. [sorry, this is exaggerating slightly for illustrating the worry.] Of course, it is a way of reaching some kind of reflective equilibrium in one’s moral thought, so I guess it might not be that bad.

  10. Ben,
    I don’t think the claim that every action has a maxim is an empirical claim. For Kant and many Kantians, at least, it operates as a conceptual claim: a maxim is simply a description of one’s intention, every action has an intention, so every action has a maxim. So if you intentionally eat food, then there’s a description of what you’re doing (or, more precisely, intending to do), and that’s your maxim.
    Jussi,
    I think you’re right that for Kant we must have morality somewhere in mind in order to be morally good, but I don’t think we have to have any particular motive to act rightly. There are, of course, objections to Kant’s theory of moral goodness, but I think that the distinction between rightness and goodness immunizes Kant’s theory of rightness from such objections.
    Also, I agree that there’s substantial disagreement over the nature of the contradiction in the FUL test. I once tried to sort it all out and I mapped it for a class handout, which, if you’re interested, can be found at:
    http://faculty.oxy.edu/jglasgow/Courses/Phil377-KantsEthics/377Handouts/CITestsAdvanced.htm
    While I’m not sure whether to characterize what everyone’s doing as virtuously or viciously circular, I think it would be more obviously vicious if the argument were “We know X is wrong because the FUL test says so, and we know the FUL test says so because X is wrong.” What’s happening might be more plausibly read as “We know X is wrong because FUL says so, and we’ve tweaked our understanding of FUL to match our ordinary moral judgments (about Y and Z).”

  11. Campbell, I don’t want to interrupt the interesting discussion here, but my sense is that the answer to your question is that the vast majority of kantians have abandoned the first formulation. Those who haven’t completely abandoned it have resorted to hopefull appeals to something called a ‘middle theory’ to get from empty universalisable maxims to what to do here and now. Some drag in a hefty appeal to judgment (i.e., theft), others go more or less ‘Hegelian’ and appeal to each culture’s morals small m, to which universalization tests are to be applied. Rawls’ own version of the test does as well as one could hope I suppose (he’s got a paper on this whose name escapes me at the moment). But, in the end, the second formulation is where most have fled. I don’t see much hope for the first myself.

  12. Hi everyone,
    This is my first post here, and I hope nobody minds my jumping in. I’ve enjoyed reading this blog for a couple of months now.
    I’m interested in the exchange between Ben and Josh on the status of the claim that every intentional action has a maxim.
    It occurs to me that whether an action’s being intentional implies its having a maxim may depend on how general or specific we are in describing actions. If actions are things like “lifting a teapot”, then it seems more plausible that it is a conceptual truth that actions have maxims then if they are things like “lifting that teapot here and now”. Might this point to a difficulty with the “In ___, I____, in order to ___” schematic? If actions are things like “lifting that teapot here and now”, the circumstances-clause (the “In ___” clause) looks otiose. But I’m inclined to regard phrases like “lifting a teapot” as describing action-types, and action-types have neither intentions nor maxims.
    Jason

  13. I’m in agreement with much of what is said above in defending the idea that (according to Kant) every action has a maxim. The basic idea as I think it must go for the view to have any hope of working is that we can use the actual intention of the agent when engaging in the action to generate both a description of the action that fits how the agent it thinking of it, and a motive that encapsulates what the agent thinks the point of doing that action is. I don’t think it is too implausible to think that every intention both involves thinking of an act in some way or other or that it involves seeing the action as having some point. Perhaps this need not be conscious and surely one can be self-deceived about one’s reason for acting. But we can use this to generate something of the form “doing X in order to Y.” And that is then plugged into the universalization test.
    As I said before, I don’t myself think that this together with the rest of the Kantian apparatus for the first formulation generates a plausible moral theory (though I find it a little disconcerting that Robert who I greatly respect seems to agree since I had hopes he might convince me I was wrong). But I thought the original worry had to do with the emptiness of a universalization test if there is not some constraint on how the actions are characterized.

  14. I am no Kant scholar, but I would like to weigh in on th issue of whether acting intentionally requires that you act on some maxim.
    Perhaps Kantians could defend this idea by pointing out that intentional actions are actions that are done for reasons. When I decide to pick up the tea kettle on this particular occassion, I am acting for a reason. One feature of reasons is that there are not completely specific to the particular situation, but are, by their very nature, more general than this.
    Consider an analogy to reasons for belief. If I have a certain amount of evidence for claim P, then any person who is in the same epistemic circumstances has me as the same reason. The reasons for me to believe P cannot be in principle specific to just me on just this particular occassion. They are reasons to believe that P for anyone in relevantly similar epistemic circumstances.
    So if my action is really intentional, and done for reasons, then those reasons must commit me to accepting something more general than just “In this circumstance (here and now) I will lift the teapot.” Any action that is that specific will not be done from reasons, for Kantains, but out of some inclination or other.

  15. I too have noticed that many Kantians are abandoning the first formulation, or, like Herman, asking it to do much less for us than we might have thought it was supposed to do. I was wondering how the second formulation (about not treating people as mere means) could be thought to show what the first formulation seemed to try to show, namely that I in some sense contradict myself or, at any rate, cannot think of my reason to O as a good reason to O if my universalized maxim generates a contradiction.
    What I don’t know (and perhaps I just need to be pointed to some literature) is if Kantians see the shift away from the first formulation as going together with an abandoning of the aspiration to show the above. And if they don’t, I wonder if someone would give me a quick tutorial or point me to papers that explain how the second formulation can vindicate the above thought. Offhand it does not seem self-evident why I am necessarily committed to treating others as an end–this seems like the moral conclusion rather than a non-moral premise in generating the conclusion that immorality is necessarily irrational.

  16. David,
    here is an attempt to answer to your question based largely on Korsgaard’s reading of Groundwork (4-427-8). Now, I am not defending this view myself, and I see that there are big holes in it.
    First, Kant notes that the objects of our desires can have only value conditional to our desires (Kant, the Humean). If we did not desire them, they would not have value for us. Second, Kant notes that the value of being the object of desires is also conditional. We have sometimes mad desires and inclinations causally working in us of which we know that to be the object of them is not to be valuable. Thus, the value of the objects is conditional of them being objects of our rational desires.
    Here it gets tricky. As I read it, the next step is that we see some things valuable in the world. We pursue those things rather than others. This value is conditional on our having rational desires. This means that we must think of our own rationality as an end. Why else would we pursue things we think are valuable? And, if we thought that our rationality was not an end, then on the previous analysis of value, it might be difficult to say how anything could be of value to us.
    This is where there is something missing from the argument. But, I guess Kant would believe in some kind of supervenience theses (see also Nagel’s Possibility of Altruism). If my rationality is a necessary end for me, then if there is no relevant difference between my rationality and the rationality of others, then I must also regard their rationality as an end. To think of my rationality as an end and not that of others would be like thinking one Ferrari to be a great car and another precisely identical to it not. This does smell like a contradiction or like an inconsistency at least. And, thus if I am to take things in the world valuable, I must think of my own rationality as an end, and by the same token the rationality of others.
    I hope I didn’t think too big of a mess of the argument, and, to repeat, yes I am aware of the huge problems it has…

  17. A few tidying up comments:
    To echo Scott: Indeed, most Kantians think of maxims either as reasons for actions or as incorporating reasons for action, and they take it that a reason for action for one agent is, ceteris paribus, a reason for action for another. In this sense, talk of ‘universalizing’ a maxim is somewhat misleading, because a maxim is already a principle that aspires to govern human action in general, making universalization redundant. We can thus certainly speak of maxims as having the form “in circumstances C, I will do X in order that E result,” etc., and we may need to formulate them that way in our actual moral deliberation, so that we can then “move to” the level of considering them as universal laws, but this is not strictly speaking, what a maxim is. In my view, Kant thought of maxims as candidates for the governance of human affairs in the way that scientific laws govern nature. (Ted McNair’s paper, “Universal necessity and contradictions in conception,” Kant-Studien 91 (2000), follows a similar line.) Maxims therefore already include a universalizing element. So to speak of ‘my maxim’ is to refer to whatever candidate universal principle actually motivates me to act, not to the principle qua the actual token psychological object that moved me. (This also lends support to the interpretation of Kant as the arch-externalist in the internalist/externalist debate about practical reasons and desires.)
    I also agree with Jussi that there’s still much to be said about how maxims generate contradictions (and duties). This can be in part traced to the murkiness of what it is to *will* a principle of law. Kant seems to put great weight on how it’s the willing of a principle (rather than some logical, syntactic, empirical, etc. feature of the maxim or the acting thereupon) that generates the contradiction, but I’ve not yet been satisfied by any of the discussions of this within the secondary literature. There is also the problem of matching up the two kind of contradictions Kant identifies with the two categories of duty, perfect and imperfect. Compelling arguments can be made (Herman makes some of them in ‘Murder and mayhem’) that Kant wrongly applies this taxonomy, with the result that duties he treats as imperfect might in fact be perfect, or vice versa.
    Back to Campbell’s original query: When I teach Kant, I fanatically drill into students’ heads the mantra “the CI tests maxims, not acts!”. And as Mark rightly pointed, in order for the CI (under any formulation) to yield categorical requirements on act-types (whether on their performance or non-performance), it would have to be true that every maxim referring to a performance of that act-type fails the CI, or that every maxim referring to a non-performance of that act-type fails the CI. But that seems like an awfully tall order. So why not just interpret Kant as not aspiring to undertake that project?
    My own sense anyway is that Kant meant his ethical theory to be an ethic of maxims, not an ethics of actions. You can search far and wide for examples in Kant of maxims that have the form ‘do/don’t perform type-t acts’. So I consciously resist teaching Kant as a ‘deontologist,’ if that means a theory that thinks of act-types, independently of their motivations, etc., as bearing properties like morally right and morally wrong. I agree wholeheartedly with Herman that reading Kant as a deontologist has been one of the great sources of mistaken and unsympathetic interpretations of his work.

  18. Jussi,
    That is very helpful, thanks. And now that you say that, I can recall hearing others say similar things.
    Just off the top of my head, it at least kinda appears that such an argument reintroduces the argumentative style of the first formulation.
    But perhaps the differences are that here 1) one is held to necessarily be committed to treating one’s own rationality as an end (as opposed to the first formulation case where the maxim need not have been something one was necessarily committed to, and 2) now the thought it that, rather than one’s own commitment to treating one’s rationality as an end entailing that others should treat their rationality as an end, instead one has no grounds for treating likes differently and other’s rationality is like one’s own.
    I would wonder what is wrong with the thought that each person should treat their own rationality as an end for themselves. This does not seem to treat likes differently. Saying that each person should treat their own mother with respect does not treat likes differently.
    (I realize Jussi that you were not defending the argument you gave so please don’t think of these questions as pointed at you.)

  19. Michael,
    I just wanted to second one of the things you wrote (and perhaps all of them, but I’ll focus on one). As I understand the rationale for the universalization test, it is that one’s maxim asserts that it provides sufficient reason to do something. This reason seemingly in principle could not be that one is oneself–that is not intelligible as a reason. One’s maxim must speak of general features of the situation (which could in principle apply to others as well) and claim that anyone who fits those general features would also have reason to follow one’s plan as well. That is the price of thinking that one’s maxim provides one with a reason to do something. Thus the “universalization” stage is just a taking seriously the scope of one’s maxim.
    Admittedly Kant sometimes talks as if there were two stages where the universalization stage is conceptually distinct, but I take it the thought is just that we sometimes fail to notice the full implications of our maxim and the universalization method reminds us of the full implications.

  20. Two small notes on Michael’s post. I’d disagree with the claim that maxims are already universal, and so universalization is redundant. On the alternate reading, maxims are principles for particular actions, and in willing them I am only willing them for the particular actions in question. I am neither willing them for all other agents nor for all other times that I find myself in relvantly similar circumstances. If Kant is right, of course, then in willing a maxim I am rationally committed to extending it to other times and other agents, but that is a further (non-redundant) step.
    Also, while it’s true that Kant focuses on maxims, not actions proper, I don’t think this means he’s not a deontologist. The point of focusing on maxims rather than actions is that moral assessment should focus on what we intend, not whether we succeed in effecting our intentions in action. The deontology/anti-deontology debates can just as likely focus on intention as on action. E.g., a consequentialist who agrees with Kant about the proper target of moral assessment might nonetheless hold that the right maxims just are the ones that are aimed at maximizing good consequences. And a deontologist can deny that the right maxims just are the ones that are aimed maximizing good consequences while still holding that we should assess rightness in terms of maxims, not actions. That doesn’t seem any less deontological, anyway. (That is, I don’t think it’s best to characterize deontology as a theory that assesses acts only.)

  21. Josh,
    One thing that I like about Michael’s reading here is that it comes with an explanation of why we are committed to the universalization of our maxim–roughly, because the maxim already has a universalized form. What I don’t see is why I would count as committed to the universalization of my maxim if my maxim really only applied to me here now.
    Now of course I agree that Kant thought that people could act on maxims even when the univeralization of the maxim would lead to contradiction–so in some sense it must be possible to heed only a portion of one’s (on Michael’s view) general maxim, but still the maxim itself is general because one’s proposed reason is general and only because of this, as far as I can see, is one committed to the universalization of it.

  22. David,
    even though I am not an ‘strict’ Kantian, but rather a Kantian of a modern, ‘wishy-washy’, type, defending Kant is somewhat hobby of mine. So, again, here is a shot at the question you pose – a second argument.
    Imagine a situation in which there were no other rational agents around. And, as a Kantian you thought of your own rationality as your end for the reasons given in the first argument. Now, the requirement for you to be succesful in pursuing this end is that you are able to have rational desires.
    Here comes Kant’s questionable assumption. If your desires are dependent only on the causal forces and natural inclinations acting in you, there is no guarantee that they are rational – most likely they are not. Thus, you need a some way of ensuring that the desires you think are rational are not strictly dependant just on your natural inclinations. At this point, other rational agents come in handy. They can provide an external standard to assess the rationality of your desires. If your desires are such that other rational persons would have them too no matter of their desirative makeup, then the rationality of your own desires does not depend on your desirative makeup. Thus, in order for you to be able to be rational (which necessarily must be your end), you need to have other rational agents around – and thus also they must be your end.
    Of course, there are many problems with this argument too. But, here is an interesting one. This argument, if succesful, would show that you cannot think of only your own rationality as an end, you also need the rationality of others for that end. Yet, it does not establish that you need to regard the rationality of *everyone* as your end, only that of some large enough group.

  23. David,
    I take Kant’s argument to be somewhat different. I see it, very roughly, as holding that we should will (i.e., we are rationally committed to) only universalizable maxims because universalizability is part of the concept of morality. Of course, morality might be an illusion, and it might not really obligate us, and putting those possibilities down is the job of chapter 3 of the Groundwork and the Second Critique. But at the point where Kant introduces FUL, I take his argument to be that if morality rationally obligates us, then since morally permissible maxims are just those maxims that are universalizable, we are rationally obligated to will only universalizable maxims. But in that case, it’s consistent to hold that our maxims, to be mere maxims, need not have any universalizable character.

  24. Josh-
    I don’t think we disagree that deeply, and I may overemphasized the redundancy point earlier. I’d concur that the universalization comes in insofar as maxims are meant not simply as principles of action, but as *rational* principles of action. So perhaps we can think of maxims in terms of the actual principles that motivate particular actions, but when treated as rational principles, the universalization comes in tow with the maxim. (Quite independently of Kant, I think I’d follow David though in thinking that the idea of a purely local rational principle incoherent.)
    On the deontology point: I used to know what ‘deontology’ is. On which characterizations of that view would you class Kant as a deontologist? The distinction between consequentialism and deontology gets very cloudy for me, and I can think of a few ways deontology has been described. Witness this passage from an anthology on ethical theory (Geirsson and Holmgren, Ethical Theory):
    “Deontological theories have a fundamentally different structure from consequentialist theories. … Deontologists hold [that] the right is a more fundamental moral concept than the good. … Deontological theories tell us that certain actions are themselves right and others are wrong regardless of the consequences. We cannot legitimately pursue even the best results in just any way that we choose. The ends do not justify the means. Some action are intrinsically wrong …”
    I hear echoes of several ideas there: D as non-consequentialism, D as constraining our pursuit of valuable ends, D as prioritizing the right over the good, D as identifying certain acts as wrong as such. There’s not any suggestion in that passage that D makes act-types intrinsically wrong, though of course many theories that are thought to have a deontological structure look like that (‘Ten Commandments’ morality). How many of these are apt descriptions of Kant’s ethics?
    I only meant to emphasize above that *if* deontology is the position that rightness or wrongness attach to acts, taken as descriptions of overt behavior, then Kant’s emphasis on maxims renders him not a deontologist, and that supposing that the CI tests acts directly is not only mistaken, but also supposes that the point of the CI is to be a decision procedure wherein acts are tested directly.

  25. I have lots of ideas about much of what everyone is saying here, agree with much, but I’ll only add something regarding David’s worry, as well as to Jussi’s nice reply, re: the argument for the 2nd formula as a ‘demand of reason’. (And maybe i’ll restore Mark’s confidence!) Clearly, you can’t get to that conclusion if your conception of practical reason involves only efficiency in producing ends. An appeal to its ‘intuitiveness’ as a rational principle is hopeless, too, for reasons similar to those David gives as responses to Korsgaard’s version. Some have thought (Korsgaard included) that there is an argument for the 2nd formula buried in Ch. II of the Groundwork, but I think that’s got to be wrong, if for no other reason than the fact that Kant says over and over again that his argument is analytical in ch. II (i.e., he’s only telling us what the moral principle *is*, not justifying it). The only way of getting an argument, I believe, would have to appeal to autonomy in Kant’s special sense. And even then, the argument looks to be daunting. In particular, the appeal to the autonomy of the rational will. (To many, that will look like jumping out of the frying pan.) But the basic idea would be to argue that a rational will just cannot operate except under the idea of its own spontaneity or self-directedness, and that (somehow) requires or involves a kind of respect for the directions that reason lays down. Naturally, the whole story would have to be better than that. And it does look like pulling a rabbit out of a hat sometimes. But maybe there’s hope in that direction for a defensible argument

  26. Josh,
    You said that “I take Kant’s argument to be somewhat different. I see it, very roughly, as holding that we should will (i.e., we are rationally committed to) only universalizable maxims because universalizability is part of the concept of morality.”
    A difficulty I have with this is that it seems to make it opaque what a contradiction in conception has to do with ununiversalizable maxims. On your view, as I understand it, the problem with ununiversalizable maxims is that they cannot be moral. But doesn’t it seem that what Kant is saying is that the problem with such maxims is that they lead to a kind of conceptual contradiction? That seems to be the point of his best examples where the ununiversalizable maxim is ununiversalizable because its universalization leads to impossible states of the world (eg people still making and taking promises even when everyone knows that no one will honor them).

  27. Thanks to everyone for their helpful comments. At least I now have something more constructive to tell my students. But I have another question for Kantians (and others).
    It seems to me that there’s an ambiguity in the standard formulation of the Categorical Imperative. A little formalism will be helpful in showing this. Let a be some action and let x,y, … range over maxims. Write Ra for “a is right”, aCx for “a is done according to x“, and Ux for “x can be willed as a universal law”. Then we can distinguish four versions of the Categorical Imperative:
    (CI1) Ra < -> Ex(aCx & Ux)
    (CI2) Ra < -> ~Ex(aCx & ~Ux)
    (CI3) Ra < -> (Ex(aCx & Ux) & ~Ex(aCx & ~Ux))
    (CI4) Ax(aCx -> (Ra < -> Ux))
    (Note, here “A” is the universal quantifier, and “E” the existential.)
    Now, if we assume the following two conditions, then (CI1) – (CI4) are equivalent. The first condition states that the action is done according to least one maxim:
    (Existence) Ex(aCx)
    And the second says that the action cannot be done according to two maxims, one of which is universalisable and the other isn’t:
    (No Conflict) Axy((aCx & aCy) -> (Ux < -> Uy))
    It’s interesting to note, I think, that if either of these conditions are false, then (C1) – (C4) have different implications. First, if Existence is false, then (CI1) and (CI3) say that a is not right; (CI2) says that a is right; and (CI4) says neither that a is right nor that it’s not right. Second, if No Conflict is false, then (CI2) and (CI3) say that a is not right; (CI1) says that a is right; and (CI4) says both that a is right and that it’s not right.
    Much of the discussion above has focused on whether Existence and No Conflict are true. But it worth pointing out, I think, that even if one or both were false, this might not be such bad news for the Categorical Imperative. Clearly (CI4) is not tenable if No Conflict is false, since in that case it implies a contradiction. But it’s not clear that the any of the other versions are similarly untenable.
    So, my second question for Kantians: which of (CI1) – (CI4) gives the best formulation of the Categorical Imperative? (Or, do you feel that Existence and No Conflict are so obviously true that it makes no difference which formulation we choose?)

  28. Michael,
    Since it took me 3 chapters of my dissertation to try to sort through how to characterize deontology, I’m probably not capable of giving a concise and helpful answer here! (Though, I swear, I’m trying to get it down to one paper.) All I meant to say was that, however it’s characterized (right prior to good, etc.), its principle can be aimed either at acts or at maxims.
    Dave,
    I think that’s exactly right: immoral maxims are those that are incapable of being universalized, and the ones that are incapable of being universalized are the ones that generate contradictions when we try to universalize them. But I don’t take that to be opaque.

  29. Josh,
    It now seems to me that the thing that is clearly opaque is my question. I am not sure I now know what I was trying to ask. I’ll try to think more and see if I can come up with a clearer thing to say.

  30. Let’s suppose, for argument’s sake, that the following is true: for any action A, there exists some maxim M such that (i) the agent’s performing A is consistent with her acting on M, and (ii) M is universalisable. (This may be too strong, but it’ll do for now. I take it that a weaker claim, beginning “For most actions …”, would not be very controversial.) This seems to cause trouble for the Categorical Imperative for the following reason. (I guess this differs a little from the indeterminacy worry suggested above.) Given our supposition, the CI seems to give the agent a free license to do as she pleases. For any action that she desires to perform she’ll be able to find some universalisable maxim to adopt such that her performing the action will count as acting on the maxim and, hence, will be permitted by the CI.
    The favoured Kantian response, I take it, is to say that it’s not as easy as all that to act on a maxim. In particular, you can’t just say that you’re acting on some maxim and thereby make it the case that you are doing so. You must genuinely intend to be acting on the maxim, and you can’t just intend these things willy nilly; something more demanding is involved. (This is mostly inspired by Sobel’s reporting of O’Neill above.)
    So, two questions: (1) Have I got that basically right? (2) What more is involved in order for the agent to act on a maxim?

  31. Campbell,
    this is a tricky issue. But something many Kantians would be likely to say is that there is a much more closer connection between acts and maxims. Some (O’Nell I think at least) would even go onto claim that the identity of the act – which ever act is in question – depends of the agent’s will at the time of acting, i.e., the maxim the agent acts on. Keep in mind that the maxim *is* agent’s conception of the situation, what she does and for what purpose. And, perhaps this is not that unintuitive. Calling to someone at night seems to keep them awake seems different from calling to them to discuss an important urgent matter.
    If this is right, there cannot just be many maxims for an act. (this of course moves the problem to choosing the act – but that seems more ok.)
    Another issue is that the CI is I take it not supposed to be a test for moral rightness of acts, but more to moral worth of them. It analyses the conditions implicitly present in our judgments of moral worth of some acts. These judgments, at least in Kant’s own examples, seem to be backward-looking. They site the ends the agent whose past action is being evaluated had.
    I guess yet another response would be to ask from such a ‘chooser and adobter of a universalisable maxim’, why is she doing this. The answer would have to be along the lines as to *appear moral* towards others. The next question of course is this is *this* universalisable. Can she want to only appear moral and that everyone else does the same – only wants to appear to be moral without really being such? Here I take it the person would run in to contradictions, and thus if the fiddling around with maxims goes on the agent wants to allow special things for herself she denies from others. This, I believe, strikes us wrong.

  32. It looks like this discussion might be wrapping (or wrapped) up, but I wanted to throw two more cents in. I think there are two interpretive claims that we should not abandon:
    (1) The CI is a principle for rightness, not moral worth. This is just to say that the CI determines one’s duty, while moral worth is about the motive one should act on when doing one’s duty–rightness and motive are two different things. Kant’s famous shopkeeper example illustrates the intuitive point that one could do the right thing from a morally neutral (or even bad) motive, and vice versa. And, for Kant, the criteria for rightness and moral worth are different. One acts in a morally worthy way when one acts from the motive of duty, while one acts rightly when one acts in accordance with the CI.
    (2) The CI is a principle for determining the rightness of actions, not merely maxims. Both the Formula of Universal Law and the Formula of Humanity tell us how to act (“Act only on that maxim…” and “Act so that you treat humanity…”). For Kant our actions are to be assessed through our maxims, but I think it’s safe to say that this is still an assessment of the action.

  33. Josh,
    sorry about this. I am still a bit uneasy about (1). Here is one worry. You say that
    (1) “the CI determines one’s duty”
    and
    (2) “one acts rightly when one acts in accordance with the CI”.
    Sentence before (2) you refer to the “criteria of rightness”, and for that reason it is tempting to put (2) as
    (2*) the CI determines the rightness of one’s acts.
    This would mean that the CI determines both what is one’s duty and what is right for one to do. Yet, if the CI test is meant as the same one here all the way through this would mean that what is one’s duty and what is right for one to do is exactly the same thing. This is hard to believe. In the shopkeeper case, if the shopkeeper acts from a bad motive, she can perhaps do the right thing, but it is hard to believe that she is in this case doing her duty. This would mean that the extension of ‘one’s duties’ and ‘what is right for one to do’ differ. Then, the CI cannot be what determines the both differing extensions – it would either have to determine the duties or what is right. Which one would you choose here?
    My bet would be on duties. This would fit the idea that the discussion of the CI takes place in the context of the discussion of moral worth of actions. It would make sense then that if the CI is criteria for what is one’s duty, then the CI also sheds light on the moral worth of actions, when acts have moral worth when done from the motive of duty – the realisation that one’s maxims one can will to be laws.
    But, if this is right, then Kant *would* lack the criterion for acts’ rightness as this would have to be a larger set than what the duties are. Or, if it is not, then he lacks the criterion for what is one’s duty, and then the CI is irrelevant for what moral worth one’s actions have. I am puzzled. This also has consequences for your second interpretative claim.

  34. Jussi,
    Apologies for being a bit sloppy. To clear it up, I was thinking of “right action” and “dutiful action” as roughly equivalent. I want to say that, by giving correct change, the shopkeeper is both doing her duty and (equivalently) doing the right thing. So I guess I don’t share the view that she’s not doing her duty. Why not think this? After all, she acts “in accordance with duty” (to borrow Kant’s phrase), even if her motive is less than admirable.
    Also, I agree with the claim that the CI is insufficient for determining the moral worth of an action. The moral worth of an action is determined by whether or not the agent acts from the motive of duty. Of course, for Kant, acting in accordance with duty (which I take to be acting in compliance with the CI) is not sufficient to act from the motive of duty.

  35. Josh,
    I have to apologise as well. The shopkeeper example may have been a bad one in this respect. But the view that the classes of right actions and duties are equivalent just seems pretty bizarre to me and I find difficulties in finding support for it in Kant. Kant talks little about rightness. Take your (or mine or any normal person’s) average day. During that day we do x amount of actions and x here probably is in hundreds or thousands. I wouldn’t be surprised if all (or at least vast majority) these acts would fall into the category of right acts. But, I would be very surprised if they all (or even most of these acts) were our duties. Take the next opening of the door of the room you are at right now. Surely that needs to be a right thing to do, but it just cannot be your duty. If it was then, there would be a easy way to become an admired moral saint. Just remember to each time be motivated by the idea that your doing your duty, and your every action gets moral worth. That’s just got to be a reductio. So, there seems to be a difference between right acts and duties.

  36. Jussi, it soulds like you want to include both the optional and the obligatory under right action, and only the obligatory under duty. If that’s correct, I’m fine with it. (In fact, I’d better be fine with it, since I’ve tacitly relied on that narrower notion of duty in print!) Still, though, on this understanding, I’d want to say that the CI is a principle of rightness, as it determines whether an act is optional, obligatory (dutiful in the narrower sense), or forbidden. So the CI determines the deontic status of all acts, but still it does not speak to moral worth, since moral worth is about one’s motive and the CI is not.

  37. Josh,
    Yes – all that sounds about right. I think I was confused about quite a few things. Sorry. Anyway, the idea that the CI can determine the deontic status of all acts also requires that there are two different applications or different ways of using the Categorical Imperative – one of which outcome is the set of right acts in the narrow sense (duties (perfect or imperfect), i.e., acts required in one sense or another) and one the outcome of which is the set of right acts in the wide sense (permissible acts).
    I guess my question can now be formulated: If you take the pretty much standard application of the CI – you take your maxim, attempt to universalise it and find out that you can simultaneously will it to be ‘a law of nature’ without contradictions of either sort – which outcome do you get? The narrow or the wide rightness? And, in what other way do you have to apply it to get the other outcome?

  38. Jussi,
    With the standard application of FUL, as you describe it, I think you get both outcomes. If, on the one hand, universalization fails, then acting on that maxim is forbidden, and, furthermore, acting on that maxim’s opposite is obligatory. So, for example, if neglecting the needs of others can’t be universalized, then it’s forbidden, and beneficence is obligatory. If, on the other hand, the maxim can be successfully universalized, then it is at least permissible. To find out whether it is also obligatory, you’d have to test the maxim’s opposite. For example, a maxim of helping others presumably passes the test, so it’s at least permissible. To find out that it’s obligatory, one would have to test a maxim of non-beneficence, where one would find that it is forbidden, making beneficence actually obligatory. By contrasat, if the maxim of wearing dark, instead of light, socks passes the test, and if its opposite passes the test too, then it’s merely permissible to wear dark socks.

  39. The indeterminacy objection…
    Read O’Niell, Conisistency in Ethics, in her Constructions of Reason book.
    Easily evades that objection, as posted.
    More to be said obviously, but she makes the first moves.
    Nick Zangwill

Comments are closed.