Like all of Kant's writings, The Foundations of the Metaphysics of Morals is in serious need of explanation. It is in need of explanation, first, how the several major distinctions and claims Kant makes -- each of which can be, and has been, the subject of whole articles of commentary in its own right -- fit together into a unified whole. I am thinking of the distinctions between a priori and empirical knowledge, appearances and things in themselves, hypothetical and categorical imperatives, and heteronomy and autonomy, for instance. Second, one or two odd statements of Kant's which have become the objects of attempts at reinterpretation by commentators, in conscientious attempts to make Kant reasonable, are in need of explanation. I am thinking here, in particular, of the claims that the only thing good 'without qualification' is a good will and that only an action which one had no desire to do could ever have any moral worth.
Once Kant's theory of morals has been (hopefully) made clear, we will be able to proceed with criticism of it, that is, with stating some of the numerous objections against his views and the elliptical arguments for them that he gives.
At the outset, I should make a brief comment on methods of interpretation. In exegesis, the so-called principle of charity is frequently invoked on behalf of famous philosophers. In contrast, I give primacy to the strong 'he said what he meant' principle. This is the principle by which one attempts to take the most simple and direct reading of a passage, assuming that the author was using language with its standard meanings. This approach is based (in part) on the conception of interpretation as a cognitive endeavor, viz., to find out what a certain person was thinking as he wrote a certain book. It is in contrast to the idea of interpretation (central to contemporary literary analysis) as the attempt to come up with the most interesting and clever thing that a person might have said. Nor do I see reason to assume a priori that great figures in intellectual history are any more immune to mistakes and contradictions than the rest of us. This is not the place to elaborate on these points, but great thinkers need not be particularly logical or correct; usually, they are great mainly for being original and for dealing with important topics.
Before we discuss the content of our moral obligations according to Kant, it is well to begin with his second-order moral theory, since that (second-order ethics) is probably approximately what he means by the foundations of the metaphysics of morals.
The distinction between a priori and empirical knowledge looms large in the Kantian epistemology. His distinction between the world of appearance and things in themselves is equally legendary. In the Foundations, he unites the two distinctions into a distinction between the 'intelligible' world and the 'sensible' world, with the intelligible world being, apparently, the world both as understood through pure reason (a priori) and as it is in itself, and the sensible world being the world as we experience it and as mere appearance:
[A]ll conceptions, like those of the senses, which come to us without our choice enable us to know the objects only as they affect us, while what they are in themselves remains unknown to us . . . . This must furnish a distinction, though a crude one, between a world of sense and a world of understanding . . . . A man may not presume to know even himself as he really is by knowing himself through inner sensation. For since he does not, as it were, produce himself or derive his concept of himself a priori but only empirically, it is natural that he obtains knowledge of himself only through inner sense and consequently only through the appearance of his nature and the way in which his consciousness is affected.(1)
This passage clearly says that all empirical knowledge is knowledge of mere appearances. Kant gives as the reason why self-knowledge would be mere appearance that it is empirical. The passage at least implies, further, that if one had an a priori concept of oneself, then one could have knowledge of true reality in this case -- and that, more generally, a priori knowledge (or rather, more precisely, knowledge which is thought through a priori concepts), if there is any, would be knowledge of things in themselves. Moreover, merely calling the world as it is in itself "the world of understanding" or "the intelligible world" implies this. It's true that he also periodically proclaims the impossibility of knowledge of things in themselves, but this is not to be taken too strictly since he also grants us knowledge of the freedom of the will as an instance of such knowledge.
Now this distinction can be equally well applied to the realm of practical reasoning as to descriptive science. The propositions of practical reasoning (at least for the most part?(2)) are what Kant calls 'imperatives', where an imperative is, as I assume, a statement that tells people what to do. Reasons for acting will either derive from pure reason or from inclinations (i.e., emotions, instincts, or some such). The imperatives of reason, then, are the a priori practical principles, while the imperatives deriving from inclination are merely empirical, and hence of an inferior grade:
From what has been said it is clear that all moral concepts have their seat and origin entirely a priori in reason . . . In the purity of their origin lies their worthiness to serve us as supreme practical principles, and to the extent that something empirical is added to them, just this much is subtracted from their genuine influence and from the unqualified worth of actions. (p. 32)
According to Kant, an a priori concept or item of knowledge is one that the mind comes up with spontaneously, by itself, whereas an empirical idea or item of knowledge is one that is externally caused (see the above quotation where he refers to "conceptions, like those of the senses, which come to us without our choice"; see also p. 80: "For though the latter [the understanding] is also a spontaneous activity and does not, like sense, merely contain conceptions which arise only when one is affected by things, being passive . . ." -- here he is clearly talking about a priori concepts contrasted with empirical ones.) Imperatives that reflect our inclinations are empirical, in the first place, because the inclinations they are derived from, like sensory experiences, come to us without our choice and are externally caused; and in the second place, because they are contingent, because they could have been otherwise in case we'd had different inclinations. The external causes of our inclinations, and hence of the imperatives that reflect them, are the objects of desire that we seek to obtain.
From what has just been said it emerges that any 'consequentialist' practical principle, that is to say, any imperative based on what are the efficient means to obtaining some desired end, is empirical, for the desire is caused by something external, viz. its object. Thus we can understand the distinction between hypothetical and categorical imperatives: hypothetical imperatives tell one how to achieve some end, are therefore empirical, and hence are inferior to categorical imperatives, belonging as they do to the world of appearances. Categorical imperatives, belonging to the world of things in themselves, as the truly moral imperatives, must prescribe actions which are intrinsically worthy, i.e., ends in themselves, in order that they be a priori.
It is here that the doctrine of the good will comes in --
Nothing in the world -- indeed nothing even beyond the world -- can possibly be conceived which could be called good without qualification except a good will. (p. 11)
Further elaboration of what I take to be the same idea:
From the preceding discussion it is clear that the purposes we may have for our actions and their effects as ends and incentives of the will cannot give the actions any unconditional and moral worth. Wherein, then, can this worth lie, if it is not in the will in relation to its hoped-for effect? It can lie nowhere else than in the principle of the will irrespective of the ends which can be realized by such action. (p. 19)
The observant reader may notice the slip here between the claim that actions prescribed by categorical (i.e., according to Kant, moral) imperatives must be intrinsically good, and the claim that the willing of them must be intrinsically good. Kant does not seem to make a distinction between these; perhaps he regards the basic action as really boiling down to willing, which is then followed by some effects.
Lastly, the Kantian theory of free will: Recall that according to Kant a priori ideas are produced spontaneously by the mind, whereas empirical ideas are induced in the mind, which is passive, by external objects. His definition of free will is almost identical to Kant's description of a priori knowledge: the freedom of the will is its being determined according to laws of its own making, as opposed to being determined to act by external causes. Since according to Kant practical reason and the will are the same thing, this means that a free will is one that acts on a priori practical principles rather than empirical ones. "Therefore a free will and a will under moral laws are identical." (p. 74) For this reason, Kant thinks that the validity of morality depends on the freedom of the will. He thinks that because a free will is one that obeys moral laws, in order to show that there are moral laws, one has to show that the will is free. But instead of actually demonstrating this, he purports to show that we must presuppose that the will is free. Somehow this is supposed to be good enough --
Now I say that every being which cannot act otherwise than under the idea of freedom is thereby really free in a practical respect. That is to say, all laws which are inseparably bound up with freedom hold for it just as if its will were proved free in itself by theoretical philosophy. (p. 75)
I beg to be excused from explaining this laconic remark. Here is the entirety of the equally laconic argument for why we must presuppose freedom of the will:
Now I affirm that we must necessarily grant that every rational being who has a will also has the idea of freedom and that it acts only under this idea. For in such a being we think of a reason which is practical, i.e., a reason which has causality with respect to its objects. Now, we cannot conceive of a reason which consciously responds to a bidding from the outside with respect to its judgements, for then the subject would attribute the determination of its power of judgement not to reason but to an impulse. Reason must regard itself as the author of its principles, independently of foreign influences; consequently, as practical reason or as the will of a rational being, it must regard itself as free. (pp. 75-6)
I think the argument is that if you thought of yourself as acting in response to external stimuli, then you wouldn't call your actions acts of your will or of practical reason (will = practical reason) -- that is, it would not be your will that was acting, but something else (an impulse) -- and a rational being with a will is by definition one whose reason acts. I have to admit that I am puzzled by Kant's argument here. I thought that whether reason could be practical was the very point at issue, and Kant was concerned to show that it could. To give, as if it were an independent reason for the conclusion, "For in such a being we think of a reason which is practical, i.e., a reason which has causality with respect to its objects" surely begs the question. I confess I cannot figure out how the premise is supposed to differ from the conclusion. Again, why can't we think of ourselves as externally determined? Because "then the subject would attribute the determination of its power of judgement not to reason but to an impulse" -- yes, but so what? I thought whether our judgement (or action, as the case may be) was to be attributed merely to impulses was precisely the point at issue.
Well, passing over this oddity, we come to Kant's reconciliation of determinism with free will, where the intelligible and sensible world again make their appearance. It seems that not only are a priori ideas ideas of things in themselves, but they themselves are things in themselves; and not only are sensations ideas of the world of appearance, but they are parts of that world themselves. I say this because Kant says we are members of the 'intelligible world' insofar as we are intelligent, and members of the sensible world insofar as we have sensations:
That it [i.e., presumably, any person] must think of itself in this twofold manner rests, with regard to the first, on the consciousness of itself as an object affected through the senses, and, with regard to what is required by the second, on the consciousness of itself as intelligence, i.e., as independent from sensuous impressions in the use of reason and thus as belonging to the intelligible world. (p. 87)
It's not obvious why our use of intelligence puts us into the intelligible world, unless Kant is just asserting that instances of the use of reason are things-in-themselves, beyond the fact that they impart knowledge of things in themselves. It's not clear why our having sensations places us in the sensible world (sensations are not themselves sensible) unless Kant is telling us in addition to the fact that sensations impart (merely) apparent knowledge that they are themselves (merely) apparent things. Having knowledge of x is not the same as being x, so reason's giving us knowledge of the intelligible world isn't sufficient for saying that it makes us part of the intelligible world.
So as things-in-ourselves, we are autonomous, acting on moral imperatives, while as appearances, we act on inclinations and are determined by natural laws; and Kant assures us that there is no contradiction in this. Determinism is true in the world of appearances and free will in the world of things-in-themselves.
At this point, unable to avoid the feeling of inadequacy with this, I wish that I were making up this theory in order to set up a straw man, and that Kant really had a more convincing defense of freedom, but this is what he says. I shall evaluate it below, in section III.
Now that we understand the nature of morality, we can deduce its content. Kant tells us that there is, surprisingly, only one categorical imperative. He argues obscurely for the categorical imperative twice, in section 1 and again in section 2:
But what kind of a law can that be, the conception of which must determine the will without reference to the expected result? . . . Since I have robbed the will of all impulses which could come to it from obedience to any law, nothing remains to serve as a principle of the will except universal conformity of its action to law as such. That is, I should never act in such a way that I could not also will that my maxim should be a universal law. (p. 21)
But if I think of a categorical imperative, I know immediately what it contains. For since the imperative contains besides the law only the necessity that the maxim should accord with this law, while the law contains no condition to which it is restricted, there is nothing remaining in it except the universality of law as such to which the maxim of the action should conform; and in effect this conformity alone is represented as necessary by the imperative.
There is, therefore, only one categorical imperative. It is: Act only according to that maxim by which you can at the same time will that it should become a universal law. (p. 44)
Presumably by saying that this is the only categorical imperative, Kant does not mean to deny that, for example, "One should keep promises" is also a moral imperative. He must, therefore, mean that all other categorical imperatives derive from or are simply applications of this one. The argument for this categorical imperative appears to start from the premises that
(1) moral imperatives must be universal laws, and
(2) as discussed earlier, they must not be in service to some end beyond themselves, nor (this is supposed to be the same thing) can they be restricted to some contingent condition under which they are valid.
We start with the positive fact of universal lawhood on the part of any moral imperative, and when we exclude from these imperatives any ends to be achieved or conditions of applicability, nothing is left 'in' them except their universal lawhood. This conclusion would be warranted if
(3) universality (or particularity) and some end to be achieved were the only two possible aspects to an imperative, and if
(4) an imperative contained these things in a manner similar to how a jar can contain both purple and red jelly beans.
If we are assured that the jar can have only these two kinds of jelly beans but that it can't have any red jelly beans, then it must have only purple ones. Kant's argument seems to intend to be of that form. He implies premise (3) by calling universality the 'form' of an imperative and any desired end the 'material' of it. Form and matter or form and content are traditionally distinguished from each other as mutually exclusive and jointly exhaustive categories. Kant implies (4) by arguing that since the moral law doesn't 'contain' any condition (or end), there isn't anything left 'in' it except its universality. A final step is needed in the argument which consists in interpreting the idea of an imperative that only contains universality as meaning the imperative to only act in a manner you could will to be universal. I don't know why the imperative containing only universality, if there were such a thing, wouldn't tell you to act only on a maxim that is a universal law, instead of merely one that you could will to be a universal law. Once again, I wish that I were inventing a straw man here.
The examples of the application of the categorical imperative that Kant gives, which I go into here because they are, however brief, instances of first-order ethics, are equally convincing:
1. Committing suicide out of self-love is contrary to the categorical imperative because there is "a contradiction in a system of nature, whose law would be to destroy life by the feeling whose special office is to impel the improvement of life." Kant does not seem to be claiming -- what would at any rate be very implausible -- that it would be impossible for everyone to commit suicide, or impossible for anyone to will that everyone should do so. Instead, he says that if this happened, "it would not exist as nature; hence that maxim cannot obtain as a law of nature." (p. 45; emphasis added) The claim seems to be based on the ideas that the destruction of life is incompatible with its improvement and that nature always chooses organs adapted to their purpose (p. 13), so that nature couldn't (or wouldn't?) allow self-love to be used in a way contrary to its purpose.
2. Making false promises is contrary to the categorical imperative because the universal making of false promises would be impossible because if everyone broke their promises the institution of promising would collapse -- no one would believe promises or accept contracts that they knew would be broken. (p. 46)
3. Everyone must cultivate his talents because he could not will the neglect of his talents as a universal law of nature "for, as a rational being, he necessarily wills that all his faculties should be developed, inasmuch as they are given to him for all sorts of possible purposes." (p. 47) I beg to be excused, once more, from explaining this curt argument.
4. People must give to charity and help other people, because nobody could will universal indifference on the part of men towards other men, because everyone would want to receive help when in need. That is, because everyone would want others to show kindness and generosity towards him, he could not will universal selfishness. Therefore, he must not himself be selfish. (p. 47)
Kant's second formulation of the categorical imperative, putatively really the same principle, follows more naturally from our previous discussion of the nature of morality. Recall that moral, or categorical, imperatives do not aim at any end and that a will in accordance with them is an end in itself. And recall from the discussion of freedom that all rational beings (albeit presumably only to a limited extent) act according to moral laws. Because of this, "rational nature exists as an end in itself" (p. 53). Kant concludes
The practical imperative, therefore, is the following: Act so that you treat humanity, whether in your own person or in that of another, always as an end and never as a means only. (p. 54)
Why does Kant claim, contrary to appearances, that this is equivalent to the first categorical imperative that he identified? Both versions of the categorical imperative derive from the fact that only a will in accord with moral laws is intrinsically, and not merely instrumentally, good. If this is true, then we should treat beings with such wills -- in other words, all of humanity -- as if they are intrinsically and not merely instrumentally good. We should treat them as being what they are. Also, since a will in accord with moral law is intrinsically good, we should make our own wills good by means of conforming them to "the universality of law as such," which means the first form of the categorical imperative.
Kant thinks the first form of the categorical imperative, in which we have merely to make sure that our maxims are possible universal laws, follows from the denial that they should serve any ends (because of the purple jelly bean argument). He also thinks that the denial that (actions according with) categorical imperatives serve some ends means that such actions are ends in themselves, which means that the willing in accordance with the law is itself intrinsically good. And this is why we humans are intrinsically good.
Thus morality and humanity, so far as it is capable of morality, alone have dignity. (p. 60; emphasis added)
Autonomy is thus the basis of the dignity of both human nature and every rational nature. (p. 61)
These remarks indicate that, as per my interpretation, it is our acting in accord with moral laws that makes humanity an end in itself (Kant explains that by "dignity" he means "unconditional and incomparable worth" (p. 61)). From this is derived the second form of the categorical imperative.
There has been some controversy about what Kant meant by his doctrine that only action from the sense of duty had 'moral worth.'(3) I have assumed that he meant the same thing by calling a good will 'good without qualification' and by attributing 'moral worth' to conscientious action, and again by attributing 'dignity' to morality and humanity. I see a great internal coherence in Kant's book. Fundamentally he makes one distinction between two kinds of value. (Value, worth, and goodness are all the same thing.) Kant distinguishes intrinsic from instrumental value. Intrinsic value corresponds to imperatives that are categorical, a priori, universal, necessary, rational, and moral. Instrumental value corresponds to hypothetical imperatives -- also called "conditional imperatives" -- which are empirical, contingent, and based on inclinations. The 'condition' of the conditional imperative -- that is, the condition under which the imperative is valid -- is that one desires a certain end. Since people's desires differ, hypothetical imperatives are not universal. Since our desires could have been otherwise, hypothetical imperatives are contingent. (When the hypothetical imperative is conceived as of the form "If you want A, you should do B," it is the consequent of this conditional that is contingent, non-universal, etc.) When Kant said that the good will was good 'without qualification' he meant the same as 'unconditionally good', i.e., intrinsically good. And when he spoke of moral worth he meant, again, intrinsic and unconditional value, because it is the essence of his theory of morality that morality corresponds to categorical imperatives, intrinsic values, etc. He repeats the distinction over and over again throughout the Foundations, and it, with the identification of morality with the intrinsic, categorical side of the dichotomy, is the one thread unifying the work.
To the unconditionality and intrinsicness of moral value, Kant also adds its incommensurability with other values:
And, regarded for itself, it [the good will] is to be esteemed incomparably higher than anything which could be brought about by it in favor of any inclination or even of the sum total of all inclinations. (p. 12)
Whatever has a price can be replaced by something else as its equivalent; on the other hand, whatever is above all price, and therefore admits of no equivalent, has a dignity. (p. 60)
This esteem lets the worth of such a turn of mind [viz. moral virtue] be recognized as dignity and puts it infinitely beyond any price, with which it cannot in the least be brought into competition or comparison without, as it were, violating its holiness. (p. 61)
The incommensurability of moral value is the doctrine that moral virtue is infinitely better than anything else -- literally, incommensurability of some values means that there is no ratio (among real numbers) of the value of one thing to the value of the other. Kant, in saying that a thing with 'dignity' has no equivalent, appears to disallow not only comparison of moral with non-moral values but comparison between two things of moral value -- so, for instance, it's impossible to compare the value of one person (who has dignity) with that of another (another thing with dignity).
Some of Kant's mistakes are more demonstrable than others. Generally speaking, arguments that Kant gives are demonstrably invalid, while his claims, taken as flat assertions, even when false, are still not decisively refutable.
It is open to serious doubt whether this distinction is intelligible. What is a thing in itself? I've never seen one of them. I can see cats, cars, and clouds, for example; but Kant assures me those things are not things in themselves. They are mere appearances. So it appears that a Kantian 'mere appearance' is what I understand by a "thing", and what a Kantian 'thing in itself is' I have no idea. But I doubt the difference between Kant and myself is simply verbal. Like the regular skeptic, he seems, at least prima facie, by calling ordinary things "appearances" to be claiming that the nature of the world is radically different from what we generally take it to be. And how it really is we can never know. A very pessimistic view. If this is the meaning of the appearance/thing-in-itself distinction, the skeptical part of it is in desperate need of justification. If this is not what it means, then Kant has failed to explain what he did mean. However, this issue really harks back to the Critique of Pure Reason and is not particularly ethical, so we should move on.
Even granting the validity of the distinction, it is very doubtful whether Kant was right in equating all empirical knowledge with the 'mere appearance' side (at which rate, incidentally, it is obscure what entitles such beliefs to be considered 'knowledge'). Kant says that sense-perceptions enable us only to know objects 'as they affect us,' but it's not clear why we can't tell also how they really are, perhaps on the basis of how they affect us (isn't that the common-sense view of perception?) He was also wrong to equate the a priori with knowledge of reality, since beliefs not based on experience and beliefs involving a priori concepts are just as subject to error as empirically derived ones. I myself have frequently made errors in mathematics -- and the diversity of opinions in philosophy shows it must be even more subject to error. It seems that in contrasting the sensible world with the intelligible world, Kant did not really mean to say (absurdly) that beliefs based on perception are always false and beliefs based on armchair reasoning are always true. Rather, it emerges that, though both kinds of beliefs can be either true or false, some true beliefs are more true than others -- empirically based beliefs, even when true, do not correspond with reality as much as true a priori beliefs do. But this, of course, only raises again the question of the intelligibility of the distinction.
This issue is not irrelevant to the Kantian ethics. It is essential to Kant's argument. The equation 'empirical = appearance, a priori = reality' is the reason -- at least the only reason I can discern -- why Kant insists, as he does repeatedly, that empirical principles are unworthy to serve as parts of morality. And the appearance/thing-in-itself distinction is at the heart of the reason he gives in section three for why we should obey the dictates of reason rather than inclinations, granting that they can conflict:
Moreover, since it is only as intelligence that he [i.e., everyone] is his proper self (being as man only appearance of himself), he knows that those laws [i.e., moral laws] apply to him directly and categorically, so that that to which inclinations and impulses and hence the entire nature of the world of sense incite him cannot in the least impair the laws of his volition as intelligence. (p. 87)
Kant is here relying on the idea that moral laws are true in the realm of things in themselves, to explain why their importance must override inclinations. If, therefore, we are right to doubt the intelligibility of Kant's distinction, or to doubt whether empirical knowledge should only be counted 'appearance,' then the reason Kant gives for his moral theory collapses.
I still think it's true that we should always follow the dictates of reason; I just don't think the reason has anything to do with appearance vs. things in themselves (rather, I think it's just an analytic truth).
Kant motivates his definition of free will with the premises (1) that every event must have a cause, and (2) that causation must be according to laws. From these he concludes that free choices must conform to laws, "otherwise a free will would be an absurdity" (p. 74), but (since they are different from non-free actions) the laws must be "of a peculiar sort." Kant suggests that the laws to which a free will conforms are peculiar in that they are given by the will to itself. Inanimate objects are always, according to Kant, moved by external causes, whereas humans sometimes act according to internal causes. These special laws that the will obeys -- by reason of which it is 'autonomous' -- are none other than the laws of morality (which we have previously been assured are entirely a priori).
Now the confusion in this account is palpable. The kind of 'laws' which comprise morality are decidedly not causal laws. They are not the same kind of 'laws' as the laws of nature. A causal law is a description of how things actually behave (needless to say, a moral law is not). Nor can we hope to save the theory of the laws of freedom by replacing moral laws with some correlative causal laws, stating that people do in fact behave morally (which is the only natural intelligible interpretation of what Kant meant) -- for then such laws would be false (barring the unreasonably optimistic view that everyone always does his duty) and therefore we would have to conclude that we are not free after all, if freedom means acting in accord with such laws. Kant's account appears to entail that no one could ever be blamed for acting wrongly: for if a free will is the same as a will obeying moral laws, then to the extent that anyone transgressed a moral law, we would have to say, on that occasion, that his action was not free; that, if he gave in to some temptation, then his action was caused by inclination, i.e., externally caused; and therefore he could not be responsible.
Kant tells us, "The proposition that the will is a law to itself in all its actions, however, only expresses the principle that we should act according to no other maxim than that which can also have itself as a universal law for its object." (p. 74) He doesn't exactly explain how this is so, and it is especially puzzling since the first proposition is a descriptive one, while the principle he says it 'expresses' is normative. The proposition that the will is a law to itself, I should think, expresses the principle, not that we should, but that we do act only according to maxims, not that we could, but that we do will to be universal laws. That 'principle,' incidentally, is not true, even if the normative one with which Kant confuses it is.
Again, these objections dispel the whole main argument of the Foundations. In sections 1 and 2, Kant purports only to derive what morality would dictate if there were any moral laws. That there is such a thing as morality is supposed to be shown in section 3, on the basis of a proof of the freedom of the will. That argument becomes a non sequitur once the confusion in Kant's account of the connection between freedom and morality is uncovered.
Furthermore, the defense of the freedom of the will is unconvincing. Kant argues that free will is at least possible, consistent with the laws of nature, because freedom may reign in the realm of things in themselves while natural necessity rules in the realm of appearances, "for there is not the least contradiction between [sic] a thing in appearance ... being subject to certain laws of which it is independent as a thing or a being in itself." (p. 87) This makes it sound as if the sensible and intelligible worlds were really two different places, like California and New York. There is no contradiction in a person acting one way in California and an opposite way in New York. But to say there is no contradiction in a person acting one way 'in appearance' but a different way in reality, is either obviously false or else merely a sneaky way of saying that he could appear to act one way while he was in fact acting differently. It would be the same if one said that there is no contradiction in unicorns existing in my dreams but not in reality. But this way of 'resolving' the contradiction is merely to deny one of the two conflicting propositions and not at all to remove their conflict. That is, on the simplest interpretation of the appearance/reality distinction, what Kant says about freedom vs. physical determinism amounts to that the doctrine of free will is true and determinism, while it appears to be true, is false. This may be -- indeed, probably is -- a misinterpretation of the appearance/reality distinction in Kant, for Kant's attempt to reconcile freedom with determinism is ostensibly demanded by the fact that determinism, or so he thinks, is true. But then I think Kant's conception of the distinction is confused, since, as I say, he talks about these two different worlds as if they were related as California is related to New York. Any distinction which enables you to reconcile being subject to certain laws -- that is to say, subject to them everywhere and at all times -- with being independent of those same laws is probably a mistake.
Finally, the very distinction between internally and externally caused actions, on which the definition of freedom is based, is dubious. Of course, Kant is not the only one who uses this concept (it's implied in current debates about how much of a person's behavior is explained by his environment). Suppose that I take an ice cube out of the freezer. The ice cube's melting point is 32°F, and the room's temperature is 70°F. The ice melts. Now which factor was more responsible for the melting: the ice's chemical properties (internal cause), or the temperature of the room (external cause)? Such a question is senseless. Any action that any object -- whether human or inanimate -- partakes of is the outcome of the properties it has together with the circumstances in which it is placed. There is no answer to the question which factor is more responsible. And there is no such thing as an action that is caused solely by internal factors. Dutiful action on the part of human beings is certainly not an example of such a thing, for supposing that, for example, I give someone ten dollars in payment of a debt, because I believe in a duty to keep promises: then my action is surely caused, among other things, by my having the ten dollars and by my belief that I made a promise, which belief is empirical. Or if I give money to a poor man because I believe in a duty of charity, then my action is caused, among other things, by the presence of this man and his being poor. He would be the cause of my action no more nor less than any object of the inclinations is ever a cause of action. There is, therefore, no reason for declaring action from the sense of duty 'autonomous' and action from (emotional or instinctive) desire 'heteronomous'.
Nor are the actions of inanimate objects examples of purely externally caused actions. If physical determinism is true, then the brain processes that (in part) determine our behavior are no less internal than any beliefs or other psychological cause might be. So there is no reason for equating natural necessity with external causes and freedom with internal causes.
Kant associates hypothetical imperatives with inclinations and with actions aiming at ends beyond themselves. It is his conviction of the a priori and rational nature of morality that convinces him that morality can have no basis in means-end calculations. But this is a mistake. The distinction between intrinsic and instrumental values has nothing to do with the distinction between a priori and empirical principles, nor the distinction between reason and inclinations. Take utilitarianism as a common foil for the Kantian moral theory: utilitarians counsel us to act always to achieve the end of happiness. Nevertheless, their theory is just as a priori as Kant's. They do not appeal to experience to show that actions which maximize happiness are right. Although they do have to appeal to experience to determine which actions will produce happiness, a similar thing could be said of followers of Kant: the fact that a given action will fulfill a promise, or aid the needy, or develop my talents (all duties that Kant mentions) is no less empirical a fact than the fact that it will produce pleasure. This is especially obvious for the case of the duty of helping others -- for helping them, presumably, means helping them to be happier and to satisfy their inclinations, and what will help other people in this way is surely no less empirical an issue than is what will help myself.
Actions that are performed for their own sake may also be based on inclinations just as well as on reason. Suppose, for example, that somebody likes to dance just for the sake of dancing. The fact that the dancing would not be aimed at some further end beyond itself does not mitigate the fact that it is motivated by inclination rather than reason.
So we see that, contra Kant, instrumental values can be rational, and intrinsic values can be based on inclination. Kant therefore has lost his basis for insisting on the purely deontological (as opposed to consequentialist(4)) form of morality.
By "categorical imperative #1" I mean Kant's claim that we should act only according to a maxim for which we could will that it should be a universal law. Its problems are fourfold.
First, its justification is based on what I have called the purple jelly bean argument: if a jar contains only purple and red jelly beans, then if you remove all the red jelly beans it will have nothing but purple jelly beans left in it. It seems wrong to think that an imperative can contain universality and an end in the same way a jar can contain two flavors of jelly beans. If Kant is right to call these things the "form" and "matter" of the imperative, then their relation is quite different: notably, they cannot exist independently. There is no such thing as form without matter or matter without form. If you 'remove' the matter of a proposition, then you remove the proposition. There is nothing left. The notion of a proposition that has nothing to it except a universal form, without any particular content, is meaningless.
This brings me to the second problem, which is with interpreting what it is that the jelly bean argument has supposedly established. The reason obeying the categorical imperative is supposed to be a good thing is that to do so constitutes conforming one's actions to "law as such" or "the universality of law as such." That there is no such thing as conformity to 'law as such' should be evident. One cannot just conform one's action to law 'as such' without conforming to any particular law, any more than one can have a car that merely has 'color as such' but without any particular color.
Third, and probably as a result of that problem, categorical imperative #1 isn't actually an imperative of the ordinary sort; it doesn't really tell us what actions to perform. Instead, it tells us what sort of maxims to act on, as if our deliberations typically were not about what actions to perform but about under what maxims to perform them. The categorical imperative represents a test on maxims or rules of conduct -- a necessary condition of their acceptability. This would be perfectly fine, if only the condition were strict enough to determine a unique set of acceptable rules, or, barring that, at least a greatly limited number. Given that Kant claims his categorical imperative is the sole principle of morality (the only categorical imperative, that is), we are entitled to expect that it determine the principles of morality uniquely, since if it leaves multiple incompatible sets of maxims open, we will have no basis for choosing among them, there being no other principles of morality on which to base the choice besides categorical imperative #1. Our expectations are disappointed. Take, once again, the utilitarian moral theory for an example. Surely it would be possible to will the universal maximization of happiness? The principle of utility, then, passes Kant's test; and, contrary to popular opinion, it turns out Kant's moral theory is consistent with utilitarianism. To my knowledge, no moral system has ever been articulated that would not pass the test of categorical imperative #1. This might at first be thought to be good, as a sign that the test is valid (it doesn't rule out anything that shouldn't be ruled out), but it also unfortunately means the condition is too weak to be of any use to us. How about a religious ethics, for another example? It surely seems possible to will universal conformity to the Bible (assuming the Bible is itself internally consistent). To be even more antagonistic to Kant, why wouldn't it be possible to will that everybody should just act on the basis of arbitrary whims? Or, why couldn't one will universal egoistic hedonism?
Kant claims that the last is impossible in his discussion of the duty of charity, but this brings us to his fourth difficulty: the applications Kant derives from CI #1 do not follow from it. He lists four duties:
a) The duty to keep promises. He claims universal promise-breaking would be impossible because it would mean no one would ever believe a promise made to him, and this would cause the institution of promising to collapse. But it is not logically impossible that promising could continue to exist. It is only an empirical hypothesis that if promises were generally broken then people would stop believing in them, and a further empirical hypothesis that if no one believed promises then people would stop making them. If we are allowed to consider empirical facts in morality (contra Kant's repeated insistences upon its a priori nature), these facts still do not show that people would be unable to carry on universally making and breaking promises -- only that they would probably not wish to.
Furthermore, it isn't clear that universal promise-breaking is what I would have to be able to will, if I plan to break a promise. For the maxim I act on may not be 'to break promises' but rather only 'to break promises when it would be especially difficult to keep them,' or even, 'to break promises when no one is likely to find out.' I may keep most of my promises but only break them when it is especially much in my interests to do so. Since this would be fairly rare, the institution of promising could survive the universalization of my maxim (in fact, such a maxim probably is nearly universally in practice already).
b) The duty of not killing oneself. Here Kant premises that the 'purpose' of self-love is to impel the improvement of life. It may be objected that the 'function' of self-love, if there is any, is rather to promote one's happiness and prevent one's suffering. With that purpose suicide could be consistent. Kant premises, second, that the maxim of someone who commits suicide is to kill himself out of self-love. It's interesting that here the maxim of one's action is taken to include a description of one's motive. But that might not be the motive. If the 'office' of self-love is indeed to impel the improvement of life, then perhaps the victim of suicide would better describe his motive not as "self-love" but as "dislike of suffering." His maxim is then to shorten his life out of an aversion to suffering when he anticipates suffering in the future -- in which there is surely no inconsistency. Kant also implies that death is incompatible with the improvement of life. But if a thing is sufficiently bad, then destroying it may very well be an improvement.
Kant's argument appears to require, finally, the premise that it is impossible for a thing to be used in a way contrary to its natural function. But even if the rest of the argument were correct, this surely could not be. For since people do in fact commit suicide, there obviously is no contradiction in it.
c) The duty to develop one's talents. About this a similar complaint may be made: Kant claims a person could not will the universal neglect of talents, "for, as a rational being, he necessarily wills that all his faculties should be developed, inasmuch as they are given to him for all sorts of possible purposes." If the South Sea Islanders really are, as Kant says, neglecting their talents and idling away, then, since presumably they are doing so willfully, clearly it cannot be true that every rational being necessarily wills that his faculties be developed. d) The duty of helping others in need. About this, Kant argues that one could not will universal indifference on the part of people to each other's need because every person wants people to help him when he is in need. Kant concludes that therefore every person must be willing so to help others, in order to be consistent. But instead of showing that refusal to help others is immoral, why doesn't this just show that wanting help from other people is immoral? In other words: there is a conflict here: when it is myself who is in need, I want people to give to charity, but when others are in need I do not want to give to charity. If this shows that my maxim of not giving to charity is not universalizable, why doesn't it show that my maxim of accepting charity is not universalizable? The alleged reason I cannot will the law of universal selfishness is because I would not want that others should follow it towards me; but then, we can also argue that I cannot will the law of universal altruism because I would not want that I should follow it towards others.
Actually, though, it is not impossible for me to choose to give to others in need, so it is not impossible to will universal altruism. But then, it obviously is not impossible to choose not to receive help from others either, so it is not impossible to will universal egoism. Even if it's true (as I quite doubt) that everybody in fact wants to receive help when he is in need, it certainly would not be impossible that a person should choose not to be helped.
Moreover, the question is not whether one could will each instance of the universal law in question; the relevant question is whether one could will the law as a whole, i.e., whether, if offered the whole package, in which other people do not help you but you don't help them either, anybody could accept it. So far from it being the case that no one could choose this, I say that a great many people, perhaps the majority, would choose it.
These examples show how very little about morality categorical imperative #1 tells us. Basically, the only sorts of things that people could choose to do that CI #1 rules out are actions whose maxims include (by definition) the fact that other people are not doing the same thing. For example, if I were to write some books by way of trying to become the best philosopher in the world, or if I tried to be the best at anything, my maxim would not be universalizable. I could not choose that everybody be the best. Incidentally, this is of course a counter-example to the Kantian principle. There is nothing wrong with trying to be the best. There is nothing wrong per se with doing things that, if you do them, other people cannot.
Categorical imperative #2 is the proposition that one should always treat humanity, in oneself and others, as an end and not merely as a means. Its first problem is ambiguity. I know how to treat a state of affairs as an end (namely, try to bring it about), but I don't know how to treat an entity as an end. Does treating a person as an end mean treating his existence as an end? Should we, then, simply try to maximize the number of rational beings existing? Kant does not say anything that sounds like this, but in his second discussion of the meritorious duties towards others, he implies that treating people as ends means treating their happiness as an end. This is why we have a duty of helping people to achieve what they want. But how this flows from the rest of Kant's theory is obscure, since he previously insisted upon the utter worthlessness of all objects of inclination. If moral virtue is infinitely better than happiness and pleasure, then, just as we must always sacrifice our own pleasure to moral virtue, must we not also sacrifice others' happiness for the sake of moral virtue? Wouldn't the proper thing be to treat people's moral virtue as an end, rather than either their happiness or their existence? Thus we should have to go about preaching constantly to each other. This need not involve detriment to their happiness, but with the Kantian doctrine of the incommensurability of values, it is unlikely anybody's happiness would ever influence our actions. Although it's a theoretical possibility that two actions could result in the same amount of moral virtue in the world but one produce more happiness, so that the latter should be chosen, it is infinitely unlikely this would ever happen. That two quantities should be exactly equal, or that an action influencing our happiness should have no effect whatsoever on our moral virtue, is an infinitesimal possibility. We therefore can for all practical purposes disregard any concerns of benefitting others or ourselves, and focus all our energies on trying to get people to obey the categorical imperative.
The second problem with CI #2 is a circularity problem: Why should we treat people as ends and never merely as means? Because they in fact possess intrinsic, incommensurable value. Why do they possess such value? Because they are capable of morality, and dutiful action is the sole intrinsic, and incomparable, good in the world. But for this to be so surely presupposes that there are duties (presumably Kant didn't think that action according to false moral principles was good). Moral action is good only if we do in fact have obligations. And when we inquire into these duties we come full circle at last, learning that the only duty there is is the duty to treat people as ends (which we were trying to figure out why it existed to begin with). Ultimately, this alleged duty has no basis. The absolute worth Kant ascribes to human beings would only be warranted, from his argument, if he could show independently that there were obligations. If there were obligations to do some things other than just try to see to it that people follow obligations, then there could also be an obligation to do that. But the morality I suggested in the previous paragraph (of maximizing virtue) is an impossibility: there cannot be a moral imperative to do nothing but make sure people follow that very imperative. This is what would result if the sole principle of morality were (as Kant tries to make it) the importance of people's adhering to morality. This would be like a government whose only law said to treat law-abiding citizens in a certain way.
Finally, the illustrations that Kant gives of CI #2 are, while more convincing than those for CI #1, mostly unsuccessful:
a) He claims that suicide is a case of treating oneself as a means instead of an end. Well, it is a case of treating one's existence (or non-existence) as a means for one's happiness. The ambiguity in the categorical imperative resurfaces. If we are supposed to treat the mere existence of human beings as an end, then no one should commit suicide. But if the happiness (and absence of pain) of humanity is the end, then perhaps some of us should. It is interesting that when it comes to oneself, Kant thinks one should treat the mere existence of humanity as an end in itself, whereas when it comes to others, he thinks we should treat their happiness as an end in itself.
b) Kant tells us that breaking a promise constitutes treating another person as a means, namely the person to whom the promise was made. Since the breaking of a promise rarely kills anyone, we can assume the complaint is that it fails to treat the other person's happiness as an end in itself. This is not always the case either, for the breaking of a promise need not make the promisee unhappy; in theory, it might even increase his happiness. Nor is it obvious why the person who wishes to break a promise would be treating promisee as a means any more than promisee would be treating promisor as a means by holding him to the promise. If defaulting on a debt is immoral because it uses another person to promote your own happiness, then why isn't collecting a debt also immoral, because it also promotes your own happiness at the expense of another person? Why, indeed, isn't any sort of business dealing which one does for profit immoral?
c) When it comes to the duty to develop one's talents, Kant introduces a third meaning of 'treating humanity as an end' -- now it means treating the perfection of humanity as an end in itself. Since this may interfere with taking our happiness as an end in itself, we should be equally entitled to say that the person who develops his talents is failing to regard himself (his happiness) as an end, as to say the person who neglects his talents fails to regard himself as an end.
d) Kant reverts to the idea of treating humanity's happiness as an end when he discusses the duty to help others. But once again we can turn Kant's principle against him. He uses it to show that egoism, and refusing to help others, is wrong, but can't it just as well be used to argue that altruism is wrong? For isn't the altruist just using himself as a means for the sake of others' happiness? Indeed, if taking money from other people to use on yourself is treating them merely as means and not as ends (and presumably taking their money for any purpose, even to help some third party, would also be treating them as means) then surely taking your own money and using it on others, by analogy, is treating yourself as a means.
With regard to any possible action where people have conflicting interests -- the action serves one person's interests while harming another's -- it could be argued that the person who performs or refrains from the action is treating one of the parties as a means, no matter what he chooses to do. If I have ten dollars which I can give to either of two people (perhaps one of whom is myself), then whomever I give it to, I will, perhaps, be accused of treating the other person merely as a means -- or at the least, of failing to treat him as an end. If this is so, it is impossible to adhere to the categorical imperative. Moreover, since the reason given for CI #2 is the absolute and incomparable value of human beings, there is no way to make choices between people. Since one cannot compare the value of one person to that of another person, or to anything else, in a situation in which one is given a choice between different people (suppose there are a limited number of life rafts on the Titanic), there is nothing one can do. If values are incommensurable, we cannot say a hundred deaths are worse than one.
Frankly, I find this doctrine irrational. It ignores the fact that there are situations in which we are forced to compare values. We have to decide, for instance, how much money to spend on health care -- if we choose to take health as an incommensurable value, then there is little doubt it can consume the entirety of the gross national product of the country, and leave no time or resources for anything else. This is what any 'absolute value' in this sense will do: it will destroy everything else. For this reason it is logically impossible to have multiple absolute and overriding values -- they must come into conflict, in which case they can not both be treated as absolute. Moreover, the concept of incomparable values is just mathematically absurd -- the notion of a quantity that is neither greater nor less than, nor yet equal to, another quantity (both being quantities of the same thing, as value, or mass, etc.) is mathematically absurd. Kant appears to want this incoherent idea to justify his principle of never treating people as means. If he admitted comparison of values, then one might sometimes be required to sacrifice one person for the sake of others. If he merely said all people were equally valuable, then two people would be twice as valuable as one, so you could kill one person in order to save two. That, of course, is only on consequentialist assumptions. Kant's theory of the incomparability of values (his theory of 'dignity') is his way of undermining consequentialism. I think it is unsuccessful, although I do not think it is the only possible way of undermining consequentialism.
The Foundations can be looked upon as a sustained argument leading to the categorical imperative and its alleged applications, but it is an argument that I believe I have undermined at every step. The argument has the following structure: First, Kant attempts to show that the form of morality must be deontological (that is, not consequentialist), universal, and unconditional. He argues that this is the nature of the concept of morality. From this he proceeds to argue, next, that there is only one possible moral principle satisfying those conditions, which can be stated in two different forms (actually he says three, but the third one, which I have not discussed herein, is hardly distinguishable from the second), and that principle is 'the categorical imperative'. From this principle he deduces four more specific moral obligations as illustrations. Finally, he attempts to demonstrate the validity of the concept of morality (i.e., that there is such a thing as morality), as the necessary final step in the argument -- that is, he has argued that the categorical imperative, as he articulates it, is the only principle that can satisfy the concept of a moral imperative, and he now (section 3) argues that the concept of a moral imperative is valid; therefore, the categorical imperative is true. For this step, he argues that a free will and a will under moral laws are the same; and that all rational beings have free will.
I have argued, on the contrary,
(1) That the form of morality need not be deontological, for, contra Kant, consequentialist principles can be equally as rational and a priori as deontological ones; nor are they any more 'conditional' or contingent than deontological principles;
(2) That in any case the first form of the categorical imperative as the sole moral principle does not follow merely from the assumption that morality must be universal and deontological;
(3) That the second formulation of the categorical imperative is ambiguous, and the only interpretation of it on which it actually follows from Kant's argument for it also makes it something that could not, on pain of circularity, be taken as the sole moral principle;
(4) That none of the examples of moral duties that Kant gives are, as he claims, actually applications of the categorical imperative;
(5) That the validity of morality cannot follow merely from the freedom of the will, which is a purely descriptive proposition; and finally,
(6) That Kant's positive argument for free will merely begs the question, while his defense against determinism, trying to show it is compatible with free will, is based on a confused conception of the relation between appearance and reality.
1. Pp. 78-9. All references are from the Foundations of the Metaphysics of Morals: text and critical essays (Indianapolis: Bobbs-Merrill, 1969), Lewis White Beck, tr.; Robert Paul Wolff, ed.
2. "Billy is a good man" isn't an imperative.
3. See, e.g., Richard Henson, "What Kant Might Have Said: Moral Worth and the Overdetermination of Dutiful Action" (Phil. Review, January, 1979, pp. 39-54).
4. By "consequentialism" I mean the view that the sole obligation is to maximize the good, or expected goodness. That is, whatever it is that is good, we must produce as much of it as possible.