Kant is usually recognized for his unique emphasis on moral duty, but like every other prominent moral philosopher, he works to leave room for happiness. In both the first and second Critiques, he claims that "the highest good" is a necessary end to pursue. "The highest good" is defined as "happiness distributed in exact proportion to morality." In the first Critique, Kant claims that "all hope concerns happiness" and then rejects the "pragmatic" principle that "advises us what to do if we want to partake of happiness" because it "is grounded on empirical principles; for except by means of experience I can know neither which inclinations there are that would be satisfied nor what the natural causes are that could satisfy them." (CPR A806/B834) On Kant's view, such a principle would be so random and unsteady as to leave us with no hope for happiness. Kant instead advocates his own moral theory which instructs us to do "that through which you will become worthy to be happy." (CPR A808/B836) We gain this worthiness through virtuous moral action, which God will then reward with a proportionate amount of happiness. According to Kant, this is our only sure hope for happiness.

But Kant's theory leaves even less room for happiness than he himself supposes. Happiness is to come as a result of virtuous action; but as Kant describes virtuous action, it is impossible for mere humans to discern and act on an entire category of virtue. Because humans lack future-directed omniscience, they can neither perform actions to fulfill the imperfect duties that Kant has already described for them, nor can they determine what further imperfect duties they have to fulfill. Consequently, their entire hope of happiness must rest on the few and formal perfect duties which they might still be able to discern and act upon.

Kant himself recognizes some of the obstacles to the sort of hope for happiness he suggests; one such hurdle is "The Antinomy of Practical Reason". The antinomy is this: as a necessary end, we must promote the highest good, but we cannot act virtuously because we wish to receive happiness (for such action is not virtuous, by Kantian lights) nor can we make our genuinely virtuous actions result in happiness (for we lack the power to do so). In the second Critique, he notes the consequences for his conception of the moral law should this antinomy stand unresolved.

Naturally, Kant does not believe that the antinomy actually does prove his theory to be false; it is resolved with the postulated existence of God. The above claim is only that it would prove his theory false if it could not be resolved. His point here is a strong one, and it can be made more generally applicable. Broadly construed, it runs something like this: if a moral theory requires something that is "impossible in accordance with practical rules" of the agents to which it is meant to apply, it is seriously flawed and cannot be the right theory for these agents.

This idea is surely not foreign to Kantian moral theory. It seems to be just another instance where "ought implies can," and an agent cannot be obligated to do to what is impossible for that agent. Not only can the agent not be obligated to perform such actions, but the agent cannot even be expected to successfully perform such actions. It is in this spirit that Kant makes a specific objection in the Groundwork against eudaemonist moral theories.

This objection seems to be an elaboration on the objection Kant offered before against "pragmatic" principles. Not only do the eudaemonists present principles that are morally flawed, but the principles they present are practically flawed as well. This is because, according to Kant, only an omniscient agent could use a eudaemonist principle as a guide for action. Human agents are finite and thus require a different theory; that is, one that they can use as a guide for both their practical and moral action.

Kant presents what he takes to be just such a moral theory, one that is applicable not only to humans, but to all rational beings. He intends his moral theory to be not only applicable to rational beings, but successful in guiding their actions to be moral and in securing happiness, their necessary and most valued practical end.

As his theory is meant to apply to humans, it should not require omniscience in order to guide action, and Kant takes his theory to be free of such a requirement. Regardless of what Kant might think about his own theory, I will endeavor to show in this paper that, at least regarding imperfect duties, Kant's theory does require that agents be omniscient to act on and derive these duties, and they should thus not obligate finite rational agents. Because this class of virtuous actions is not available to humans, it also changes the status of the happiness that might have resulted from these actions, from an end that was apparently obtainable, to one that is beyond the reach of finite human beings.

To do this, I will first explain Kant's objection to the eudaemonists, why it is a threat to their theories (especially the Epicureans) and why Kant's requirements for practical success are so stringent. Next, I will illustrate how a parallel objection can also be leveled against Kant's own theory. This objection will show acting on and deriving imperfect duties to be impossible for finite rational beings. Finally, I will note how this impossibility thwarts much of the hope of happiness.

It is certainly nothing new to observe that Kant rejects any moral theory that attempts to base moral action in the pursuit of happiness. Any action for the sake of happiness will be heteronomous and lacking moral worth, on Kant's view. But the objection Kant offers in the Groundwork goes even further than merely claiming that acting to maximize happiness should not be a human agent's principle for action, but that it cannot.

The first obstacle to achieving happiness is its vague definition: "unfortunately, the concept of happiness is such an indeterminate one that even though everyone wishes to attain happiness, yet he can never say definitely and consistently what it is that he really wishes and wills." Even the wisest and most powerful human cannot determine what actually comprises his happiness. "Now it is impossible for the most insightful and at the same time most powerful, but nonetheless finite, being to frame here a determinate concept of what he really wills." (G 418)

The difficulty this vagueness presents is compounded by a second more serious obstacle that would persist even if a determinate concept of happiness could be established. Without knowledge of what the outcome of his actions will actually be, an agent cannot determine which actions will actually promote his happiness. To illustrate this, Kant gives us four brief examples of actions that initially appear to promote happiness, but actually result in greater unhappiness for the agent. Should he pursue greater wealth, he might just be pursuing greater anxiety and receive the envy of others. Pursuing greater knowledge may only result in dreading the unforeseen but unavoidable future evils. In seeking greater longevity, he may actually be seeking to suffer a painful ailment longer. Even the simple goal of greater health could allow the agent to fall into harmful excesses that he might otherwise have avoided. Because even the best-intentioned and most carefully planned actions can fail to bring their intended result, Kant believes they cannot be the basis for even practical action, let alone moral action. "In brief, [an agent] is not able on any principle to determine with complete certainty what will make him truly happy, because to do so would require omniscience." (G 418)

The four examples that Kant provides here are a bit too brief and vague to fully illustrate his point, but we can borrow another famous Kantian case to make his point more vivid. The case comes from Kant's "On a Supposed Right to Lie Because of Philanthropic Concerns." Here, he imagines a murderer who comes to your door to inquire if his intended victim is in your house. Wishing to save the intended victim's life, you lie and tell the murderer he is not. Consequently, the murderer leaves your door, finds the intended victim who had slipped out of your house, and subsequently murders him. Beyond Kant's own didactic purpose in telling the story, it illustrates how even given a clearly defined end and a seemingly obvious means to it, acting on those means is no certain method of promoting that given end. Presumably, you acted for the sake of your happiness and that of the intended victim, but because you did not (and as a finite being could not) know the outcome of the action, you actually helped bring about precisely what you intended to avoid. Only an omniscient agent could expect to avoid such difficulties.

Such objections would not hold against a modern eudaemonist who only wishes to maximize expected utility. She might feel the need to defend herself against the claim that she has no determinate concept of her own happiness; but that events go awry is no problem for her theory, as she aims only to maximize expected utility, and the complicating circumstances are all unexpected. Obviously, Kant was not addressing modern Utilitarians, and these objections are more serious when considered in light of the Hellenistic eudaemonist theories that Kant often used as foils to his own.

Stoics and Epicureans alike offered definite formulae to achieve happiness. The Stoics urged us to find happiness in performing our virtuous actions, while the Epicureans believed we would find happiness by removing all pain. Thus, to claim that no one can ever "say definitely and consistently what it is that he really wishes and wills" in order to be happy is directly contrary to the central claims of both of their moral theories.

The second objection from unexpected circumstances seems only to hold against the Epicureans, but not the Stoics. For Stoics, the emphasis is on finding happiness in performing virtuous actions, not in the success of those virtuous actions. R. W. Sharples characterizes the Stoic sage as "an archer whose goal is not to hit the target, but to do the best he can to hit the target" and then quotes Cicero who says of an agent "That he should do everything to achieve his purpose would be…his ultimate good…but that he should hit the mark is as it were to be selected, but not to be desired." Thus, if an unexpected circumstance prevents the success of a virtuous action, this is no worry for a Stoic, as long as the action was virtuous.

Kant's objection speaks more against the Epicureans who both claim a determinate idea of happiness, and require successful actions to achieve this idea. Epicurus himself, in his Key Doctrines, says "He who knows the limits of life knows how easy it is to obtain that which removes pain caused by want and that which makes the whole of life complete." Even considering the simplicity that Epicurus advocates, the complications Kant has highlighted work to undermine the claim that the complete life is easy to obtain. Perhaps if an agent is omniscient, in the relevant sense, the complete life would be easily obtainable, but humans are not such agents and cannot expect success using such a theory as a guide to their action.

It could be objected at this point that Kant is simply being too stringent in his requirement that practical action be so successful. This is because Kant simply cannot accept that circumstances might go awry because he must leave room for the hope of happiness. The existing Hellenistic theories are merely "pragmatic" and, lacking knowledge of our actions' outcomes, we can never be sure that our actions will actually produce or increase our happiness. If we use eudaemonist principles, our happiness is left to erratic events and random chance, and such circumstances leave us in a position that is hopeless. Contrary to this, Kant offers us a theory that provides us with happiness insofar as we are virtuous, and given these circumstances, we can at least hope to achieve happiness. We need not know what comprises our happiness, or which particular actions might promote it; we merely need to be virtuous, and happiness will result.

Kant's rhetorical strategy when he levels this objection in the Groundwork is clear. Before presenting his own theory, he wishes to clear away the competing moral theories by showing them to be impossible or inadequate, so that his theory can be presented without distraction. But instead of another moral theory remaining to challenge Kant's theory, it is this very objection that persists to become an objection to the theory he presents. We have already seen what this objection is and how it works against the eudaemonist theories. I will now show how the objection applies to Kant's own theory, requiring omniscience for an agent to act on or derive imperfect duties.

Only three Academy pages after this objection to the eudaemonists, Kant presents the Universalizability Formulation of the Categorical Imperative, from which both perfect and imperfect duties will be derived: "Act only according to that maxim whereby you can at the same time will that it should become a universal law." (G 421)

Perfect duties are the first kind of duties derived from this universalizing procedure. Perfect duties are those which must be observed at all times, and require us to avoid conceptual contradictions. These arise from maxims that, when universalized, result in conceptual incoherence. One of Kant's own examples concerns false promising. (G 422) He imagines a man who universalizes a maxim to the effect that "I will promise to repay a debt I do not intend to repay, to get money when I need it." Kant claims that such a maxim cannot be willed as universal law, because if everyone were to make false promises to get loans, no loans would be given, because no creditors would expect to be repaid.

Imperfect duties, on the other hand, require us to avoid contradictions in our wills. A contradiction in the will would occur when an agent wills both a maxim and its negation. One of Kant's examples here is a maxim of non-benevolence. If an agent wills the maxim "I will not help others in need" this will not result in an incoherent world when universalized: a world in which no one is helpful can be easily imagined. The contradiction is instead in the agent's will, because the agent will inevitably need the assistance of others and then will certainly will that others help her, contradicting the proposal that everyone will not help those in need.

The duties against false promising and of benevolence are only half of the duties Kant here provides as examples. He also cites a duty against suicide and a duty to cultivate one's natural abilities, but these four duties should not be taken as an exhaustive list of the duties that the Categorical Imperative proposes. Immediately after presenting these four, Kant states "These are some of the many actual duties, or at least what are taken to be such, whose derivation from the single principle cited above [universalizability] is clear." (G 424)

Perfect duties, because they arise in response to conceptual contradictions, are immune to Kant's objection to the eudaemonists. They do not depend on what an agent will be required to will in the future, so they do not to require that an agent be omniscient in order for the agent to derive and act on them.

Conversely, imperfect duties do depend on what a particular rational agent must will, and are vulnerable to Kant's objection in two important ways. First, to act in accordance with the given imperfect duties of benevolence and the cultivation of talents, an agent must have the same kind of foresight that is required to act on a eudaemonist principle, and that finite beings lack. Second, in order to derive any of the further imperfect duties that Kant promises, a human agent must have even more detailed knowledge of the future than is required to act on a eudaemonist principle, and that is beyond him, given his finite nature.

When a finite agent discovers his imperfect duty of benevolence, he is faced with a practical problem almost identical to the one Kant believes he would face if he were attempting to act on a eudaemonist principle. Kant gives us two very similar but distinct notions of what benevolence should be considered to be, the first of which should obviously be rejected in light of his objection to the eudaemonists, and the second of which will be shown to be impossible for finite agents.

One account of benevolence is to act so as to promote the happiness of others. It should be obvious how using this as a guide to action requires omniscience just as much as a eudaemonist principle. Merely because an agent now attempts to act for others' happiness instead of his own does not grant him any greater certainty about the nature of happiness, or what the outcome of his actions will be. All the practical problems Kant believes arise in pursuing one's own happiness should appear in this case as well. Thus, promoting another's wealth may just promote his anxiety, promoting another's longevity, his continued ill-health, and so on. If this conception of benevolence is correct, then it requires that an agent be omniscient, just as much as the eudaemonist theory did.

Thus we must interpret benevolence the second way, as mutual aid. On this reading, benevolence is performing the permissible actions others ask of you, that they believe will promote their happiness. This reading is supported in the Doctrine of Virtue, when Kant says of benevolent action "What [others] count as belonging to their happiness is left up to them to decide; but I may decline many of these things which I do not regard as so belonging, if they otherwise have no right to demand them of me."

This interpretation of benevolence as mutual aid might initially appear immune to the Kant's objection to the eudaemonists, as it seems simple enough to merely do what one is told. On closer inspection, it still requires that an agent have a kind of knowledge beyond that of a finite rational being.

The problem is less obvious in simple and trivial cases. If someone merely asks to be handed a wrench that is beyond their reach to complete a mechanical task, there seems to be no difficulty in performing this action. But such cases are too trivial to really qualify as part of anyone's happiness --rather, they seem to be more a matter of mere convenience.

Beyond trivial cases of mere convenience, people tend to seek aid to their own happiness in far more personal and difficult circumstances. A plausible example might be one in which a friend asks for help with their sister. The difficulty here is much more obvious. Imagine that the sister is seriously depressed and that you are asked to help bring her out of her depression. Given such a task, you may not even know how to begin. Even if you should decide on a course of action, you must know that your actions will actually assuage the sister's depression, rather than exacerbate it; otherwise, you are not helping. Imagine that you buy her a kitten, but her only reaction is an allergic one. You make her a steak dinner, without realizing that she's a vegetarian. You try to distract her by discussing history, only to learn that her philandering husband was a Historian. To actually perform the action of helping bring the sister out of her depression, your actions must be helpful. And the only way to be sure that your actions will be helpful is if you can know their outcome. Beyond the simple and trivial cases, it seems that people will tend to seek aid in similarly vague and difficult circumstances, and one cannot help without adequate knowledge of how his actions will turn out.

However benevolence is cast, Kant's own given imperfect duty of benevolence will require a kind of omniscience (or at least accurate foresight) that finite beings lack. If this requirement of omniscience qualifies as a reason to reject eudaemonism as an ethical theory for humans, it is also a reason to reject the imperfect duty of benevolence as a possible guide for their action.

A similar difficulty arises in conjunction with the second imperfect duty that Kant enumerates in the Groundwork, the duty of cultivating one's talents. I must cultivate my talents to avoid a contradiction in my will. My talents are given to me "for all sorts of possible purposes", presumably my own purposes and the purposes of others. (G 423) If I am to universally will that no one cultivate their talents, I contradict my will when I will any of these "possible purposes" as my end, and they require a particular talent (of mine or another's) to be pursued. Consequently, I have an imperfect duty to cultivate my talents.

But when an agent is faced with the practical task of cultivating her talents, it is not at all clear how this is to be achieved. First, she must be sure that her suspicion that she actually has a particular talent is accurate. Kant describes the situation as a man who "finds in himself a talent whose cultivation could make him a man useful in many respects." (G 422-423) But this is not a state that is easy to verify. Suppose an agent suspects that she has a talent for music, and spends years of her spare time attending lessons, practicing her instrument, and attending concerts. Despite all of her efforts, she simply has no rhythm and never manages to produce even a competent performance of any piece of music. At the outset of her endeavor, she couldn't have known that she had no rhythm and wouldn't develop any, and she couldn't have known that her musical efforts would not be a case of cultivating her talents, because she simply does not have the talent she suspects herself to have. All of her efforts have been wasted, in terms of fulfilling her imperfect duty of cultivating her talents. She has labored to cultivate a talent she lacks, and has wasted the time she might have spent cultivating a talent she actually possesses. Only an omniscient agent could accurately know what her particular aptitudes actually were, and which she could develop. A finite agent might spend his entire life laboring tremendously to cultivate talents he mistakenly identifies himself as having. And despite all of his efforts, he has still not managed to fulfill his imperfect duty, because no matter how great his efforts, if he is working to develop a skill he simply lacks the aptitude for, he is not cultivating one of his talents.

Kant's given imperfect duties of benevolence and talent cultivation both require omniscience to guide action. Because humans are finite beings who lack omniscience, these particular imperfect duties cannot apply to them, because they are "impossible in accordance with practical rules" for finite beings.

But the problem is not isolated to these two imperfect duties that Kant provides. An agent will also require omniscience to determine what further imperfect duties he has, given his contingent nature. This is an even deeper omniscience requirement. Even when an agent was uncertain how to act to satisfy his imperfect duty of benevolence, he at least knew which imperfect duty it was he needed to satisfy. Kant clearly states that the four examples he provides are not an exhaustive list of the duties of a rational agent. (G 424) An agent will need to discern what his further imperfect duties are, and this will depend on what he must will in the future, but the finite beings who lack knowledge of the future cannot know what they must will, and cannot therefore know what their further imperfect duties are.

It may be surprising to notice that different rational beings will have different duties according to the Categorical Imperative. Kant has already stated that he does not intend the moral law to guide only human action, but the action of all rational beings. "The principles should not be made to depend on the particular nature of human reason…but…should be derived from the universal concept of a rational being in general." (G 412) I will not try to explain what Kant means by "rational" here. I only mean to point out that Kant already suggests the idea that there might be many kinds of rational beings, distinguishable by their contingent natures and circumstances. As Barbara Herman has pointed out, "all rational beings are subject to the same fundamental practical principle-- the Categorical Imperative…[but] not all rational beings will have the same duties. The duties they have… vary as their natures vary." She illustrates her point by inviting us to consider entirely self-sufficient rational beings she calls "angels". These beings can know a priori that they will never require the aid of another rational being, so they have no imperfect duty of benevolence. Since they know that they will never be forced to will that others help them, it is no contradiction in their wills to act on a maxim of not helping others in need. (PMJ 59) Imperfect duties depend on what an agent will have to will in the future, and they depend on the circumstances and limitations of the particular rational beings to which they apply.

Finite rational beings such as humans have general imperfect duties of benevolence and talent cultivation as a group, but they will also have more specific imperfect duties depending on what they must will individually. These further imperfect duties might be considered subsidiaries of the imperfect more general duty of benevolence. Kant himself would likely resist this reading. In the Doctrine of Virtue he states that

Given Kant's statements on this point, I will consider these further imperfect duties as distinct imperfect duties relative to a particular rational being, not as duties constitutive of the more general duty of benevolence, which has a subtly different ground of obligation.

Beyond the general duty to benevolence which has already been shown to require omniscience, more specific duties will be required of the being who is bound by the imperfect duty of benevolence, specifically those actions that the being will later need performed. Because the being is finite, it simply cannot perform all benevolent actions all the time, so to ensure that its imperfect duties are covered it must choose among possible benevolent actions.

A finite being must choose the benevolent actions specifically tailored to its future (and unknowable) needs. This is a direct result of the way imperfect duties are derived. The general imperfect duty of benevolence is the result of the inevitable fact that this finite being will require the aid of others. But needing aid generally is inevitable for all finite rational beings in a way that needing a specific type of aid is not. Consider any human, let's call her Fiona, who will surely require some kind of aid in her life, but might never require the specific aid of injected insulin. Fiona still has the general imperfect duty of benevolence toward her fellow rational creatures, but she has no duty to the specific aid of injected insulin. Because Fiona will never be forced to will that others help provide her with insulin, she has no imperfect duty to help provide others with insulin. For Fiona, there is no contradiction in her will to universalize the maxim "I will not aid in the production or administration of insulin". Such a world is surely possible, and though it might contradict the wills of other rational creatures who would require insulin, it is no contradiction of Fiona's will that she will and act on this maxim.

At this point one might object that this problem could be avoided if an agent acted benevolently in all the ways the agent might later require aid. But this is not an option open to a finite being such as a human, because the number and variety of misfortunes in which a human might require assistance will certainly overwhelm the energy and time available to any human. Finite beings such as humans simply can't perform every plausible type of benevolent action they might require, and so they must choose some and avoid others.

But it is of critical importance which benevolent actions humans choose and which they exclude, because they must match up with the particular human's contingent and mysterious circumstances. Although it turned out that Fiona had no imperfect duty to produce or administer insulin, surely she has some imperfect duty to a benevolent action of another kind which she will later require. Let's say Fiona has a heart attack and requires a pacemaker when she is 50 years old. Because she eventually requires a pacemaker and wills that others help her with the production and implantation of this pacemaker, she has an imperfect duty to the production and implantation of pacemakers. Fiona could not will the maxim "I will not help with the production or implantation of pacemakers" without a resulting contradiction in her will. For Fiona, or any particular human, what particular duties bind her will be dictated by the contingent facts of her life. Should Fiona not consider the particular way in which she will require help and not act on a maxim to promote that kind of help, she will have failed an imperfect duty.

There is nothing to suggest that an agent errs in following an imperfect duty that does not apply to her, unless it conflicts with the fulfillment of other duties she actually does have. For finite beings, acting on imperfect duties you don't have decreases your finite supply of time and might prevent an actual duty from being fulfilled.

Thus, imperfect duties have different content for different rational beings. Rational beings are bound by the imperfect duties that do (or will) apply to them, and finite beings cannot prepare for every imperfect duty that might apply to them; instead they must choose to will particular maxims, and they must choose correctly.

The fundamental problem for finite rational beings is that they lack the "God's eye view" we have assumed in describing them. We know, because we have stipulated, what Fiona's imperfect duties are and are not, but she cannot know this "because to do so would require omniscience" or at least a kind of foreknowledge she lacks. Kant claims that happiness cannot be the basis for the practical reasoning of finite creatures, because they cannot know with any certainty which actions will produce happiness. This same lack of knowledge prevents finite rational beings from being able to determine what their further imperfect duties are.

The omniscience required to discern further imperfect duties is even more far-reaching than the one required by a eudaemonist principle. Obviously, we can't know with the kind of certainty that Kant requires what action will produce the most happiness, but we can still have fairly plausible expectations. Greater wealth and longevity might well improve our happiness more than detract from it, so we aren't entirely foolish in taking these means to the end of happiness. But, there is no comparably plausible way to predict what particular needs we may have or misfortunes we might suffer in the future. The shaky medical predictions that can be made on the basis of a family history, for instance, provide only a tiny and uncertain glimpse of the complete picture of the kind and number of needs a finite agent will have, and thus which maxims that agent is required to will or avoid willing.

As we have now seen, the Kantian requirement of imperfect duties requires omniscience of its agents in two crucial ways. First, precisely parallel to the way omniscience is required to maximize happiness, omniscience is required to act on the general and given imperfect duties of benevolence and the cultivation of talents. Second, omniscience is required in order to determine what a particular agent's further imperfect duties will be. Now it is obvious that humans lack this required attribute of omniscience, so for humans, acting on and deriving imperfect duties should be considered "impossible in accordance with practical rules." Humans, and other finite beings who lack future-directed omniscience, lack the abilities necessary to fulfill their imperfect duties, so this aspect of Kant's moral theory should not apply to them.

Kant believed that eudaemonist moral theories were both morally illegitimate and practically ineffective in achieving their stated goal of achieving happiness. Similarly, the fact that imperfect duties require omniscience both to be derived and performed will show both that these duties cannot morally apply to finite rational beings such as humans as well showing that they cannot provide a hope of happiness.

To ensure that happiness can actually be distributed in accordance with virtue, Kant provides a "postulate of pure reason" to the effect that God will ensure that virtuous actions are rewarded with happiness. Such a God must be postulated in order to hope for happiness. As Kant states in the second Critique "for the sake of this wish [for happiness] the step to religion has been taken -- then for the first time can this ethical doctrine also be called a doctrine of happiness, because it is only with religion that the hope of happiness first arises." (CPrR 130)

Even if virtuous action and happiness are connected by God, it should now be clear that there is a serious rift between human action and the virtuous actions that come from imperfect duties. For finite rational beings, acting on imperfect duties is impossible; this entire class of virtuous actions is inaccessible. If there is to be any hope for happiness or even virtuous action in Kant's system, it must come from perfect duties. These duties have formerly been seen as formal and very few. If virtue and happiness are to be salvaged for Kant, this will require a radically new and more robust interpretation of these perfect duties.