Knowledge that a particular action will produce some good result need not be derived from practical reason.1 I might know that by easing trade restrictions international economy would be improved. But if I have no opportunity to make such changes, my knowledge remains no more than a theoretical estimate. Practical reason comes into play when the agent’s own action is what is in question. But not all actions spring from judgments of practical reason. A man may act on blind impulse: he may go beserk and kill anyone he happens to meet. We would call a man in this condition irrational. Another may act from instinct, as when he jerks his hand off a hot surface. Such action would be nonrational, involving no reasoning at all.
Under what conditions would we call an action rational? For Kant, the answer is simple: an action is rational, done with practical reason, when the agent acts from some guiding principle. It is not enough that the agent knows what he wants and how to get it done. The madman may know he wants to kill and how to do it, but he would not be acting by any rule. Roughly speaking, a person acts rationally when his action arises from and is consistent with some personal policy. If I walk to the corner to cross the street rather than cross in mid-block, I act from the policy, “I will cross streets at corners.” Such a policy, or general rule of acting, Kant calls a principle. There are, of course, many possible reasons for crossing at the corner. I might happen to be there at the time and simply cross. But if I walk down to the corner to cross, because it is my policy to do so, then I am acting on principle. In Kant’s words, I am acting not only according to the policy, but because of it, or, “from the idea of it.”
Everything in nature works according to law. Only a being who reasons can act from the idea of law, that is, act on principle. We call this ability the will. And since the ability to determine one’s actions on principle requires reason, we can see the will and practical reason as one and the same ability.
Every event, human or nonhuman, occurs according to some law. Walking down the block accords with the law of gravity, as much as does the flow of water over a waterfall. Rational beings act not only according to laws of nature but use them to advantage. Applied science is nothing more than using laws of nature to our advantage (and in some cases to our disadvantage). In contrast to laws of nature, to which we always conform, there are other laws which we ourselves create: civil and moral laws. The basic difference between the laws of nature and laws of man, is that the former are purely descriptive. They are universal statements of the unvarying ways natural events occur. Being descriptions only, they do not command; so we do not speak of disobeying them. Laws of man, on the other hand, are prescriptive; they do command and are indeed often violated. But mere knowledge of a law of man does not by itself result in compliance. The link between knowledge of the law (principle, rule, or policy) and the complying action is the will, which Kant identifies with practical reason, our capacity for translating policy into action. Given favorable circumstances, when we will to act, we act. And if we act because of some policy, we act rationally. This process—knowing, willing, acting—is the exercise of practical reason.
Many times we do not act in accordance with our policies. The philanderer who has the policy, “Take advantage whenever you can,” may on occasion resist his impulses. The saint may occasionally slip. When we act against principle are we therefore acting irrationally? Not necessarily. We may act contrary to one principle, true, but that does not mean that we act without any principle whatever. Sometimes we reject one policy in order to act from another policy. To do so, of course, implies an inconsistency somewhere in our values. When we act rationally we act on some policy or other—but not always on a good policy.
Should there be anyone whose reason infallibly determined his will, he would necessarily choose to perform those actions which reason recognized to be objectively necessary. His will would necessarily concur in the decisions which reason, free from the influence of desire, commanded as necessary for him to perform, that is, as good. But for someone else, reason may not be strong enough to coerce the will, because of the influences of subjective inclinations and desires which oppose reason’s decision. Whenever a will lacks complete harmony with reason—as in our human condition—then those actions which reason recognizes to be objectively necessary may not be chosen by the will. When such a decision is subjectively contingent, then reason’s influence on will through objective laws is called constraint. Constraint is defined as the relationship between objective law and a will which is less than completely good; it denotes the influence which reason, in recognition of law, exerts on the will of a rational being, a will which need not necessarily obey.
Consider the case of an individual who always and by his very nature acts in complete accord with his reason. His reason will have as its basic policy: “Do that which is good in itself, rather than what you want to do.” If a particular action is good in itself, then it is good on its own merits, that is, objectively good. And if objectively good, then reason will dictate that it be chosen without reservation. In Kant’s language, such an action is objectively necessary. The person whose reason completely rules his life will always choose to perform these objectively necessary actions. They become subjectively necessary by becoming the objects of his personal policy; and they become such on a priori grounds, not from inclination. We might express this in more contemporary language by saying that such a person naturally chooses to do what reason tells him he ought to do.
But for one in “the human condition,” beset with the pull of desires, inclinations, and incentives contrary to what reason commands, actions do not always conform to reason’s direction. We may know that something ought to be done, yet choose to do quite the opposite. There is nothing very mystifying about it. We simply choose to do other than we ought to do. When reason’s influence is weak, i.e., when we lack will power, we follow the sweep of inclination.
But even when we shun doing what we ought to do, we still feel a certain tug toward our duty. This “feeling,” which we have already discussed, Kant calls constraint, the relation of a will to its obligation. The relation in question is that between a will not entirely determined by a recognition of its obligation and reason which recognizes that obligation. Such is the case of a will not completely good, yet which still feels “that spark of celestial fire called conscience.”
Since man, as a rational being, is capable of acting from principle, we must examine the principles from which he acts, so that we may understand the relation of reason to will and of morality to acting.
The rational awareness of an objective principle which constrains will is a command of reason; and the expression of this command is called an imperative.
This little paragraph is disarmingly simple, but two important points are raised. First the command of reason is a priori, arising in the moral understanding. It is the recognition of a law which constrains a will not completely good. For a completely good will there would be no constraint, since the will would naturally follow the objective principles. But in our imperfect state we tend to take an objective principle as a maybe-I-will-maybe-I-won't kind of thing. Our reason says, “Do!" while our will says, “Not necessarily.” For us, such principles are constraining, morally impelling, but not subjectively necessary. We are aware that nothing compels us to act on them; and distaste, aversion, or laziness sometimes lead us to do something else. Nevertheless, the rational recognition of the principle is a part of the situation: for the awareness itself that a principle is objectively binding constitutes a rational command for us.
The second point is important for an understanding of Kant’s terminology. Commands of reason, he says, are expressed (formulated) as imperatives. If someone were to rewrite the Foundation, he might substitute the expression “Moral rule” or “ought-statement” for “imperative.” Imperatives are frequently associated with orders or other verbal commands. Kant does not use “imperative” in a grammatical sense but in a logical sense of “prescription,” which includes both grammatically imperative sentences and ought-statements. Any ought-statement pertaining to human action will be classified as an imperative.
All imperatives are expressed by “ought,” a word which indicates the connection between reason’s objective law and a will which, because of its own nature, is not bound to obey this law. The connection itself is called constraint. Imperatives assert that something would be good to do or that something else would be good not to do; but they assert such judgments to a will which may not choose to act simply because something is suggested to it as a good thing to do. However, a practical good does determine the will by means of ideas of reason: in this case will is determined not on subjective grounds but on objective grounds, those which are valid for every rational being as such. Subjective grounds, such as the pleasant, affect the will through personal experiences; they provide no universally valid principles of reason since they relate only to the sensations of this person or that.*
Although Kant asserts that “all” imperatives are expressed by an “ought,” his own formulations of the categorical imperative never use the term “ought” but are in the imperative mood. This is only a minor point. “Ought” does express the relation between an objective principle and a will subject to the pull of inclination. “Ought” expresses constraint.
"Imperatives assert that this would be good to do, or that would be good not to do.” This fundamental axiom interrelates a good action, its recognition by reason, and the human will. Kant assumes the ancient principle that the will is determined by the recognition of goodness and seeks only what appears to it as good. When we choose to act, we act for some goal which appears to us, in one form or another, as good.2 But there are many kinds of good things which we might catalog in the following way:
a. Subjective goods: things good for me, as I see them; things which I judge good in their relation to me. For example, my health, security, job, money, etc. A thing may seem good to me yet bad to someone else, as in the case of my marrying a person desired by another.
b. Objective goods: things good in themselves, for everyone, in every situation, not dependent upon particular persons or upon particular situations. For example, happiness, virtue, love, knowledge, salvation, etc.
We must not make the distinction as narrow as Kant does, as being between desire for pleasure and reason. Some subjective goods may be unpleasant; the pleasure may not be the main considration. One need not consider pleasure at all when he judges something good for himself. Must a political candidate look upon the presidency as a pleasant goal? (He may, but I daresay he should know better.) Kant’s point does not depend upon a hedonistic distinction. Objective and subjective goods are distinguished in one basic way: whether or not what is judged good is judged on the basis of reason or on the basis of personal inclination or interest. If reason represents something as good for every rational being (and virtue is the example par excellence), then it is an objective good; but if reason represents it as good only for me or only for a particular personal situation, then it is a subjective good.3
Both kinds of good, objective and subjective, are recognized by judgments of reason. Rational activity may be directed toward a subjective or objective good or toward both at once. It would be a mistake to say that an object of one class can be chosen only by reason and that an object of the other can be chosen only by desire. Objective goods are indeed chosen by reason, or better, recognized by reason, but so are subjective goods. The difference is in the basis for choice or recognition. Objective goods are good in themselves, valid goals for everyone; subjective goods are good for me but not necessarily for everyone nor in every situation. Furthermore, as we shall see, both varieties of good can be related to the will by an “ought,” but not on the same grounds.
In a footnote to the last paragraph, Kant presents an interesting corollary.
* When desire depends upon an experience of some need, we call it inclination. On the other hand, when a will which is not necessarily subservient to objective laws of reason follows the rules of reason, we call it interest. (Thus the divine will cannot be said to have an interest.) But the human will can have an interest in something without choosing to act from it. To have an interest means to be interested in the action: the action itself interests me, since my will is guided only by the laws of reason. But to act from interest means to be interested in the result of the action: my will follows the counsel of reason solely to achieve the anticipated pleasant results of the action. In Section I we said that when we perform an action from duty we must disregard any interest in the results and confine our interest to the action itself and its basis in reason (i.e., law).
Although not a new point, this footnote presents a partial answer to the question, “When we act from duty, don't we want to act that way? Don't we have an inclination to do our duty?" The answer is both yes and no: an inclination for the performance of duty, yes, but not necessarily for what a specific duty entails. We can choose to do something without any desire for the results which the action will produce. To some extent we do want to do our duty, over and above the awareness that we ought to. And it is a different kind of “feeling” than in those cases where we find some pleasure in acting. But exactly what the difference is has not been made clear.
We have been talking mostly about a human will. What of a Holy Will, or God’s Will? If we can consider God, and perhaps the angels, as rational, do objective principles bind them too? “Ought” God do anything?
Although a perfectly good will would in every case be bound by objective laws (laws of goodness), we cannot think of it on this account as obligated to obey these laws since by its very nature it is moved to act only by the idea of goodness. Thus no imperative binds the Will of God nor, more generally, a Holy Will. “Ought” does not apply to such a will, since its choice necessarily accords with law. Consequently, imperatives express only the relation between objective laws and the subjective, imperfect will of this or that rational being (e.g., a human will).
No, we can never say that God, or a Holy Will, “ought" to do anything; the word “ought” loses its moral meaning in such a use. For “ought” (in a moral sense) expresses the relation between something which reason presents as good without condition (a good-in-itself, an objective good to be sought by all rational beings), and a will which is so constituted that it does not have to choose that good. But if there be a will which by its very nature is completely guided by reason, then such a will cannot know obligation. In plain terms, such a Holy Will always and necessarily wants to do what it ought to do. Its inclinations are always and necessarily subject to an objective, rational appraisal of the good. Moral obligation is meaningless for such a being, as much as for non-rational beings. The latter have no conception of ought; the former does not require it. “Ought” applies only when an alternative choice is possible, but for a Holy Will there is no such possibility. “Ought,” then, and the whole of morality, is a human concern. God has to become man in order to drink the cup of duty.4
Kant’s analysis to this point can be summarized as follows:
- Rational beings not only act in accordance with laws of nature but may also act because of reason’s recognition of certain laws.
- When reason and will are completely harmonious (as in a Holy Will), the dictates of reason are always followed. Human wills lack such a harmonious relation with reason: what reason dictates may not always be obeyed.
- Constraint, that unique tension between reason and will, occurs when reason presents a particular action as objectively necessary (good for everyone) but the will is inclined to an opposite course of action.
- We express this particular stress between reason and recalcitrant will by an “ought.” “Ought" specifies the necessity of performing some action which is not necessarily desired by the will. Such “ought” statements, called imperatives, apply only to wills not completely good.
All imperatives command either hypothetically or categorically. A hypothetical imperative commands the will to perform a particular action as a necessary means to some end which the will desires (or might desire). A categorical imperative commands the will under objective necessity, without reference to any ends.
Now every practical law points to some possible action as being good and thus necessary for any will which is guided by reason. All imperatives then are expressions which determine an action to be necessary in virtue of some principle found in a will which is good to some degree. If the action appears good merely as a means to something else, the imperative is hypothetical. But if the action is shown to be good in itself and hence necessary as a matter of principle for a will that follows reason, then the imperative is categorical.
The distinction between hypothetical and categorical imperatives is related to the distinction between hypothetical and categorical assertions. A hypothetical (conditional) assertion is one which shows the relation between an event and its condition or supposition: “If it rains, the vote will be small.” A categorical assertion is an unqualified statement: “Some Vietnamese are short.” Similarly, a hypothetical imperative expresses a command to the will but does so under some condition: “If you want to be polite, you ought to study a book of etiquette.” The “ought” is conditional upon the desire to be polite. Not so with the categorical imperative; it depends upon no conditions whatever. “Keep your promise” expresses a command which is unconditionally necessary.
These paragraphs make explicit some premisses which explain how imperatives interrelate the will and reason. The imperatives commanding the action as necessary are implicit judgments that the action commanded is good. These are not two separate judgments: rather, the judgment that an action is good is also a command that it be performed. Furthermore, if practical reason judges an action as good, it judges it to be necessary. An action which is good ought to be done. A hypothetical imperative expresses the necessity of pursuing the means to some desired end: “If you want a strong bookcase, you ought to use screws.” The use of screws is presented as a necessary step in building the bookcase wanted. A categorical imperative, on the other hand, expresses an objective necessity, an action which is not good as a means to something else but good in itself and thus necessary for its own sake. Only an action good in itself can be presented as an action necessary without any conditions attached to it.
The moral problem is a double one: how does reason know what is good; and, v/hen the good is known, how does reason move the will to choose it? Imperatives, then, serve a dual function, corresponding to the two sides of the moral problem.
An imperative, then, tells me which of those actions I am able to perform are good; it does this by stating the rule commanding such an action and relating this rule to a will which may not perform the action merely because it is a good action. For in some instances a person may not know that a particular action is good; and even if he knew that it was good, he might still have a maxim which conflicts with the objective principles of practical reason.
In order to see how an imperative can inform us of the good, we have to remember that goodness is a determination of value and that valuating is a function of reason, not of will. Let us suppose I am in a situation which confronts me with a number of alternative courses of action. I do not know initially which of these will be good to do and which not. If, however, I can dertermine which of the courses of action is commanded by reason (whether hypothetically or categorically), I can conclude that that action at least is a good one. For reason could not consistently command an action as necessary if reason did not at the same time judge that action to be good. But I cannot reverse the procedure: the judgment that an action is a good one does not permit me to conclude that it is commanded by reason. There might still be an alternative action in that situation which is also good and which is commanded by reason, and that action must take precedence over the first good action. For instance, suppose I am weary but have promised to visit my sick friend. Now surely taking a nap would be a good thing to do, but since I have promised, reason commands me to keep my promise. Practical reason, then, forces me to conclude that not only is keeping my promise a good thing to do, but it is better (in this situation) than taking a nap.
The more serious problem arises in determining how reason, simply by judging that visiting my sick friend is better than taking a nap because it is my duty, can move my will to choose my duty rather than to take a nap. Where does reason get its power? To answer this question we must look more closely at hypothetical imperatives, which are directives based upon subjective interests. By understanding how they function as commands, we will be better able to comprehend how an objective command of reason can move a will without relying on desire.
The hypothetical imperative merely says that the action is a good means to some possible or actual goal. If it relates the action to a possible goal, we call the imperative problematic; if the goal is actually sought, the imperative is called assertoric. But the categorical imperative, which declares an action to be objectively necessary in itself without any reference to any goal whatsoever, we call an apodictic imperative.
The terminology used in this section is not currently in vogue. Kant borrowed his terms from the names of the logical forms of judgment, as he did the terms “hypothetical” and “categorical.” The term “problematic” means, roughly, “contingent.” A problematic judgment is based upon some condition which may not obtain, such as: “Iƒ he reaches the 30-yard line, it will be a first down.” An assertoric judgment (one which “asserts") is likewise a conditional statement, but one in which the condition has already been fulfilled. Thus, “If he reaches the 30-yard line—and he has!—its a first down.” Or, “Since he reached the 30-yard line, it’s a first down.” The term “apodictic” means roughly the same as “objectively necessary.” It is a statement of necessity, such as: “A team must progress ten yards in order to make a first down.” Apodictic judgments are unconditional.
The first kind of hypothetical imperative to be discussed is the problematic imperative (later called an imperative of skill).
Whatever a rational being can do, that we may take as a possible objective for some will. Consequently, there are an unlimited number of possible principles of action, referring to the means necessary to achieve these objectives. Every practical science proposes certain goals to be attained and uses imperatives to describe the means for attaining the goals. These we can call imperatives of skill. In such imperatives the relevant question is not whether the goals are reasonable or valuable, but only what must be done to attain them. The means used by a physician to cure his patient and the means employed by a murderer to poison his victim have equal value to the extent that each accomplishes the respective purpose.
A young child cannot know all the goals which he will seek during his lifetime and so his parents teach him skills in choosing the correct means to a great variety of arbitrary goals. Since parents cannot be certain which goals will be chosen, they teach skills regarding goals which might be chosen. But in their zeal to teach the skill of choosing the correct means, they frequently neglect to train the child’s judgment about the value of those things he may later choose as his goals.
The imperative of skill commands nothing more than the means required to achieve a certain end. We learn the means-end relationship from many sources, and we are well aware that there may be many ways to achieve a particular end. Take the situation of a man who wants to burn a pile of leaves. He might use a match, a magnifying glass, a cigarette lighter, a napalm bomb, a flame thrower, or a fire-by-friction kit. A variety of things will start the fire. But in a particular situation (with the available means at hand) we would not say, “If you want to burn leaves, you ought to use napalm,” but would say something like, “If you want to burn the leaves, you ought to use a match.”
Where is the necessity in this situation? There is surely no obligation to use a match. There is only the practical necessity of using the appropriate means to achieve the desired end. We are concerned only with the means necessary to achieve some end, whatever its value as an end. Cyanide is much better than aspirin for poisoning someone. Kant is aware of the dangers of such abstraction. We may become so engrossed in the means to some end that we fail to evaluate the end itself. Scientists during World War II were more intent on detonating a nuclear device than in evaluating the results they hoped to achieve thereby. But whatever the value of the end, the imperative of skill expresses the necessary means for achieving it.
But we must remember that the skill imperative is conditional. The necessity of performing the action rests on the desire for the end. If a person does not want to burn the pile of leaves, reason has no basis for its command to use a match. Furthermore, since it derives its practical authority from the person’s desires, it is a subjective imperative. The necessity relates to one person only and involves only one particular goal for that person. Some other desired object or purpose will demand other means, involving a different imperative. This individuality of ends will help to distinguish the problematic imperative from the assertoric imperative.
One goal, however, we can assume that all rational beings actually pursue (at least those rational beings who can be subject to imperatives in the first place, i.e., dependent beings) and, furthermore, that everyone pursues this goal through some natural necessity. This goal is happiness. Those imperatives which show certain actions as necessary for achieving happiness are called assertoric imperatives. The action is proposed not as a practical necessity for some merely possible goal, but as necessary for a goal we know for sure that everyone pursues, since it springs from his very nature. When someone has skill in choosing the means to his own well-being (i.e., his happiness), we call him prudent, in the narrowest sense of the word.* Imperatives which refer to the means of achieving happiness, while they are rules of prudence, still are merely hypothetical: the action is commanded as necessary, not in itself but as a means to an end.
* “Prudence” has two meanings: (a) the interpersonal sense, denoting a man’s skill in influencing others in order to use them for his own purposes; and (b) the personal sense, referring to his own ultimate benefit. Thus any advantage of interpersonal prudence must finally be grounded in personal prudence. One who is prudent in the interpersonal sense, but not in the personal sense, should be called clever or shrewd, but not prudent, properly speaking.
The second type of hypothetical imperative involves a different kind of good. First, we know that men do seek the goal, happiness. Secondly, it is an end which encompasses all our activities to such an extent that some philosophers (e.g., Aristotle) believed happiness to be the proper foundation of morality.5 Thus, the imperative of prudence is assertoric: “If you want to achieve happiness—and you surely do!—then you ought to. . .” The more we know of the proper means to happiness and act on the knowledge, the more prudent we are. The belief that some action is a proper means to happiness finds expression in an imperative of prudence. The necessity of acting still depends upon our desire for happiness but, since everyone desires happiness “through some natural necessity,” the authority for the imperative is a broader one than that for an imperative of skill. It is not derived from a desire for some particular objective but from our human nature itself. Yet the necessity is not thereby objective, since it is not based upon rationality, but upon our specifically human nature. We know man to have this goal through our experience, not through reason.
Our natural goal is happiness, true, but reason is not capable of leading us to it, even if we had holy wills. Too much depends on circumstances beyond our own control, such as health, sufficient necessities of life, harmonious relations with our fellow men, etc. There are no guaranteed means to happiness. Must we despair of achieving our goal in life? Not at all. Although it is highly unlikely (due to stepmotherly Nature) that we will fully achieve our natural goal, we can surely achieve our moral or rational purpose, which is to be worthy of happiness. Thus the rational goal is in accord with man’s natural goal, and not in opposition to it. The seeming opposition arises when an action which reason represents as necessary for being worthy of happiness is not the same as that action which desire or inclination suggests as a means for achieving happiness itself. This situation is the model case of moral struggle and exemplifies the fundamental difference between the hypothetical and categorical imperatives.
Finally, there is that imperative which straightforwardly commands us to do something without basing the command on some goal to be achieved by the action. This is a categorical imperative. The categorical imperative disregards the particular circumstances of the situation as well as the proposed results of the action; instead, it commands in virtue of the form of the action and from the law governing such situations in general. The essential moral goodness of such actions depends on the motive, no matter what the results might be. Such is the imperative of morality.
When a particular action appears necessary as a means to happiness, its necessity is still conditional, depending upon the agent’s desire for that end. When reason represents an action as necessary in order to be worthy of happiness, however, it does so regardless of whether happiness will actually be achieved. The action, in fact, is not presented as a means at all; reason does not present an objectively necessary action to the will in this way. No, the will is presented with an action which ought to be done solely because the action is necessary on its own account. The imperative which presents the action to the will does not rest upon some condition, nor does it depend upon some desire of the agent. Rather, it presents the action as categorically and unconditionally necessary, necessary in spite of any possible conflicting desires or inclinations. The goodness of such an action is not found in the result to be achieved but in the action itself or, more accurately, in the willing of the action, in our motive for doing it. No matter what the result may be, the action must be performed because it is the moral thing to do. Thus, says Kant, only the categorical imperative is the imperative of morality, for only a categorical imperative commands what is unconditionally and morally necessary.
The difference between these three kinds of imperative can be stated in the different ways each constrains the will. To make the distinctions clear, let us call them respectively rules of skill, counsels of prudence, and commands (laws) of morality. We might also call the first a technical imperative (concerned with art), the second a pragmatic imperative (concerned with one’s own benefit),* and the third a moral imperative (concerned with free conduct as such, namely, with moral action ).
Now only law contains the idea of unconditional, objective, and universally valid necessity; thus laws must be obeyed, no matter what our desires may be in the matter. Counsels, on the other hand, while they do involve a kind of necessity, do so under a contingent and subjective condition, that a man believes the action counseled to be a means to his happiness. But the categorical imperative is not subject to any conditions and is a command in the strict sense of absolute, practical necessity.
* This seems to me the proper meaning of “pragmatic.” Pragmatic sanctions, for instance, are derived from statutes enacted for public welfare, not from any natural law pertaining to states. Pragmatic history is that which is written to teach prudence, by showing how we might do better for ourselves than other peoples have done in the past.
These, then, are the three types of imperative:
a. Technical Rules of Skill. These indicate the necessity of employing a particular means to a particular end. The necessity of acting is subject to a desire for that end. If one desires to achieve that result, then he must (ought to) use these means.
b. Pragmatic Counsels of Prudence. These depict the necessary actions one must perform in order to achieve happiness. The necessity depends upon the agent’s belief that the action will indeed produce happiness. A person desires happiness by his very nature as a human being and believes the indicated action to be a means to happiness; then he must (ought to) perform the action.
c. Unconditional Laws of Morality. These command certain actions as necessary without regard for the desires of the agent, nor for the goodness of the results of the action. The action must be willed because it is objectively necessary, an action which is good in itself.
We might also catalog these imperatives according to the breadth of the conditions from which they derive their authority. A rule of skill is dependent upon one particular desired goal of one agent. A counsel of prudence, extending to all the actions of an agent, is still dependent upon his own desire for happiness and belief that the particular action in question will help achieve it. A law of morality is universal, unconditional, and independent of any agent’s desire. Moral law prescribes that anyone in a given situation must obey that law if he is to be moral. Morality is a matter of unconditional necessity and does not depend upon wanting to be moral. A duty is a duty, whether or not the agent wants to do it.
These then are the three ways by which reason commands the will to perform an action. The question which we have postponed—how does reason influence the will?—may now be raised.
But this raises the question: how are all these imperatives possible? The question does not ask how we come to perform the action which the imperative commands, but simply how an imperative constrains the will by imposing obligation on it.
The question of the possibility of imperatives is not the psychological question which deals with the processes of human decision and resultant action. Kant is asking how the imperatives impose necessity on the will. How does an imperative constrain the will? The question deals with the relationship between reason and will, not with actual human behavior.
We do not require a lengthy explanation to show how the imperative of skill is possible. Anyone who chooses a particular goal must choose the necessary means to that goal—at least if he chooses with reason, and if the means are within his power. So this imperative is an analytic command for the will. When I choose to act in order to produce some effect, I must at the same time think of myself as the causal agent who uses the means to that end. The imperative, consequently, simply commands me to perform those actions necessary to obtain the goal I have chosen. Of course, synthetic propositions must be used for deciding what means will produce the desired result, but they refer only to the actions which will realize the goal; they do not refer to the basis in the will for choosing to perform the actions. For instance, if I want to bisect a line according to an exact method, I must draw from each end two intersecting arcs. I know this by means of synthetic propositions of mathematics. But when I recognize that this is the only way to produce what I want, then it follows analytically that in choosing to bisect a line I also choose to perform the operation necessary to do it. To think of something as an effect which I cause by a particular action, and to think of myself as acting in that particular way, are one and the same thing.
Suppose I want to make ice cream. I have learned through experience that ice cream can be made only by using some type of milk base. The proposition, “Ice cream is made with a milk base,” is not analytic for in the future chemicals may serve as well as milk base for making ice cream. But as things now stand, if I chose to make ice cream, I must also choose to use a milk base in making it. That is, if I desire a certain end, then to be consistent I must also will the means.
In a consistent way of willing things, if we will a particular end, we must likewise will the means to that end. Or as Kant puts it, in the idea of willing the result, we necessarily inelude willing the means to that result. And this is what Kant means by calling the imperatives of skill analytic. If I fully choose to obtain the end (and do not merely wish I had it), I must by necessity choose to perform the means to the end. Even though the cause-effect relationship is learned through experience, it is the means-end relationship7 which practical reason recognizes as necessary. Whether or not I actually perform the necessary means is another matter and does not involve the means-end relationship itself. No one outside the trade has an obligation to make ice cream, but a man who wants to make ice cream, who knows that a milk base must be used, and who yet persists in trying to make ice cream with cement and turpentine, is simply irrational.
If we could precisely define happiness, then the imperatives of prudence would be the same as the imperatives of skill and thus would be analytic. Here, as well as in the former case, choosing the end would, if the agent chooses rationally, impose a rational necessity of choosing the means (within the agent’s power) which would produce that end. Unfortunately, the idea of happiness is so vague that no one can definitely and consistently say what he wants and chooses, even though he wants and chooses happiness. The reason is that all the components which he includes in his idea of happiness are empirical, i.e., derived from experience; but his idea of happiness is itself absolutely complete, including total wellbeing, both in the present and in any future circumstances.
But not even the most farsighted and capable finite being can get from such an idea any definite conception of what he really wants. If he believes happiness to consist in wealth, how much trouble, envy, and plots will he thereby bring upon himself? If he thinks it consists in knowledge and foresight, how many inescapable evils (now concealed from him) will he discover? How many more desires will be aroused—as though he did not already have enough—by his discovery of new needs? If he thinks happiness to be long life, can he be sure that it will not be merely one long misery? Or suppose he simply takes happiness to be health; how many times has he been held back by physical illness from those excesses which perfect health would have made alluring? We could go on and on. In brief, we are not all-knowing; and so we simply cannot determine from principle and with certainty what to do to become happy. Since we cannot find any definite rule for achieivng happiness, the only rules we can act on are empirical counsels, such as rules of diet, frugality, courtesy, restraint, and so on, which we know from experience will best promote a happier life generally.
We can conclude, then, that imperatives of prudence cannot command, strictly speaking: they cannot present actions as being practically necessary on objective grounds. These imperatives are better called counsels (consilia) rather than laws of reason (praecepta), since it is impossible to find a sure and universal answer to the question, “What action will make a rational being happy?" Happiness is an Ideal of the imagination, not of reason, and is derived from—experience so it cannot be the foundation for a command of reason. We can never expect experience to show us, beyond doubt, the action which would produce the entire series of consequences—a series which is in fact endless.
Even if we could determine with certainty the means to happiness, the imperative of prudence would then be an analytic proposition; it would differ from the imperative of skill only in having its goal actually desired, whereas the imperative of skill states its goal merely as possibly desired. But since both command an action as a means to some desired end, both imperatives are analytic, for they command that one who wills the end must in each case will the means. And so we see no problem as to how the imperative of prudence is possible.
Briefly put, the imperative of prudence, or counsel of prudence, would be analytic if we knew the precise means-end relationship for acquiring happiness as well as we know those relationships in the technical rules of skill. In making ice cream, we can have a precise idea of what we are hoping to achieve and what must be done to achieve it. And reason perceives that, as matters stand, if we choose to make ice cream, then by practical necessity we must choose the actions which produce ice cream, whatever they may be. In willing happiness, on the other hand, reason still perceives the necessity of acting to produce happiness. The important difference, however—so important that a counsel of prudence is not a command at all for finite beings—is that we have neither a clear idea of what happiness is nor definite ideas of how to achieve it. We have certain notions about happiness, true, but they are vague, confused, obscure, and uncertain. Happiness is so all-encompassing that it would require a superhuman mind to conceive it fully. Consequently, the counsels of prudence are commands only in an advisory sense.8
The only real problem we must investigate is how the imperative of morality is possible. Since it is not hypothetical, we cannot base its objective necessity on a presupposition as we did with hypothetical imperatives. Nor can we show by examples from experience that there is in fact an imperative of morality. It might possibly be that every imperative which appears to command categorically may be subconsciously a hypothetical imperative. Consider the precept, “Do not make a false promise.” Now, we take this as something more than advice for avoiding some evil, as though it said, “Do not make a false promise for you may lose your credit if you get caught at it.” On the contrary, we judge a false promise to be bad in itself and that the imperative forbids it categorically.
But no example can prove beyond question that a will which obeyed a command, apparently influenced by the law alone, was not in fact determined by other influences. It is always possible that some hidden fear of disgrace or a secret dread of other dangers may unconsciously influence the will. How could we prove that no such hidden cause exists when we have no experience of it? But if such a hidden cause does exist, then the so-called moral imperative, which seems to command categorically and unconditionally, turns out to be no more than a pragmatic rule, one which shows us how best to act in our own interests.
The initial problem is that the categorical imperative is not analytic. There is no means-end relationship, since the categorical imperative depends upon no conditional factor as does the hypothetical imperative. It commands universally and unconditionally, with a moral necessity. We cannot justify the imperative of morality by any appeal to experience. We have already seen that no historical instances can be given as unquestionable illustrations of purely moral behavior, simply because we can never be sure there were no hidden motives. In fact, it is possible that there have never been instances of purely moral behavior; it is therefore impossible to use actual examples as proof.
Consequently our investigation of how a categorical imperative is possible must be a priori. If we could prove by experience that there really is such an imperative all we would need to do would be to validate it. But we do know this much: only a categorical imperative can be a practical law. Other imperatives may be called rules of choice, but not laws, since anything which must be chosen merely as a means to some arbitrarily selected goal may itself be considered contingent. As soon as we lose our desire for the goal, the imperative loses its force. But an unconditional command allows the will no option to choose otherwise; it alone commands with that necessity which we expect from law.
The principal reason that examples will not work is that the categorical imperative is a priori: it binds the will under necessity. Necessity can be justified only by an appeal to rationality, not to experience; and since the categorical imperative does command without any conditional presupposition, it is a priori. In the case of making ice cream, we find that using a milk base is part of the process, so in willing to make ice cream, reason demands that we use the milk base. But reason does not demand that we make ice cream. On the other hand, we must keep our promises, no matter what the situation. We have no allowable alternatives.
We have seen how the hypothetical imperative is analytic: the concept of willing the end contains the concept of willing the appropriate means to the end. But in the categorical command of reason, there is no such means-end relationship. Nor will logic alone reveal any particular action to be one’s duty. Any relationship between reason and will with respect to a dutiful action is nonanalytic: Kant calls such a relationship synthetic. The categorical imperative expresses a synthetic a priori relationship between reason and will.
In the second place, there are weighty reasons for our difficulty in determining how the categorical imperative (the law of morality) is possible. This imperative is a synthetic a priori practical proposition,* and, because so much trouble arose in validating synthetic a priori propositions in theoretical knowledge, one could easily conclude that equal problems would arise in practical knowledge.
* I relate the action to my will by an a priori (necessary) connection, without assuming as a condition any desire on my part to perform the action. The relationship is completely objective, based on an idea of reason which has full power over subjective motives. This practical proposition, therefore, cannot assume an analytic relationship between willing an action and having some inclination to perform it. (We do not have such a perfect will.) Rather, this synthetic proposition connects the action directly to the idea of a rational will, yet the action is not analytically contained in that idea.
The discovery of the synthetic a priori proposition has often been cited as Kant’s major contribution to philosophy. While this is not exactly true—others before Kant had been aware of such a class of propositions—Kant was the first to analyze explicitly their role in human knowledge. Kant agreed with Hume that propositions are divided logically into two groups: analytic propositions and synthetic propositions. Analytic propositions are those which are true by the very concepts involved. We need only understand what the subject of the proposition means in order to know that the predicate is true of that subject. Thus, “All bachelors are unmarried,” “All fathers are male,” and “All circles are round,” are analytic propositions. Synthetic propositions are simply propositions which are not analytic. If the predicate adds something to our understanding of the subject, then in forming the proposition we have “synthesized” two nonrelated concepts into one proposition. Thus, 'Washington was the first President,” “All men are mortal,” and “Rose Ann has blue eyes,” are synthetic propositions.
Besides the logical distinction of analytic versus synthetic, there is the epistemological distinction of propositions into a priori and a posteriori. A priori propositions, characterized by necessity and universality, are grounded in reason, while a posteriori propositions are derived from experience and thus lack the characteristics of necessity and universality.9 “All men are mortal,” while true of every man in the past, is not true by rational necessity, since there is nothing inconsistent in the idea of an immortal man. We know that all men die because experience has so taught us; thus the proposition is a posteriori. On the other hand, “All bachelors are unmarried,” is necessarily true, anywhere, anywhen, past, present, and future. It is true because it is a truth of reason and not merely of fact. The concepts are understood by reason to be logically related. The proposition, accordingly, is a priori.
Hume had proclaimed that there is only one way of pairing these propositions: all analytic propositions are a priori, and vice versa; all synthetic propositions are a posteriori, and vice versa. Kant, however, disagreed strongly. It is true, he said, that a proposition which is analytic is a priori, since analytic propositions are constructed by reason; and it is also true that a posteriori propositions are synthetic, since experience does not provide us with material for statements true by the very meaning of their terms. Nevertheless there are propositions which are universally and necessarily true, Kant noted, which are not true because of the conceptual equivalence of their subjects and predicates. They are synthetic a priori propositions. Examples of such propositions are, “Every event is caused,” “The shortest distance between two points is a straight line.” “For every action there is an equal and opposite reaction,” and “Something endures through every change.” These propositions are necessarily true, and true of any time or place in the universe. What is more, their truth is not derived from experience; in fact, they are used to judge experience. For instance, I know that the straight distance between two given points is ten feet, but a line drawn from one point to the other is measured to be ten feet and 1/16 inch. I may then conclude that no matter how straight the drawn line may look, it is not straight.
Practical propositions—i.e., imperatives expressing what a person ought to do—are likewise distinguished as analytic and synthetic, a priori and a posteriori. We have already seen that hypothetical imperatives are analytic: they command under necessity and are thus a priori. The force of the command to choose the means follows from the practical meaning of the wills choice of the end. But moral commands, since they do not command an action in order to achieve some end, command an action as necessary (a duty) for its own sake. Such practical propositions are a priori but are not analytic: they are synthetic a priori practical propositions. Reason relates an action to the idea of a good will and duty, but the meaning of good will and duty do not include any reference to a particular action. How then can reason command a particular action as unconditionally necessary? Or, in Kant’s words, how is a synthetic a priori practical proposition possible? How does reason give moral law? The difficulties of the Critique of Pure Reason, in which Kant sought the basis of synthetic a priori theoretical propositions, seem equally great for moral propositions.
In order to give a rational foundation for obligation, to justify the possibility of the categorical imperative, we must first determine whether we ever make such judgments of practical reason. It is certainly easier to validate a known proposition than some abstract idea which is never expressed. We have a fairly clear understanding of what the categorical imperative must be:
a. It must express an unconditional command of reason.
b. It must be independent of desire or inclination.
c. It must be applicable to every moral situation, not limited to this person or that lifetime.
Kant proposes to combine these characteristics in a single formulation, which will then be the supreme principle of morality.
To solve this problem, let us first ask whether from the very idea of a categorical imperative we may derive a formulation which contains all that a categorical imperative must be. Even if we find this formulation, there still remains the special and arduous task of finding how such an absolute law can be possible. We will postpone the latter difficulty until the last section.
The bare idea of a hypothetical imperative tells me nothing until I know under what conditions the imperative comes into being. But when I think of the categorical imperative, I know immediately what it contains. Besides the specific law applying to the situation, the categorical imperative commands only the necessity that my maxim* conform to that law. But since the law is not subject to any conditions, the categorical imperative can command only that my maxim conform to the universality of law as such. The conformity itself is what the imperative commands as necessary.
* A maxim is a subjective rule of acting and must be distinguished from the objective rule, the practical law. The former is a personal policy of acting which reason determines in view of subjective conditions, such as desires or even ignorance. The maxim is the rule according to which the agent does in fact choose to act. The law, on the other hand, is an objective rule—an imperative—valid for every rational being; it is the rule by which the agent ought to act.
Consider the idea of a hypothetical imperative once again. What is the command of reason here? We do not know until we know what end has been chosen. Reason cannot command me to use milk base until I choose to make ice cream. That is why hypothetical imperatives are not commands at all in the strict sense but are what we might call rules of practical consistency. Surely I do not disobey any command if I try to make ice cream with cement and turpentine, unless it be the law of practical consistency. The most that could be said is that I act foolishly, unreasonably, or such like. Although my action cannot be justified as productive of ice cream, it is not an immoral action.
But examine the categorical imperative. In it we see a command in the strict sense. When I disobey it I am immoral, for my maxim—that policy or rule which is my actual and personal basis of choice—does not accord with moral law. The law says, “Do this,” while my maxim says, “In this kind of situation I will do something else.” The moral law is the objective principle, which is not restricted to me personally but applies unconditionally to any rational being in a comparable situation. What happens when I act morally? Simply, my maxim, my subjective rule, coincides with the objective command. When we remember that every action is done according to the agent’s own personal maxim and that he acts morally when his personal maxim coincides with the objective principle, then the general law of morality must be: make your maxim coincide with the objective principle. Or, always make sure that the action derived from your maxim is the same action which the law commands. Or more simply, choose to do what you ought to do.
This is all well and good: I must see that my maxim coincides with the objective principle. But what is the objective principle? Particular moral rules apply to particular situations. “Keep your promise” applies only when I have made a promise. But the categorical imperative, as the imperative of morality itself, must apply to every situation. There must then be something common to every situation that makes it a potentially moral one. This common characteristic is that in every situation there is some moral rule, some objective principle, which applies to that situation, which makes the situation a moral one, and which expresses what I ought to do (or what I ought not to do) in that situation. If applicability of some moral rule constitutes the common characteristic, then the categorical imperative as the supreme rule of morality must express a characteristic common to all moral rules, thereby granting to each particular moral rule its moral authority. According to Kant there is but one such common characteristic: every moral rule commands universally. Every moral rule is a law, for a law is a rule which commands universally. Every particular moral rule contains two elements: the action specified and the universality of application. This universality, the common element in every law, means that there are no exceptions to the command. Everyone in the same kind of situation must obey the law governing it. Everyone who makes a promise must keep it.
But there must be an infinite variety of situations, and it is logically possible that a different moral rule covers each situation. Must we know every moral rule there is? How can I know that my maxim coincides with the moral rule? There is only one sure way, says Kant. Only one law can consistently apply in any given situation, so if my maxim could be thought of as the law for that situation, without conflicting with any other law or norm of consistency, then my maxim conforms to the law for that situation.
There is, therefore, but one categorical imperative, which is:
act only on that maxim which you can at the same time will to become a universal law.
If I can consistently propose that my maxim be the maxim for everyone in the same situation, then my maxim conforms to the objective principle governing the situation. In so acting, I act morally. On the other hand, if my maxim proposes an exception to what is applicable to everyone, or if what I propose conflicts with another recognized moral maxim, then my maxim does not coincide with the moral rule applying to the situation. In so acting, I act immorally.
This then is the meaning of “acting morally": I act morally when I choose in such a way that my own principle or policy is one which at the same time applies to everyone in the same situation. If I can consistently (and honestly) will that everyone ought to do what I do, then my maxim must conform to the objective principle applicable to everyone. I thereby act as though my maxim were the objective law for everyone; I will my maxim to be the universal law.
Kant explicitly states that the objective principle is not itself a rule of action: it is rather a practical definition of what constitutes moral action.
Now if we can derive all imperatives of duty from this single imperative as their ultimate source of authority, we can at least explain what we mean by the idea of duty, even though we be unable (at this point) to decide whether what we mean by duty is anything more than an empty idea.
But we are understandably puzzled by his suggestion that all imperatives of duty (i.e., particular moral rules) can be derived from this single imperative, “as their ultimate source of authority.” Does he mean that from this principle alone we can deduce our duty in any given human situation? Or does he mean that we can determine whether our maxim coincides with universal, objective law? Apparently the latter is closer to Kant’s point of view. His examples (discussed below in Chapter 7) will present the situation of a man who already has a tentative maxim or who already contemplates a certain action. In them the categorical imperative will function as a standard for judging the moral status of the maxim. Only in this indirect way does the command of reason prescribe any particular activity.
What is the importance of the categorical imperative to the practical activity of morality? For Kant it is the validating form of moral action. Just as there is a form of the syllogism which “validates” certain arguments, so there is a “moralizing” form for human actions which “moralizes” the action. If the action is in the proper moral form, it is moral. This form of morality, as we learned in Section I, is universality. So the validating form of moral action is the categorical command of reason: Act only by that maxim which you can at the same time will to become a universal maxim, i.e., a law. Furthermore, when we understand the proper forms of the syllogism, we understand thereby what validity means. Likewise, we understand what morality means when we see how actions become moral by conforming to the objective principle of morality.
But understanding the meaning of a concept does not prove that there are actual instances to which it refers. Kant has stated more than once that the concept of duty might be empty; there may be no truly moral actions at all. Even more critical is the question whether moral action as such is even possible. The theory of determinism states that free human choice is but a delusion, since so-called free choices are but foregone conclusions, effects in a long causal chain. Kant has not yet argued for freedom of choice, nor against determinism. (We will have to wait for his examination of the real possibility of moral action until Section III. )
However, certain questions about the categorical imperative must be answered before we proceed to its justification. For example, how does Kant propose that we derive all our duties from one categorical imperative? Are there perhaps other formulations of this law of morality? How can we relate the idea of universalizing one’s maxim to the idea of acting from duty? To answer such questions, Kant diverges from his main exposition. The explanation of the categorical imperative will be in two parts : (1) an ethical discussion of ways in which the categorical imperative may be used in moral situations (Chapter 7); and (2) an examination of the necessary foundation of the categorical imperative, particularly, the self-legislating will (Chapter 8). The next chapter, then, will consist of a discussion of the three variations of the “one” categorical imperative and the famous Four Examples. This part of Section II may be seen as a preview of Kant’s later Metaphysics of Morals.
Kant has taken reason in its practical activity and derived from its function the same foundation for morality which he discovered in Section I. There he began with common rational knowledge about morality and discovered the underlying principle. Here he has taken practical reason as the source of action, and found in reason itself the standard for all moral action.
He first pointed out that every action called rational is one which is done from some principle or policy. Only a rational being can act from the idea of law, i.e., according to principle. Reason determines the principles; will is the capacity to act on them. Were we completely rational, every action would fully accord with reason; but since we are not, some actions do not conform to the principles recognized by reason.
When reason presents an action to the will as objectively necessary, but the will is drawn to something else, there arises a tension between reason and will called constraint. This constraint is expressed by an imperative, a command of reason, the chief feature of which is the “ought.” “You ought to keep your promise” expresses the relationship between reason’s command and a will which is not bound by natural necessity to follow reason.
How does reason recognize a certain action to be objectively necessary? Since there are two kinds of practical necessity, means-end necessity and objective (unconditional) necessity, reason presents these to the will in respectively distinct commands: a hypothetical imperative and a categorical imperative. The hypothetical imperative expresses only the necessity involved in willing the required means to some end. The necessity is analytic: the concept of “willing the end” contains the concept “willing the means.” The necessity of a hypothetical imperative is one of practical consistency, not moral obligation.
The categorical imperative, on the other hand, commands absolutely, unconditionally. The necessity is one of moral obligation, not of means-end. It does not command an action as a means to something else but as something to be done for its own sake. Since it commands with necessity, it is an a priori command; but as its necessity is not mere practical consistency, it is nonanalytic; i.e., it is synthetic. The categorical imperative, then, is a synthetic a priori command of practical reason.
Being unconditional, the imperative of morality applies equally to all moral situations. One acts morally when he makes his own personal principle, his maxim of acting, coincide with the objective principle, the moral law. Now the common feature of every moral situation is the universality of the rule applicable to that situation. Thus one will act morally whenever his maxim coincides with the universality of the rule applicable to everyone in that situation. And since in any moral situation there is but one applicable rule, an agent is assured that his maxim coincides with the moral rule of that situation when he universalizes his maxim so that it can consistently apply to everyone in that situation.
Moral action can thus be defined as action which accords with the universally applicable rule. Only one imperative expresses this universal rule of morality: act according to a maxim which could consistently be made the maxim for everyone. Thus the categorical imperative: Act only on that maxim which you can at the same time will to become a universal law.