Immanuel Kant, answered from the texts and cited to the page.
The question answers itself the moment one asks what kind of necessity moral obligation must possess. A consequence is contingent — it may or may not follow from an action, depending on circumstances no agent can fully control. An obligation grounded in consequences is therefore conditional: do this, if you seek that outcome. But a conditional imperative is not a moral command at all; it is a piece of practical advice, a technical rule for attaining a purpose one happens to have.1
The moral law must command unconditionally, or it does not command at all. The key is what I call Autonomie (autonomy) — the will's capacity to give law to itself. Every previous attempt to discover the supreme principle of morality failed, I argued in the Groundwork of the Metaphysics of Morals (1785), because it conceived the human being as subject to a law that came from outside his will — whether from divine command, social utility, or the desire for happiness.2
A law of that kind must carry some interest as its ground, some attraction or constraint by which the will is moved to comply. But then the imperative is always conditional: it commands only insofar as the agent has the relevant interest, and the moment the interest is absent, the obligation dissolves.3 One never arrives at duty on this path, only at the necessity of an action from a certain interest — one's own, or another's — and that is not moral necessity at all.
The principle I set against every such attempt is this: the will of every rational being must be regarded as a will giving universal law through all its maxims.4 Because the will is itself the supreme lawgiver, it cannot depend upon some interest as its ground; a will that depended on interest would require yet another law to limit that interest to the condition of universal validity, and so on without end.5
The categorical imperative is unconditional precisely because it is based on no interest — its ground is the mere form of universal legislation, nothing else.6 This leads naturally to what I called the kingdom of ends — a systematic union of rational beings under common laws, in which each member legislates universally while also being subject to that legislation.7
The Formula of Autonomy and the Formula of the Kingdom of Ends are not two separate principles appended to the Formula of Universal Law; they disclose the same law from different vantage points.8 The Formula of Universal Law tests any given maxim taken singly; the Formula of Autonomy identifies the common source of all such laws in the self-legislative will of every rational being; the Formula of the Kingdom of Ends represents the moral community that self-legislation constitutes.9
In the Metaphysics of Morals (1797), where I gathered the full architectonic of obligation, I stated the matter with the precision the topic demands: a categorical imperative represents an action as objectively necessary not indirectly, through the representation of some end that can be attained by the action, but through the mere representation of the action itself — its form — and hence directly.10
The ground of the possibility of categorical imperatives is simply the freedom of choice, nothing else.11 The consequentialist will object that this makes morality empty — that a form without content cannot tell us what to do. The objection mistakes the character of the formality. The imperative is empty of empirical content precisely so that it can bind every rational being regardless of circumstance, desire, or outcome.
That is its strength, not its deficiency. A moral law that held only when consequences were favorable would be no law at all; it would be prudence wearing a borrowed title.
Morality must be grounded a priori, in pure practical reason, not in any empirical motive — happiness, sympathy, sentiment, or consequence. The supreme principle is the categorical imperative, which commands unconditionally: act only on that maxim by which you can at the same time will it become a universal law; treat humanity always at the same time as an end and never merely as a means; act as a legislating member of a kingdom of ends.12
The three formulations are equivalent. The will that gives itself the moral law is autonomous; the will that takes its law from outside is heteronomous, and heteronomy can never ground unconditional obligation. Moral worth attaches to acting from duty, not merely in conformity with duty: the shopkeeper who is honest because honesty pays, and the philanthropist who helps from sympathetic inclination, have no moral worth, however good their actions appear. Consequences are contingent and cannot ground unconditional moral demands.1314
All other imperatives are technical and are, one and all, conditional.The metaphysics of morals (1797), p. 412
It was seen that the human being is bound to laws by his duty, but it never occurred to them that he is subject only to laws given by himself but still universal and that he is bound only to act in conformity with his own will.Groundwork of The metaphysics of morals (1785), pp. 117–118
in order to conform with the law, his will had instead to be constrained by something else to act in a certain way. By this quite necessary consequence, however, all the labor to find a supreme ground of duty was irretrievably lost. For, one never arrived at duty but instead at the necessity of an action from a certain interest.Groundwork of The metaphysics of morals (1785), pp. 117–118
Thus the principle of every human will as a will giving universal law through all its maxims, provided it is otherwise correct, would be very well suited to be the categorical imperative.Groundwork of The metaphysics of morals (1785), p. 117
a will that is itself the supreme lawgiver cannot possibly, as such, depend upon some interest; for, a will that is dependent in this way would itself need yet another law that would limit the interest of its self-love to the condition of a validity for universal law.Groundwork of The metaphysics of morals (1785), p. 117
just because of the idea of giving universal law it is based on no interest and therefore, among all possible imperatives, can alone be unconditional.Groundwork of The metaphysics of morals (1785), p. 117
By a kingdom I understand a systematic union of various rational beings through common laws... a kingdom of ends, which is possible in accordance with the above principles.Groundwork of The metaphysics of morals (1785), pp. 117–118
FA proposes to identify the common source for such laws (namely, the will of every rational being as autonomous or self-legislative), and it tells us to conform all our maxims to the system of laws that is legislated by such a will.General introduction, pp. 24–25
FUL proposes a test for maxims, for any given maxim taken severally and singly... But FA proposes to identify the common source for such laws.General introduction, pp. 24–25
A categorical (unconditional) imperative is one that represents an action as objectively necessary and makes it necessary not indirectly, through the representation of some end that can be attained by the action, but through the mere representation of this action itself (its form), and hence directly.The metaphysics of morals (1797), p. 412
The ground of the possibility of categorical imperatives is this: that they refer to no other property of choice (by which some purpose can be ascribed to it) than simply to its freedom.The metaphysics of morals (1797), p. 412
Or still better, converting the proposition, if there is a cate gorical imperative (i. e., a law for the will of every rational being), it can only command that everything be done from maxims of one s will regarded as a will which could at the same time will that it should itself give universal laws, for in that case only the practical principle and the imperative which it obeys are unconditional, since they cannot…Critique of Practical Reason and Other Works on the Theory of Ethics, pp. 51–52
But what interests us more here is to know that the prime foundation of morality laid down by all these prin ciples is nothing but heteronomy of the will, and for this reason they must necessarily miss their aim.Critique of Practical Reason and Other Works on the Theory of Ethics, pp. 62–63
These actions also need no recommendation from any subjective taste or sentiment, that they may be looked on with immediate favour and satisfaction: they need no immediate propension or feeling for them; they exhibit the will that performs them as an object of an immediate respect, and nothing but reason is required to impose them on the will; not to flatter it into them, which, in the case of duties, would be a…Critique of Practical Reason and Other Works on the Theory of Ethics, pp. 54–55
Hegel's emptiness charge — the imperative is too formal to yield concrete content — has been the standard internal critique since the 1820s. The utilitarian alternative (Bentham, Mill) treats Kantian deontology as rule-worship that produces monstrous counterexamples. Twentieth- and twenty-first-century Kantians (Korsgaard, O'Neill, Rawls, Habermas) defend the imperative in updated form, generally by anchoring it in the constitutive structure of agency or rational discourse. Bernard Williams's critique of "the morality system" is the most sustained recent challenge from outside the Kantian tradition.