[Reprinted, with some omissions, from MIND, 1888, Vol. XIII, No. 51.]
My aim is to show that, in different parts of Kant's exposition of his doctrine, two essentially different conceptions are expressed by the same word freedom; while yet Kant does not appear to be conscious of any variation in the meaning of the term.
[In the one sense, Freedom = Rationality, so that a man is free in proportion as he acts in accordance with Reason.] I do not in the least object to this use of the term Freedom, on account of its deviation from ordinary usage. On the contrary, I think it has much support in men's natural expression of ordinary moral experience in discourse. In the conflict that is continually going on in all of us, between non-rational impulses and what we recognise as dictates of practical reason, we are in the habit of identifying ourselves with the latter rather than with the former: as Whewell says, ``we speak of Desire, Love, Anger, as mastering us, and of ourselves as controlling them''---we continually call men ``slaves'' of appetite or passion, whereas no one was ever called a slave of reason. If, therefore, the term Freedom bad not already been appropriated by moralists to another meaning---if it were merely a question of taking it from ordinary discourse and stamping it with greater precision for purposes of ethical discussion---I should make no objection to the statement that ``a man is a free agent in proportion as he acts rationally''. But, what English defenders of man's free agency have generally been concerned to maintain, is that ``man has a freedom of choice between good and evil'', which is realised or manifested when he deliberately chooses evil just as much as when he deliberately chooses good; and it is dear that if we say that a man is a free agent in proportion as be acts rationally, we cannot also say, in the same sense of the term, that it is by his free choice that he acts irrationally when he does so act. The notions of Freedom must be admitted to be fundamentally different in the two statements: and though usage might fairly allow the word Freedom to represent either notion, if only one or other of the above-mentioned propositions were affirmed, to use it to represent both, in affirming both propositions, is obviously inconvenient; and it implies a confusion of thought so to use it, without pointing out the difference of meaning.
If this be admitted, the next thing is to show that Kant does use the term in this double way. In arguing this, it will be convenient to have names for what we admit to be two distinct ideas. Accordingly, the kind of freedom which I first mentioned---which a man is said to manifest more in proportion as he acts more under the guidance of reason---shall be referred to as `Good' or `Rational Freedom', and the freedom that is manifested in choosing between good and evil shall be called `Neutral' or `Moral Freedom'.
But before I proceed to the different passages of Kant's exposition in which `Good Freedom' and `Neutral Freedom' respectively occur, it seems desirable to distinguish this latter from a wider notion with which it may possibly be confounded, and which it would be clearly wrong to attribute to Kant. I mean the ``power of acting without a motive'', which Reid and other writers, on what used to be called the Libertarian side, have thought it necessary to claim. ``If a man could not act without a motive'', says Reid, ``he could have no power''---that is, in Reid's meaning, no free agency---``at all''. This conception of Freedom---which I may conveniently distinguish as `Capricious Freedom'---is, as I said, certainly not Kantian: not only does he expressly repudiate it, but nowhere---so far as I know---does he unconsciously introduce it. Indeed it is incompatible with any and every part of his explanation of human volition: the originality and interest of his defence of Neutral Freedom---the power of choice between good and evil---lies in its complete avoidance of Capricious Freedom or the power of acting without a motive in any particular volition.
[This] distinction helps me to understand how [it is that] many intelligent readers have failed to see in Kant's exposition the two Freedoms-Good or Rational Freedom and Neutral or Moral Freedom-which I find in Kant. They have their view fixed on the difference between Rational or Moral Freedom, which Kant maintains, and the Freedom of Caprice, which he undoubtedly repudiates: and are thus led to overlook with him the distinction between the Freedom that we realise or manifest in proportion as we do right, and the Freedom that is realised or manifested equally in choosing either right or wrong. When we have once put completely out of view the Freedom of Caprice, the power of Acting without a motive, or against the strongest motive when the competition is among merely natural or non-rational desires or aversions,-when we have agreed to exclude this, and to concentrate attention on the difference between Good Freedom and Neutral Freedom-I venture to think that no one can avoid seeing each member of this latter antithesis in Kant. It will be easily understood that, as he does not himself distinguish the two conceptions, it is naturally impossible for the most careful reader always to tell which is to be understood; but there are many passages where his argument unmistakably requires the one, and many other passages where it unmistakably requires the other. Speaking broadly, I may say that, wherever Kant has to connect the notion of Freedom with that of Moral Responsibility or moral imputation, he, like all other moralists who have maintained Free Will in this connexion, means (chiefly, but not solely) Neutral Freedom---Freedom exhibited in choosing wrong as much as in choosing right. Indeed, in such passages it is with the Freedom of the wrong-chooser that he is primarily concerned: since it is the wrong-chooser that he especially wishes to prevent from shifting his responsibility on to causes beyond his control. On the other hand, when what he has to prove is the possibility of disinterested obedience to Law as such, without the intervention of sensible impulses, when he seeks to exhibit the independence of Reason in influencing choice, then in many though not all his statements he explicitly identifies Freedom. with this independence of Reason, and thus dearly implies the proposition that a man is free in proportion as he acts rationally.
As an example of the first kind, I will take the passage towards the close of chap. iii. of the ``Analytic of Practical Reason'', where he treats, in its bearing on Moral Responsibility, his peculiar metaphysical doctrine of a double kind of causation in human actions. According to Kant, every such action, regarded as a phenomenon determined in time, must be thought as a necessary result of determining causes in antecedent time---otherwise its existence would be inconceivable---but it may be also regarded in relation to the agent considered as a thing-in-himself, as. the ``noümenon'' of which the action is a phenomenon: and the conception of Freedom may be applied to the agent so considered in relation to his phenomena. For since his existence as a noümenon is not subject to time-conditions, nothing in this noümenal existence comes under the principle of determination by antecedent causes: hence, as Kant says, ``in this his existence nothing is antecedent to the determination of his will, but every action even the whole series of his existence as a sensible being, is in the consciousness of his supersensible existence nothing but the result of his causality as a noümenon''. This is the well-known metaphysical solution of the difficulty of reconciling Free Will with the Universality of physical causation: I am not now concerned to criticise it,---my point is that if we accept this view of Freedom at all, it must obviously be Neutral Freedom: it must express the relation of a noümenon that manifests itself as a scoundrel to a series of bad volitions, in which the moral law is violated, no less than the relation of a noümenon that manifests itself as a saint to good or rational volitions, in which the moral law or categorical imperative is obeyed. And, as I before said, Kant in this passage---being especially concerned to explain the possibility of moral imputation, and justify the judicial sentences of conscience specially takes as his illustrations noümena that exhibit bad phenomena. The question he expressly raises is ``How a man who commits a theft'' can ``be called quite free'' at the moment of committing it? and answers that it is in virtue of his ``transcendental freedom'' that ``the rational being can justly say of every unlawful action that he performs that he could very well have left it undone'', although as phenomenon it is determined by antecedents, and so necessary; ``for it, with all the past which determines it, belongs to the one single phenomenon of his character which he makes for himself, in consequence of which he imputes to himself'' the bad actions that result necessarily from his bad character taken in conjunction with other causes. Hence, however he may account for his error from bad habits which he has allowed to grow on him, whatever art he may use to paint to himself an unlawful act he remembers as something in which he was carried away by the stream of physical necessity, this cannot protect him from self-reproach:---not even if he have shown depravity so early that he may reasonably be thought to have been born in a morally hopeless condition---he will still be rightly judged, and will judge himself ``just as responsible as any other man'': since in relation to his noümenal self his life as a whole, from first to last, is to be regarded as a single phenomenon resulting from an absolutely free choice.
I need not labour this point further; it is evident that the necessities of Kant's metaphysical explanation of moral responsibility make him express with peculiar emphasis and fulness the notion of what I have called Neutral Freedom, a kind of causality manifested in bad and irrational volitions no less than in the good and rational.
On the other hand, it is no less easy to find passages in which the term Freedom seems to me most distinctly to stand for Good or Rational Freedom. Indeed, such passages are, I think, more frequent than those in which the other meaning is plainly required. Thus, he tells us that ``a free will must find its principle of determination in the [moral] `Law''',  and that ``freedom, whose causality can be determined only by the law, consists just in this, that it restricts all inclinations by the condition of obedience to pure law''. Whereas, in the argument previously examined, his whole effort was to prove that the noümenon or supersensible being, of which each volition is a phenomenon, exercises ``free causality'' in unlawful acts, he tells us elsewhere, in the same treatise, that the ``supersensible nature'' of rational beings, who have also a ``sensible nature'', is their ``existence according to laws which are independent of every empirical condition, and therefore belong to the autonomy of pure [practical] reason''. Similarly, in an earlier work, he explains that ``since the conception of causality involves that of laws though freedom is not a property of the will depending on physical laws, yet it is not for that reason lawless; on the contrary, it must be a causality according to immutable laws, but of a peculiar kind; otherwise, a free will would be a chimæra (Unding)''. And this immutable law of the ``free'' or ``autonomous'' will is, as he goes on to say, the fundamental principle of morality, ``so that a free will and a will subject to moral laws are one and the same''.
I have quoted this last phrase, not because it clearly exhibits the notion of Rational Freedom,---on the contrary, it rather shows how easily this notion may be confounded with the other. A will subject to its own moral laws may mean a will that, so far as free, conforms to these laws; but it also may be conceived as capable of freely disobeying these laws---exercising Neutral Freedom. But when Freedom is said to be a ``causality according to immutable laws'' the ambiguity is dispelled; for this evidently cannot mean merely a faculty of laying down laws which may or may not be obeyed; it must mean that the will, quâ free, acts in accordance with these laws;---the human being, doubtless, often acts contrary to them; but then, according to this view, its choice in such actions is determined not ``freely'' but ``mechanically'', by ``physical'' and ``empirical'' springs of action.
If any further argument is necessary to show that Kantian ``Freedom'' must sometimes be understood as Rational or Good Freedom, I may quote one or two of the numerous passages in which Kant, either expressly or by implication, identifies Will and Reason; for this identification obviously excludes the possibility of Will's choosing between Reason and non-rational impulses. Thus in the Grundlegung zur Metaphysik der Sitten, he tells us that ``as Reason is required to deduce actions from laws, Will is nothing but pure practical reason'';  and, similarly, in the Kritik der praktischen Vernunft, he speaks of the ``objective reality of a pure Will or, which is the same thing, a pure practical reason''.  Accordingly, whereas in some passages  the ``autonomy'' which he identifies with ``Freedom'' is spoken of as ``autonomy of will'', in others we are told that the ``moral law expresses nothing else than autonomy of the pure practical reason: that is, Freedom''.
I think that I have now established the verbal ambiguity that I undertook to bring home to Kant's account of Free Will; I have shown that in his exposition this fundamental term oscillates between incompatible meanings. But it may, perhaps, be thought that the defect thus pointed out can be cured by a merely verbal correction: that the substance of Kant's ethical doctrine may still be maintained, and may still be connected with his metaphysical doctrine. It may still be held that Reason dictates that we should at all times act from a maxim that we can will to be a universal law, and that we should do this from pure regard for reason and reason's law, admitting that it is a law which we are free to disobey; and it may still be held that the reality of this moral freedom is to be reconciled with the universality of physical causation by conceiving it as a relation between the agents noümenal self---independent of time-conditions---and his character as manifested in time; the only correction required being to avoid identifying Freedom and Goodness or Rationality as attributes of agents or actions.
I should quite admit that the most important parts both of Kant's doctrine of morality, and of his doctrine of Freedom may be saved:---or I should perhaps rather say that the latter may be left to conduct an unequal struggle with the modern notions of heredity and evolution: at any rate I admit that it is not fundamentally affected by my present argument. But I think that a good deal more will have to go from a corrected edition of Kantism than merely the ``word'' Freedom in certain passages, if the confusion introduced by the ambiguity of this word is to be eliminated in the manner that I have suggested. I think that the whole topic of the ``heteronomy'' of the will, when it yields to empirical or sensible impulses, will have to be abandoned or profoundly modified. And I am afraid that most readers of Kant will feel the loss to be serious; since nothing in Kant's ethical writing is more fascinating than the idea---which he expresses repeatedly in various forms---that a man realises the aim of his true self when he obeys the moral law, whereas, when he wrongly allows his action to be determined by empirical or sensible stimuli, be becomes subject to physical causation, to laws of a brute outer world. But if we dismiss the identification of Freedom and Rationality, and accept definitely and singly Kant's other notion of Freedom as expressing the relation of the human thing-in-itself to its phenomenon, I am afraid that this spirit-stirring appeal to the sentiment of Liberty must be dismissed as idle rhetoric. For the life of the saint must be as much subject---in any particular portion of it---to the necessary laws of physical causation as the life of the scoundrel: and the scoundrel must exhibit and express his characteristic self-hood in his transcendental choice of a bad life, as much as the saint does in his transcendental choice of a good one. If, on the other hand, to avoid this result, we take the other horn of the dilemma, and identify inner freedom with rationality, than a more serious excision will be required. For, along with `Neutral' or `Moral' Freedom, the whole Kantian view of the relation of the noümenon to the empirical character will have to be dropped, and with it must go the whole Kantian method of maintaining moral responsibility and moral imputation: in fact, all that has made Kant's doctrine interesting and impressive to English advocates of Free Will (in the ordinary sense), even when they have not been convinced of its soundness,