20th World Congress of Philosophy Logo

Moral Psychology

Aristotle's Ordinary versus Kant's Revisionist Definition of Virtue as Habit

L. Hughes Cox
Centenary College of Louisiana

bluered.gif (1041 bytes)

ABSTRACT: In what follows I examine the following question: does it make a difference in moral psychology whether one adopts Aristotle's ordinary or Kant's revisionist definition of virtue as habit? Points of commensurability and critical comparison are provided by Kant's attempt to refute Aristotle's definition of virtue as a mean and by the moral problems of ignorance (I don't know what I ought to do) and weakness (I don't do what I know I ought to do). These two problems are essential topics for moral psychology. I show two things. First, Kant's definition is revisionist because he excludes from moral habit-formation what Aristotle includes, that is, (i) practice in prudential calculation of a mean, and (ii) habit-formation by repetition. This follows from Kant's insistence that an act is virtuous only if the moral agent is willing freely and universally. Secondly, Aristotle's virtues modify behavior directly, whereas Kant's virtues modify behavior indirectly by creating moral feeling which, in turn, represses the temptations of the natural inclination. I suggest, thirdly, that as one approximates Kant's ideal of perfect virtue, entailed by the broad duties of beneficence and self-perfection, the difference in kind invented by Kant between virtue and prudence, as a morally neutral rational skill, erodes and becomes a difference in degree. I conclude that Aristotle's ordinary definition of virtue is better able to modify human behavior and solve these two moral problems.

bluered.gif (1041 bytes)

The aim of this essay is to examine the following question. Does it make a difference in moral psychology whether one adopts Aristotle's ordinary or Immanuel Kant's revisionist definition of virtue as a moral habit? Suppose it is objected, at the outset, that these definitions cannot be critically compared because their moral theories are, respectively, aposteriori and apriori, and so incommensurable. Two points of commensurability and grounds for comparative evaluation are two basic problems that any theory in moral psychology must address. They are moral ignorance (I don't know what I ought to do) and weakness (I don't do what I know I ought to do).(1)

In the Nicomachean Ethics (hereafter Ethics), Aristotle maintains that the virtues are formed by repetition as are other habits (see book II, chapters 1-5). "[I]t is by doing just acts that a just man is produced, and by doing temperate acts the temperate man," he explains, and without this kind of habit formation "no one would have even the prospect of being good" (1105b9-12). Further, the "mark" of a good "legislator" and "constitution" is that they: "Make the citizens good by forming habits in them" (1103b4). And in his investigation of the virtue justice, he takes as his "starting point" the ordinary meanings of a "just and an "unjust" man: the latter is "lawless," "grasping," and "unfair"; the former is "law-abiding" and "fair" (V:1129a30-34). In short, Aristotle's intention is to clarify the ordinary meaning of virtue as habit.

In the Metaphysical Principles of Virtue (hereafter Virtue), Kant clearly rejects any concept of moral habit-formation by repetition. He writes:

Skill (habitus) is a faculty of action and a subjective perfection

of choice. But not every such faculty is a free skill (habitus

libertatis)... (66).

If "skill is a habit," that is, a

. . . uniformity of action which by frequent repetition has

become a necessity, then it is not a skill proceeding from

freedom and accordingly not a moral skill (66).

Further, "inner freedom" is a rational self-control which enables one to "subdue one's emotions and to govern one's passions" (67). Essential here is his claim that a virtue is not a "free skill" unless it is a free act of a moral will which "in adopting a rule also declares it to be a universal law" (66-67). Kant's concept of willing (i) freely and (ii) universally leads to his revisionist definition of virtue as moral habit.

The key to (i) is to be found in his insistence on the cognitive certainty of the following metaphysical thesis. Even if the phenomenal self is completely determined causally, the moral self is free because it is noumenal (see the Critique of Practical Reason 28-34, 43-52, 55-59, 100-106; hereafter Practical Reason). He claims that the noumenal self is a cause imminent in "experience" because it is an "efficient cause through Ideas" (50). In short, Kant rejects the formation of moral habits through repetition in order to protect radically the freedom of the moral agent from phenomenal and scientific determinism. The key to (ii) is Kant's rejection of Aristotle's following advice. Since the cognitive results of a kind of reasoning is determined by its subject matter, it is foolish to require of moral reasoning the certainty and precision one can expect of mathematical and scientific reasoning (see Ethics I, chapter 3). Kant does not find this foolish.

In Morals, he eliminates prudence from moral reasoning because the former is not an apriori or pure form of reasoning (1-5). Then he claims apodictic certainty for moral reasoning about maxims which satisfy the universalizability criterion of the categorical imperative (see 9-26). He does concede in Virtue that the virtue of truthfulness requires prudential calculation of a mean between "frankness" (telling too much) and "reserve" (telling too little) of the "whole truth" (95, ftnt. 10) But he insists that there is a difference in kind here; error in prudential calculation is a "fault" but not a "vice." And in Morals, he contends that the prescriptions of physician and poisoner to achieve life or death "are of equal value so far as each serves to bring about its purpose perfectly" (25). The difference is whether the will is good or evil. Hence, there is a difference in kind between practical reasoning --i.e., "rules of skill, councils of prudence--and the "commands of morality" (26). That is, prudential calculation is morally neutral even when it serves moral duty and virtue. In short, Kant provides a revisionist definition of moral habit in order to exclude from its formation both (i) habit-formation by repetition and (ii) practice in prudential calculation of a mean.

Operant conditioning, e.g., Pavlov's famous dog experiment, would seem to be a good example of Kant's definition of a habit that is a uniformity that has become a necessity by repetition. A conditioned reflex may provide a solution of sorts to the problem of moral weakness, but it is not a good solution for the problem of moral ignorance. A moral habit must also be a rational skill. Even similar situations vary enough so that a moral principle cannot be applied automatically or mechanically to all situations in exactly the same way. Now Aristotle defines virtue not only as a moral habit but also as a rational skill of determining a mean between extremes, a mean relative to particular persons (see Ethics II:1106b28-1107a25). He maintains further that "it is not possible to be good in the strict sense without practical wisdom, nor practically wise without moral virtue" (see VI:1144b30-31). The virtues are part of the good which distinguishes practical wisdom from "mere smartness" (see 1144a26-35), and without practical wisdom the virtues cannot be rational skills (see 1144b1-29). One more point. Aristotle discusses extensively the difference between voluntary acts which are morally accountable and involuntary ones which are not (see III, chapters 1-5). In short, Kant's criticism may apply to habit defined as a conditioned reflex, but it does not apply to Aristotle's ordinary definition of virtue as moral habit.

On the other hand, Aristotle insists that it makes "all the difference" whether good moral habits are formed "from our very youth"; virtuous habits modify pleasure, pain, and the natural inclinations governing our behavior (Ethics II:1103b24-25; see chapter 3). Since the young are strongly motivated by these, and not by rational argument, good government will use legal punishment and reward and restrictions in personal freedom to form virtuous habits (see X, chapter 9). Further, the virtue of temperance is a necessary condition for "preserving" the kind of "judgment" involved in practical wisdom (see VI:1140b8-14). Practical wisdom can't be morally neutral because "pleasant and painful objects destroy or pervert" moral judgments about what ought to be done (1140b14). A person "ruined by pleasure and pain" cannot "see" what she ought to choose "for vice is destructive" of right choice (see 1140b16-19). Indeed, practical wisdom is functioning well only if "truth is in agreement with right desire" (1139a30) because "choice" is "either desiderative reason or ratiocinative desire" (1139a31-1139b5) This choice is the "originating" and "efficient" cause of moral action because "choice is desire and reasoning with an end in view," and the "three things" controlling moral "truth and action" are "sensation, reason, desire" (1139a32, 1139a17-18). For Aristotle, the virtuous person enjoys the limitations of prudence and temperance on pleasure, the continent does not, and the incontinent and self-indulgent is pained by these limitations (see I:1099a17-25, II:1104b3-1105a13, and X:1172a21-26). In short, Kant reduces prudence to the mere "smartness" or "cleverness" of Aristotle; his prudence is not Aristotle's practical wisdom.

Aristotle makes a strong case for the thesis that his virtues modify Kant's "natural inclinations" so that they reinforce rather than impede the conformity of behavior to rational moral principle and the determination of the virtuous mean relative to a particular person. In Morals, Kant admits that apriori moral judgment requires aposteriori experience to (i) provide a causal connection between the noumenal moral subject and its phenomenal behavior, and (ii) sharpen the "power" of moral judgment by applying it to examples (3). Nevertheless, he claims that the "pure thought of duty and of the moral law generally" has a more powerful "influence on the human heart" than any empirical "incentives" because moral "reason in consciousness of its dignity despises such incentives and is able gradually to become their master" (Morals 22). And so we come to a pivotal question: Are Kant's virtues more or less effective than Aristotle's in modifying human behavior?

In Virtue, he agrees with Aristotle that the virtues must be "acquired" because they are not "innate," and a person cannot "will" virtuous behavior until one has gained the "power" to do so through trial and exercise (145). But unlike Aristotle, he insists that this is an exercise in "pure practical reason insofar as the latter, in consciousness of its superiority (through freedom) gains mastery over the inclinations" (145). He describes moral habit-formation as an exercise in "ethical ascetics." The aim of such "ethical gymnastics" is to "master" the natural inclinations to the extent necessary to overcome their "danger" to moral duty (154). According to Kant, there is a difference between the external constraints of reward and punishment and the internal self-constraint of the "incentive" to moral duty for its own sake (see 36-56). Virtue, the "moral power of this self-constraint," is the established disposition" of "respect for [moral] law" which issues in "virtuous action" (53). The moral strength of a virtue is achieved by "exercise" as well as by contemplation of the "dignity of the pure law of reason within us" (56-57). "Virtue necessarily presupposes apathy (considered as strength)" in which moral feeling is no longer influenced by "pathological" feeling because "respect for the law is more powerful than all of these feeling together" (68). Kant concludes: "The true strength of virtue is the mind at rest, with a deliberate and firm resolution to bring its law into practice" (68). In short, training in moral virtue is an exercise in moral apathy which constrains the natural inclinations by repressing (suppressing?) them.

This conclusion is confirmed by Kant's explanation in Practical Reason that the virtues are purely apriori; they do not directly modify phenomenal behavior. A "drive" is a "subjective determining ground of a will," and it need not conform to the moral law; a moral drive, i.e., moral "feeling," is apriori and "unmixed" because it must conform solely to the moral law (75). Negatively, this feeling "humiliates" the "self-love" ("selfishness") and "self-satisfaction" ("self-conceit") of "self-regard" (76-78). Their common assumption is that the supreme good is merely happiness, the sum of all of the pleasures of the natural inclinations, and so happiness is the sole justification for all human maxims. This assumption is falsified by moral feeling which is "respect" for the moral law itself. Moral feeling acts positively but still indirectly by weakening the "hindering influence of the inclinations through humiliating self-conceit"; this respect is the subjective "drive to obey the law" and the subjective "ground of maxims" for a whole "life conformable to the law" (79-86).

The above conclusion is confirmed further by Kant's description of "exercise" in pure virtue found in the section on "Methodology of Pure Practical Reason" in Practical Reason (see 157-168). The "first step" in moral education, especially of the young, is to develop the habit of rational moral judgment as a perfection of the noumenal (and so free) self until that habit becomes "second nature" (165). The "first point" in this exercise is to "sharpen" moral "judgments" by determining which particular duty or duties are commanded in a specific action (165-166). The "second point" is to determine whether an action has moral worth because it is done virtuously and not just in conformity to the moral law (166). The "second exercise" uses "vivid" examples of "moral dispositions" (i.e., virtues) to call to "notice the purity of the will" (167). This also brings to the "pupil's attention" the "contentment" which attends the "freedom" from sensuous inclinations and their inevitable discontents (167-168). In short, Kant does exclude from moral habit-formation all practice in prudential calculation of a mean between extremes.

He clearly claims that his theory of virtue does a better job than Aristotle's in solving the moral problems of ignorance and weakness. Appealing to the single and certain principle of his apriori morality is, to repeat, superior to an aposteriori morality (like Aristotle's) that is based on a mixture of many and uncertain, empirical counsels of prudence. Hence, the latter causes the "mind [to] waver between motives that cannot be brought under any principle," and it arrives at the good and not the bad only by "accident" (Morals 22). So Kant challenges the practical effectiveness of virtue if it is defined as a "habit acquired by long practice of morally good actions" (Virtue 41). If virtuous habits result not from "resolute and firm principles ever more purified," they are "neither armed for all eventualities nor adequately secured against changes that may be brought about by new allurements" (41). But is he correct in his claim?

Which is the better way to modify human behavior morally: Aristotle's direct modification of behavior through repetition and practice in prudential calculation of a mean? Or Kant's indirect repression of the temptations of the inclinations? A thorough investigation of this question is too tall an order to be encompassed in the time and space remaining. To choose a manageable problem, but a central one, recall Aristotle's identification of the efficient cause of moral action with "ratiocinative desire." Kant identifies this cause with "moral feeling, i.e., "respect" for the moral law, and he relates them both to "moral satisfaction" (see Morals 459-461, Virtue 377-378, and Practical Reason 50, 54-59, 76-86). But since the natural inclinations are humiliated by moral respect, it is a negative limitation on the pleasures of these inclinations; hence, moral feeling cannot be a pleasure (see Practical Reason 20, 92,120, 124-125).

On the other hand, Kant says that "virtue is its own reward" not because it is a means to pleasure, but because virtuous activity gives a moral satisfaction appropriate to doing one's duty (Virtue 48-50). And he identifies this satisfaction with moral feeling (see 33-35; see also Morals 59-62). Because moral satisfaction cannot be physical pleasure, it is an experience of one's "supersensuous" nature and "higher vocation" of moral duty (Practical Reason 92). Although continuous "pleasantness" is the goal of the natural inclinations, its pursuit is constantly accompanied by an inevitable "discontent" with human existence (124-125). The inclinations are a constant burden to a "rational being" because they "vary," increase with "indulgence," and are insatiable. Hence, the "self-contentment" of virtue is the "indirect" enjoyment of "freedom" from the natural inclinations (120, 125). Furthermore, their "mastery" requires a detachment from their power, and the achievement is something "analogous to the self-sufficiency which can be ascribed only to the Supreme Being" (125). Clearly, this mastery results from ethical gymnastics, that is, exercise in ethical ascetics and moral apathy.

In short, Kant's virtues do not directly modify phenomenal behavior. His virtues directly modify only apriori rational judging, and this produces moral feeling, which in turn constrains the temptations of the natural inclinations by suppressing (repressing?) them. But if psychoanalysis has taught us anything, it is that solving the problem of moral weakness, i.e., conflict between natural inclinations and moral duty, indirectly by repression is less effective than solving it by Aristotle's direct method of behavior modification. Here Kant is closer to Plato than Aristotle.

In the Republic, Plato condemns drama in general and tragedy in particular for undermining the work of philosophy; the former arouse the emotions which the latter must suppress in order to achieve and maintain a self-control which is rational and virtuous and, so, mature and adult (see book X, sections 603-608). On the other hand, in the Poetics, Aristotle praises tragedy for providing a healthy "catharsis" for otherwise destructively repressed emotions (see 1449b20-28). And in the Politics, he maintains that the goal of education is to harmonize the things that make a person "good"; they are "nature, habit, rational principle" (see VIII:1332a39-1332-b10). Since he identifies the efficient cause of moral action as rational desire, the pleasure of virtuous activity reinforces virtue as a habit and rational skill, and this in turn promotes the harmonization of these three components in human goodness. But Aristotle's goal of harmonization cannot be accepted by Kant. He insists that the conflict between moral feeling and the pleasures and pains of the natural inclinations can never be overcome because humans are finite and physical (see Practical Reason 85, 87, 89). And as we have already seen, he identifies moral feeling and satisfaction as the efficient cause of moral action, and they operate causally by suppressing or repressing the natural inclinations.

Where, in short, does Kant fall on Aristotle's spectrum of the virtuous person who enjoys the limits on pleasure, the continent who does not, and the incontinent who finds such limits painful? The evidence shows that his virtuous person is closer to the continent than to the virtuous person. This leads to the conclusion that Aristotle's ordinary definition of virtue as moral habit is more effective than Kant's revisionist definition as a solution to the moral problems of ignorance and weakness. And my argument can be extended further.(2)

bluered.gif (1041 bytes)


(1) Kant creates another point of commensurability when he claims to have refuted Aristotle's definition of virtue as a mean between extremes (see Virtue 62-64, 93-96). This refutation is comprised of what may be called his "good management" and "truthfulness" rebuttals. Both rebuttals fail, but it is the latter that is of particular interest here. He claims that it proves a difference in kind between virtue and a morally neutral, prudential calculation (see 95, ftnt. 10). He concedes that there is a "mean between frankness and reserve" because a person can tell "nothing but the truth" and still not tell the "whole truth." Still he insists that the "moralist" cannot explain exactly what that mean is; frankness and reserve are broad and not strict duties because they permit a "latitude of application." That is, "what is to be done can be determined by the faculty of judgment according to the rules of prudence (pragmatic rules," he insists, " not rules of ethics (moral rules)" (ftnt 10). Kant maintains that his truthfulness rebuttal proves the following claim.

. . . [H]e who obeys the principles of virtue, but in practice

does so to a greater or lesser degree as prudence

prescribes, can commit a fault but not produce a vice . . .

9ftnt. 10).

There are serious problems with this rebuttal too (see, for example, my footnote (2) below).

(2) One way to extend my argument is as follows. Kant's truthfulness rebuttal (see my footnote (1) above) of Aristotle's definition of virtue as a mean proves only that the relation between truthfulness and prudence is a complex one of degree and not a simple one in kind. Kant himself shows that judging exactly what is one's strict duty of truthfulness presupposes and requires as a necessary condition judging what is one's broad duty, and this in turn necessitates a prudential calculation of a mean between "frankness" and "reserve." Furthermore, one cannot determine when they are virtuous or vicious apart from this prudential mean. Finally, this truthfulness rebuttal ends with the "ideal" of "perfect virtue" which "demands" the fulfillment of one's broad duties of beneficence and self-perfection. I would argue generally that as one approximates to Kant's ideal, his categorical distinction between virtue and prudence erodes and breaks down, and his definition of virtue becomes more and more like Aristotle's.

I would argue, further, that the contrary cannot be true. For example, Aristotle maintains that the virtuous person chooses virtuous acts "for their own sake" (II:1105a33). And he defines happiness as excellent activity (e.g., virtue) plus the pleasure that naturally accompanies and completes the activity (see X, chapter 5). Hence, the pleasure that modifies human behavior is a moral pleasure because the quality of a pleasure is determined by the quality of the activity it completes. On the other hand, Kant admits in Virtue that:

The pain that a man feels from remorse of conscience, though

its origin is moral, is nevertheless in its operation physical, like

grief, fear, and every other diseased condition (52).

He claims, in short, that moral feeling causes physical pain. A simple interpretation of this claim is compatible with Aristotle's account of moral pleasure and so pain but not with Kant's. A simple interpretation of this claim is compatible with Aristotle's account of moral pleasure and, so, pain, butt not with Kant's.

In order to move Aristotle's position toward Kant's one must impose on the former the categorical difference that the latter creates in order to reject the former's acco7unt of virtue as habit and rational skill. But if Kant's own extended account does erode the difference he invents, as I have suggested his does, then it is easier--theoretically as well as practically--to make his difference one of degree, and treat Kant's position as an improvement on Aristotle's. For example, in the Morality of Happiness, Julia Annas maintains that Aristotle's definition of happiness is "unstable" (see chapter 18). He does not provide a coherent account of the relation between pleasure, virtue, and external goods which constitutes a complete and self-sufficient life; his definition is too dependent on personal luck to survive the vicissitudes of fate. And she has a point because Aristotle insists that we cannot call a person "happy" except in terms of a "complete life," and then only after that person is dead and completely beyond any chance of misfortune (see Ethics I, chapters 7-12).


Annas, Julia. The Morality of Happiness. New York: Oxford UP, 1993.

Aristotle. Nicomachean Ethics. The Basic Works of Aristotle. trans. W.D. Ross. ed. and introd. Richard McKeon. New York: Random House, 1941.

—— Poetics. The Basic Works of Aristotle. trans. Ingram Bywater. ed. and introd. Richard McKeon. New York: Random House, 1941.

—— Politics. The Basic Works of Aristotle. trans. Benjamin Jowett. ed. and introd. Richard McKeon. New York: Random House, 1941.

Kant, Immanuel. The Critique of Practical Reason. trans. Lewis White Beck. Indianapolis: Hackett Publishing Co., 1983

—— Grounding for the Metaphysics of Morals. Ethical Philosophy. trans. James W. Ellington. introd. Warner A. Wick. Indianapolis: Hackett Publishing Co., 1983.

—— The Metaphysical Principles of Virtue. Ethical Philosophy. trans. James W. Ellington. introd. Warner A. Wick. Indianapolis: Hackett Publishing Co., 1983.

Plato. Republic. The Dialogues of Plato. vol. I. trans Benjamin Jowett. introd. Raphael Demos. New York: Random House, 1937.

bluered.gif (1041 bytes)


Back to the Top

20th World Congress of Philosophy Logo

Paideia logo design by Janet L. Olson.
All Rights Reserved


Back to the WCP Homepage