You are on page 1of 30

humanity and evil 635

The Value of Humanity and

Kant’s Conception of Evil
M a tthe w C a s w e l l *

recent years have seen the development of a powerful reinterpretation of Kant’s

basic approach in ethical thought. Kant, it is argued, should not be read as de-
fending the stark, metaphysics-laden formalism for which his theory is so famous.
Rather, the reinterpreters claim that the heart of Kantian practical philosophy
is the absolute value of humanity, or human rational nature.1 Kant’s ethics can
thus be understood as a “theory of value,”2 in which the singular value of our own
end-setting capacity as rational agents is taken as supreme, or even as the source
of all value. On this reading, morality is just acting in such a way that respects or
promotes the value of humanity. Moreover, this value may be deduced through an
immanent, regressive argument about the conditions of practical agency as such,
according to which any adequate conception of ourselves as agents commits us,
finally, to moral norms.
The consequences of this approach to Kantian ethics for such central issues as
the doctrine of transcendental freedom, ethical formalism, the meaning of Kantian
deontology, and indeed the very picture of human moral life for which Kant’s
theory is meant to account are profound. The evaluation of the value of human-
ity interpretation ultimately must examine the challenge posed to the traditional
readings of these doctrines in a comprehensive manner. A contribution to this
evaluation, however, can be made by a consideration of the relation of the concep-

See Christine Korsgaard, Creating the Kingdom of Ends (Cambridge: Cambridge University Press,
1996) and The Sources of Normativity (Cambridge: Cambridge University Press, 1996); Allen Wood,
Kant’s Ethical Thought (Cambridge: Cambridge University Press, 1999); Paul Guyer, Kant on Freedom,
Law, and Happiness (Cambridge: Cambridge University Press, 2000). All three scholars are examined in
the second part of this essay. Other proponents of a value of humanity interpretation include Barbara
Herman, The Practice of Moral Judgment (Cambridge: Harvard University Press, 1993), ch. 10, 208–40;
Alan Donagan, The Theory of Morality (Chicago: University of Chicago Press, 1977), 229. J. David Vel-
leman implicitly subscribes to this kind of interpretation in “Love as a Moral Emotion,” Ethics 109
(1999), 365–66. Not all of the characteristics of the reading mentioned in this paragraph are present
in each of the cited scholars’ works.
Korsgaard, Creating the Kingdom of Ends, x; Guyer, Kant on Freedom, Law, and Happiness, 131;
Herman, The Practice of Moral Judgment, 210.

* Matthew Caswell is Tutor at St. John’s College (Annapolis, MD).

Journal of the History of Philosophy, vol. 44, no. 4 (2006) 635–663

636 journal of the history of philosophy 44:4 october 2006
tion of the value of humanity to a far-reaching but relatively neglected element
of Kant’s theory, namely, his theory of evil as strictly ethically accountable. In the
following analysis, I will show that the assignment of absolute value to humanity
is incompatible with Kant’s conception of evil. Accordingly, it will be shown that
humanity, at least in so far as it is understood in the influential writings of several
recent interpreters, cannot be taken by Kant as absolutely morally valuable.3
In the first part of this paper, we shall see how Kant’s theory of the “original
predisposition” (Anlage), presented as part of Kant’s investigation of evil in Religion
within the Limits of Reason Alone, denies the central claim of the value of human-
ity interpretation (Rel, 6:26–28).4 Next I show why the position defended in the
theory of the original Anlage is required by Kant’s commitment to the imputability
of evil. I will then consider whether the position of the Religion is compatible with
the Groundwork argument so favored by the reinterpreters, in which Kant claims
that the categorical imperative commands us to treat humanity in ourselves and
in others as an end in itself. In the second part of the paper, several leading pro-
ponents of the value of humanity interpretation will be examined and criticized
in the light of Kant’s views uncovered in the first part. I close with brief remarks
on the importance of the conception of evil for Kantian theory.

1. kant’s conception of evil and the normative

independence of humanity and personality
Book One of the Religion contains Kant’s definitive treatment of the issue of evil.
In his investigation, Kant is committed to the basic imputability of evil to an ex-
tent rarely matched in the history of philosophy. For Kant, everything for which
we can be praised or blamed—everything for which we are morally accountable,
and thus everything that is morally good or evil—must necessarily be the product

The following analysis will therefore support Robert Pippen’s claim that “there is no internal
Kantian reason . . . for regarding our capacity to set and pursue material ends as a fundamental value
that must be respected in all our activity” (“Rigorism and the ‘New Kant’” [“The ‘New Kant’”] in
Akten des IX. Internationalen Kant-Kongresses, eds. Gerhardt, Horstmann, Schumacher (Berlin: Walter
de Gruyter, 2001), 315.
Parenthetical references to Kant in the body of the paper are to the volumes and pages of Kants
gesammelte Schriften, herausgegeben von der Deutschen Akademie der Wissenschaften (Berlin: Walter
de Gruyter, 1902). The following abbreviations and translations are used:

Anth Anthropologie in pragmatischer Hinsicht.

Gr Grundlegung zur Metaphysik der Sitten [Groundwork of the Metaphysics of Morals], trans.
H. J. Paton (New York: Harper & Row, 1964).
IaG Idee zu einer allgemeinen Geschichte in weltbürgerlicher Absicht.
KprV Kritik der praktischen Vernunft [Critique of Practical Reason], trans. Lewis White Beck
(Indianapolis: Bobbs-Merrill Company, Inc., 1956).
KU Kritik der Urteilskraft [Critique of Judgment], trans. Werner Pluhar (Indianapolis:
Hackett Publishing Company, 1987).
MA Mutmassliche Anfang des Menschgeschichte.
MdS Metaphysik der Sitten [Metaphysics of Morals], trans. Mary Gregor (Cambridge: Cam-
bridge University Press, 1996).
MV Moral Vigilatius in Lectures on Ethics, trans. Heath (Cambridge: Cambridge University
Press, 1997).
Rel Religion innerhalb der Grenzen der blossen Vernunft [Religion within the Boundaries of Mere
Reason] in Religion and Rational Theology, trans. Di Giovanni (Cambridge: Cambridge
University Press, 1996).
humanity and evil 637
of a free will.5 This means that certain desires, or even our sensuous nature in
general, cannot be the source of evil, for our sensuousness is not a product of our
own freedom, but an imposition of unfree nature. If our will were overpowered
by sensuous influences, such that we were compelled by natural forces to deviate
from the law, we could not be considered blameworth for what results, since our
accountability only extends as far as what our free will itself produces. Another
traditional conception of evil—that it is, or arises, from a kind of ignorance of the
good—is also unavailable to Kant. Consciousness of moral obligation is for Kant
the most basic “fact of reason,” available to the most common understanding. Since
we would never be able to assume our freedom (and hence our accountability)
without consciousness of the moral law, an agent who is both responsible for his
or her evil willing and at the same time ignorant of the obligation not to will in
such a way is an absurdity.
Kant’s argument that the human will is “radically evil”—or that a freely chosen
disposition towards immorality is bound up with the human will at its very root—fol-
lows from his commitment to the strict imputability of evil, together with a certain
conception of human moral experience. Since inclinations cannot become reasons
for violating moral norms, or genuine temptations to immorality, unless they are
freely incorporated into the will as such reasons,6 the existence of temptations (or
incentives that tempt us to violate moral commands) can only be understood as
grounded in a free decision to subordinate moral requirements to our self-love;
a will that finds itself tempted must have already incorporated self-love into its
principle as a reason to disobey the law. Such a positively anti-moral act of the will
is evil; thus, a basically evil orientation can be inferred to belong to any will for
which morality is a struggle against temptation.7 Moreover, it is only for such wills
that virtue as a strength of character becomes a necessary ethical project.8 The
acknowledgement of genuinely imputable evil as a central problematic of the hu-
man moral condition is, for Kant, tightly bound up with his conception of virtue
as struggle, and in that sense belongs to the most fundamental commitments of
the Kantian ethical view.
Rather than further investigate the details of Kant’s complicated and admittedly
controversial deduction of radical evil, I want to focus on one of the key steps in
Kant’s explication in Book One of the Religion of human moral psychology, or his
a priori theory of human moral nature, in which radical evil is to be discovered.
The theory of the “original predisposition” presents a clear statement of Kant’s
position on the value of humanity or rational nature, according to which (contrary
to the value of humanity reinterpretation at issue) mere humanity is denied any

See, for example, Rel, 6:44: “Man himself must make or have made himself into whatever, in a
moral sense, whether good or evil, he is or is to become.”
For “an incentive can determine the will to an action only in so far as the individual has incorporated
it into his maxim” (Rel, 6:24, emphasis in original). This claim has been singled out as one of Kant’s most
significant and far-reaching insights, and identified as the “Incorporation Thesis,” by Henry Allison in
Kant’s Theory of Freedom (Cambridge: Cambridge University Press, 1990), 39–40.
See especially Rel, 6:37–38.
Thus, the doctrine of radical evil is a necessary element of our “moral discipline” (Rel, 6:51).
The reciprocal relation between radical evil and the duty of virtue is a central theme of Book Two of
the Religion. See Rel, 6:57–60.
638 journal of the history of philosophy 44:4 october 2006
intrinsic moral status. This move, we shall see, is crucial for securing the possibility
of strictly imputable evil.

The Theory of the Original Predisposition

In section I of Book One, Kant attributes to the human will an “original predis-
position [Anlage] to good,” which is separable, according to function, into three
“elements” (Rel, 6:26). These are understood as the essential constituents of hu-
man nature, such that we could not be considered a human being if we lacked any
of them. It is in this sense that they are “original, for they are bound up with the
possibility of human nature” (Rel, 6:28). The examination of the three Anlagen
thus constitutes a systematic inventory of the faculties or capacities that inhere
in the human will. The first predisposition is to “animality” and constitutes our
existence in so far as we are merely living beings with inclinations and desires (Rel,
6:26). Secondly, in so far as we are considered rational beings who can determine
our faculty of desire according to concepts in order to achieve satisfaction, we
are endowed with the predisposition to “humanity.” The third predisposition to
“personality” concerns one’s capacity “as a rational and at the same time responsible
being” (ibid.): our capacity for moral action. The two crucial claims about these
essential features which arise in Kant’s discussion are that the higher predisposi-
tions are not entailed by the lower (while the lower seem to be presupposed by
the higher), and that we are not in a position to choose or reject any of them. Our
animality is not something we choose, for inclinations and desires are not them-
selves products of freedom. Nevertheless, it is possible to misuse this predisposition
and direct it away from the good, giving rise to what Kant calls the “beastly vices”
of physical excess and ruin. Note, however, that this misuse is not attributable to
the predisposition itself (Rel, 6:28).
Kant’s discussion of the predisposition to humanity proceeds in much the
same way. The capacity for rational end-setting and reflective comparison is not
something we are free to adopt or not to adopt. This feature of our nature extends
far beyond our animality, and yet presupposes it. For we could form no idea of
happiness, of total satisfaction, through comparison and imagination if we had no
“material” in our faculty of desire which called out to be satisfied. This predisposi-
tion can also be misused, contrary to law, and thereby gives rise to the “vices of
culture” or the “diabolical vices” (Rel, 6:27). These include desires for superiority
and domination which are made possible through our capacity to (through reason)
set ends for ourselves which are not given concrete content by nature, and so can
only be defined by comparison of our status with that of others.
Most of what is important regarding Kant’s distinction between humanity and
the third “predisposition to personality” is presented in a crucial footnote to the
opening of section I, which must be quoted at length.
We cannot consider this predisposition [to personality] as already included in the
concept of the preceding one [to humanity], but must necessarily treat it as a special
predisposition. From the fact that a being has reason it does not at all follow that,
simply by virtue of representing its maxims as suited to universal legislation, this
reason contains a faculty of determining the power of choice unconditionally, and
hence to be practical on its own; at least, not so far as we can see. The most rational
humanity and evil 639
being of this world might still need certain incentives, coming to him from the
objects of inclination, to determine his power of choice. He or she might apply the
most rational reflection to these objects––about what concerns their greatest sum as
well as the means for attaining the goal determined through them––without thereby
even suspecting the possibility of such a thing as the absolutely imperative moral law
which announces to be itself an incentive, and, indeed, the highest incentive. Were
the law not given to us from within, no amount of subtle reasoning on our part would
produce it or win our power of choice over to it. Yet this law is the only law that makes
us conscious of the independence of our power of choice from determination by all
other incentives (of our freedom) and thereby also of the accountability of all our
actions. (Rel, 6:26 n.)

Kant’s argument here is of apiece with his theory of the fact of reason. Given the
background premise that our freedom in the strong, transcendental sense is re-
vealed to us through our obligation to categorical moral principle—and indeed
that it is only on such grounds that we may infer our freedom—Kant holds that
consciousness of such moral obligation itself must be taken as an inscrutable, non-
deducible “fact of reason.” The theory of the fact of reason closes off all possible
deductions of the moral law from non-moral starting points, such that a deduction
could only succeed by sacrificing the unconditional character of the morality it was
trying to deduce (KprV, 5:42–43, 91).9 Recalling the more extensive discussion in
the second Critique, Kant presents a succinct summary of the crucial background
premise in a footnote later in Book I of the Religion, claiming that “we can quickly
be convinced that the concept of the freedom of the power of choice [Willkür]
does not precede in us the consciousness of the moral law but is only inferred from
the determinability of our power of choice through this law as unconditional com-
mand” (Rel, 6:49 n.). In the theory of the original predisposition, Kant cashes out
the fact of reason as requiring a basic distinction between humanity (or practical
reason in general) and personality (or pure practical reason). That is, the theory
of the fact of reason entails such a distinction.10 Thus, while humanity “is rooted
in a reason which is indeed practical, but only as subservient to other incentives,”
personality “alone is rooted in reason practical of itself, i.e., in reason legislating
unconditionally” (Rel, 6:28), and no amount of examination of the concept of
the former will uncover the concept of the latter. Our moral capacity must be
taken as “special” in the sense that it could not arise from any “subtle reasoning”
(herausklügeln) upon the conditions inherent in forming maxims and setting ends,
but can be established only by the inscrutable “fact” of our moral legislation. Or
as Kant puts it in the second Critique, it is precisely in so far as one cannot “ferret
out” (herausvernünfteln) consciousness of the moral law from features of reason
external to morality that it can be called “a fact of reason” (KprV, 5:31).

For Allison’s discussion of the theory of the fact of reason, see especially Kant’s Theory of Freedom,
Kain, on the other hand, finds the humanity/personality distinction in the Religion in tension
with the theory of the fact of reason, since the predisposition to humanity “seems to presuppose
spontaneity,” even though Kant holds that “genuine spontaneity” can only be grounded in the “fact of
reason, the consciousness of the moral law” (Patrick Kain, “Prudential Reason in Kant’s Anthropology,”
in Essays on Kant’s Anthropology, eds. Brian Jacobs and Patrick Kain [Cambridge: Cambridge University
Press, 2003], 258.) This contradiction disappears if the spontaneity required for prudence (humanity)
falls short of what Kain calls “genuine spontaneity”—that is, transcendental freedom.
640 journal of the history of philosophy 44:4 october 2006
Note here a fundamental difference between the ways in which the first two
Anlagen are “given,” compared with the third. Kant claims that we are not free to
adopt or reject any of the three Anlagen; we could never decide to no longer feel
the pull of inclination, to cease to be rational end-setters, or to be free from moral
obligation. The first two features of our nature are “given” in the familiar sense
of being imposed by nature: nature gave us a faculty of desire, and nature gave us
the capacity to determine this faculty by concepts. Animality and humanity are, in
this sense, ‘facts of nature.’ But when Kant writes of the moral law being “given to
us from within” through the third Anlage, he must be understood as claiming that
we give ourselves the law (Rel, 6:26 n.). Our autonomy is not an object of choice,
and yet it is also not an imposition of nature like the first two Anlagen. It is rather
freedom itself: the freedom of our will as pure practical reason with which we
self-legislate the moral law. This is why we are bound by the third Anlage in the
unconditional, moral sense.11 For Kant, neither animality nor humanity involves self-
legislation and therefore are not sources of moral value. This consideration allows
us to compare the kinds of normativity at work among the three predispositions.
Animality is a “normless” domain; through our first predisposition, particular,
empirically given instincts and desires give rise to actions without the intermedia-
tion of concepts or principles. Humanity is the domain of prudential norms: our
selection of subjective practical principles is guided by hypothetical imperatives,
which, while rational, are always only conditional. Personality is the domain of
moral or unconditionally obligatory normativity, whose supreme principle is the
categorical imperative.
While it does not explicitly address the sort of value that attaches to our different
practical faculties, Kant’s theory of the original Anlage has decisive implications for
it. Since it is only in virtue of the predisposition to personality that we are moral
agents, it is only in so far as we are persons that we have a conception of the moral
law, along with its correlated conception of unconditional value or dignity. As mere
humans, we would have no access to a conception of absolute value; thus mere
humanity is not a ground of moral or unconditional value. It is clear that according
to the theory of the Anlage, there would be no duty to treat a mere human (who
was not at the same time a person) as an end in himself, anymore than there is a
duty to treat a mere animal as an end in itself.12 While this last point may give us
pause, we must keep in mind that the notion of a merely “human” being reflects
only a conceptual possibility for Kant. Kant uses the term ‘human’ in these pas-
sages not as the name of a species, or even to refer to an existing class of beings
at all. ‘Humanity’ here has a technical meaning within Kant’s moral psychology,

Thus, Pippen makes the valuable point that “there is no way Kant could count ‘being rational
end-setters’ [our humanity] as providing reasons to act without contradicting his most important claim
in moral theory: we are bound only by what we bind ourselves to” (Pippin, “The ‘New Kant’,” 325).
This does not exclude the possibility that we might have duties regarding the treatment of such
hypothetical merely human beings, just as we have regarding the treatment of animals. In neither case,
however, would these duties arise from the inherent dignity of the being in question, since in Kant’s
view neither mere animals nor mere humans possess such dignity. Moreover, we may have duties to
treat the capacities of humanity and animality in ourselves (“in our own person,” in the language of
the Groundwork) as possessing value. But in that case the value of the two lower predispositions would
be grounded in the third, and our duties regarding them would not arise from their inherent dignity,
but from a moral command given through the predisposition to personality.
humanity and evil 641
referring to the capacity for rational agency determined by empirical incentives.
Indeed, while speculations about a merely human being follow naturally from the
theory developed here, Kant did not intend the theory of the original predisposi-
tion to be used primarily as a catalogue of possible hypothetical beings. On the
contrary, this is a theory of the elements of “human nature,” and their separation
is an abstraction we perform upon ourselves in thought. Thus, each of the predis-
positions can be seen as involving a qualified idea of humanity. “Animality” is just
the idea of humanity taken “physically and merely mechanically,” or the biological
conception of humanity (Rel, 6:26); “humanity” is this same idea taken as embodied
reason, or the anthropological conception of humanity; and “personality” (as Kant
puts it­­) is just “the idea of humanity in the wholly intellectual sense,” that is, the
moral conception of humanity (Rel, 6:28).13
I have read the theory of the Anlage as claiming that mere humanity is not a
ground of moral value. This may seem to be contradicted by Kant’s assertion that
our original predisposition is “to the good.” This view is announced in the title of
section I, as well as in the following, somewhat cryptic remark:
All of these predispositions in man are not only good in a negative sense (they do
not conflict with the moral law), they are rather also predispositions to the good (they
support [befördern] conformity with the law). (Rel, 6:28)14

Kant is not claiming here that animality and humanity give rise to non-moral goods
(e.g. satisfaction or happiness).15 What makes animality and humanity predisposi-
tions to the good here is their relation to morality, which itself is given through a
separate predisposition; that is, they are non-moral predispositions to the morally
good. Such a claim is only intelligible from the standpoint of personality, as it were,

The theory of the original Anlage appears in a variant form in Anthropology from a Pragmatic
Point of View. Kant there discusses three Anlagen which distinguish the human species “from other
living beings on earth” (Anth, 7:322–23): the “technical,” the “pragmatic,” and the “moral.” The first
consists in our capacity for skillfully pursuing ends as such, the second in our ability to pursue the
overarching end of happiness in a social context, and the third in our status as “subject to a law of
duty.” Both the “technical” and “pragmatic” predispositions cited here would be included under the
Religion’s “predisposition to humanity,” while the “moral,” of course, corresponds to the “predisposition
to personality.” No reference to animality is made here, since Kant is examining those Anlagen which
we do not share with other known beings. While Kant does not explicitly claim in the Anthropology text
that the moral predisposition is not entailed by either of the others, he does not suggest that it is. If
we assume the underivability of higher predispositions from lower, it is of note that Kant might here
allow the possibility of a rational but merely technical being, or one that set ends without engaging in
reflection and comparison of those ends with others in a social context. Compare also the four “steps”
of the historical development of reason in Conjectural Beginnings of Human History. The first three steps
parse humanity or mere rationality into the distinct capabilities of “comparison” (MA, 8:5), “refusal”
(MA, 8:8) and “anticipation” (MA, 8:9), while moral consciousness comes only with the fourth and
“final” step (MA, 8:10).
The translation of this sentence in the Cambridge volume of Kant’s Religion and Rational Theology
incorrectly renders befördern as “demand” rather than “support” or “promote.” This error is perhaps
connected to Wood’s interest in finding grounds of obligation (and thus moral “demands”) within
the predisposition to humanity. See my discussion of Wood below.
Nevertheless, Kant certainly could claim here that the other predispositions (or at least that
to humanity) gave rise to other, non-moral goods. Indeed, Kant’s separation of the predispositions
entails a pluralism of value for human agency, since the subjective, “highest physical good” (KU, 5:450)
of happiness has its source in a capacity of the will conceptually independent from that which gives
rise to the objective moral good of virtue (cf. KprV, 5:60). The Anlagen are also identified as good in
the Anthropology (7:323).
642 journal of the history of philosophy 44:4 october 2006
since moral value is only at issue for persons bound by moral law. Obviously, from
the standpoints of animality or humanity themselves, these capacities are morally
worthless. Kant means that the incentives given through the non-moral capacities
ought to be incorporated into the will through morally legislative freedom, under
moral principles, such that they become part of a morally good, whole character.
The development of virtue thus requires “the reformation of our sensible char-
acter [Sinnesart]” (Rel, 6:47) in which originally non-moral drives given through
our non-moral predispositions, such as our natural sympathy, are recruited as
“moral facilitators,” promoting more consistent, effective virtuous action.16 Such
a project clearly presupposes a freedom oriented by moral principle, and thus the
predisposition to personality.

The Humanity/Personality Distinction and the Imputability of Evil

The normative independence of humanity and personality articulated in the theory
of the Anlage is a crucial thesis for Kant to bring forward in his investigation of evil,
because it is necessarily required by his conception of evil as imputable.17 What we
might call the ‘official’ Kantian reason why the conception of evil requires the hu-
manity/personality distinction is noted by Kant at the close of the footnote quoted
above. Since the predisposition to humanity or mere finite practical reason is not
connected with a purely formal, practical principle (or a categorical imperative), it
could never allow us to infer our transcendental freedom from natural determina-
tion. But then we would not be conscious of our accountability and would accord-
ingly be unable to consider our evil as something strictly imputable to us. In other
words, it would always be open for merely human agents to chalk up their putative
evil to nature’s chain of cause and effect. This position, expressed so clearly in the
Religion, is grounded in Kant’s views that transcendental freedom is a necessary
condition of moral responsibility and that obligation to formal categorical principle
provides the only adequate grounds for inferring our transcendental freedom. It
is not possible here to offer a direct defense of this central Kantian claim (now
known as Kant’s ‘Reciprocity Thesis’).18 However, a closely related reason for why
the imputability of evil requires the humanity/personality distinction arises from
a consideration of the normative structure of the two capacities.
A merely rational agent (a mere human) sets ends through the adoption of
maxims, and does so in accordance with the counsels of hypothetical imperatives.
It is in precisely this, of course, that his or her practical rationality consists. Now, if
the normative independence of humanity and personality did not hold, and the

The term “moral facilitators” is Allison’s (Kant’s Theory of Taste (Cambridge: Cambridge Uni-
versity Press, 2001), 233).
The thesis of normative independence is defended extensively by Kain, although from the
other side of the distinction, as it were. He argues that “prudence [humanity] can be seen to be a
genuine manifestation of rational agency, involving a distinctive sort of normative authority, an au-
thority distinguishable from and conceptually prior to moral norms” (“Prudential Reason in Kant’s
Anthropology,” 232). Where Kain’s extremely helpful account is aimed primarily at saving a Kantian
conception of prudence, the present account can be seemed as using the same thesis regarding the
normative independence of humanity and personality to save the Kantian conception of morality itself,
specifically, its constituent conception of evil.
Allison, Kant’s Theory of Freedom, 201.
humanity and evil 643
predisposition to personality was seen not as “special” but as already entailed by
mere practical rationality, it would be possible to derive moral principles from the
rational norms inherent in the procedure of choosing maxims and setting ends.
In other words, moral norms would just be norms of practical reason as such, suf-
ficiently reflected upon (or herausgeklugelt). But this would mean that the failure
to be moral would be normatively equivalent to the failure to be rational, for evil
would be a form of irrationality. However, the actions attributable to a rational
agent—that is, those for which the agent’s reason is taken as the cause—must
qualify as rational. In other words, we are rational agents precisely in so far as we
act rationally. Indeed, the rationality of our actions is a necessary condition of
their status as products of a free will: a will determined by principles of practical
reason. In order for an evil agent’s acts to count as evil in the genuinely ethical
sense, they must be fully attributable to his or her free, rational agency.
Does this mean that someone could not freely and accountably decide to act
irrationally? An agent conceivably could form the policy to carry on in a contradic-
tory way, exhibiting as much as possible the failure to take the necessary means to
his or her ends. For such a failure would constitute the violation of the “analytic”
practical principle of instrumental reason, which Kant claims every agent follows
“in so far as reason has decisive influence on his actions”–– that is, in so far as he
or she is a rational agent (Gr, 4:417). One might decide to act this way in order
to convince a jury that one was insane, for example. In this case, an agent’s deci-
sion to behave irrationally would be made on the basis of the judgment that such
behavior will contribute to the agent’s own happiness—that is, to staying out of
jail. The selection of the policy of irrationality is itself quite rational—clever, even.
Slightly stranger would be the case of an agent who genuinely preferred to act
irrationally—who somehow enjoyed living a life of contradiction and nonsense,
perhaps because of the rebelliousness such an attitude conveyed. But even here,
irrationality is being counted by the agent as part of his or her own happiness.
When we blame this absurdist rebel for not taking life seriously, we are implicitly
assuming that the policy of irrationality itself is made on rational, probably self-in-
terested grounds. In both cases, the accountability of the irrationality at issue turns
on a deeper, entirely rational (though perhaps immoral) principle of the agent’s
will. Thoroughgoing irrationality—or irrationality that was not itself grounded
in a free (and therefore rational) decision—can be understood neither as a free
choice, nor as a choice for which one is accountable. We typically respond to thor-
oughly irrational agency in ourselves and in others—if it deserves even to be called
agency—not with blame, but with something like concern. The irrational “agent”
is a paradigmatic case of decisively diminished responsibility: someone whose will
has become determined by non-rational forces external to the will, in place of the
free determination by reason. If evil is understood as irrationality, the evil agent
would have to be taken as less free­­, and therefore less accountable for his or her
conduct as the good agent—more fully a free rational actor. This is plainly in con-
flict with the Kantian conception of evil willing, which must be both rational and
free in the strong sense, in order to qualify for genuinely moral condemnation.
Kant’s claim that humanity and personality are normatively independent—that
the norms of morality are neither equivalent to nor derivable from the norms of
644 journal of the history of philosophy 44:4 october 2006
rationality as such—is itself required in order to preserve a conception of evil as
just as rational (and hence just as free) as good action.
Many thinkers take weakness of the will in non-moral or prudential cases as a
paradigm example of imputable irrationality. In such cases, it is argued, we fail
to take the necessary means to our own end—irrationally yet culpably violating
the instrumental principle. On such a view, because irrationality does not void
responsibility, it would be possible to construe evil as a form of just this sort of
(imputable) failure to be rational. Thus, the imputability of evil as such would not
itself require the normative independence of humanity and personality. The very
possibility of prudential akrasia on these terms seems to stand as an objection to the
present interpretation, according to which Kant’s distinction between humanity
and personality (by severing the authority of moral norms from that of the norms
of practical reason as such) is required to preserve the culpability of evil by preserv-
ing its rationality. However, Kant’s theory of freedom (implicit in his articulation
of the original Anlagen) constitutes a deep challenge to this way of framing the
issue of akrasia. On Kant’s view, prudential ends are not given us by nature, but are
ends we give ourselves by freely incorporating them into our maxim (Rel, 6:24).
Accordingly, a putative case of prudential weakness––say, a failure of resolve in
sticking to one’s diet––would need to be reinterpreted as success in carrying out
one’s perfectly resolute (though perhaps only implicit) policy of enjoying fine
pastries, at least in so far as responsibility for the akratic action was assigned to the
agent. Thus, failure to take the necessary means would count as evidence that an
agent was in fact committed to a different end. The indeterminacy of the ideal of
happiness yields rampant conflict among our prudential ends, and how we settle
those conflicts—how we prioritize our ends—is seen in how we choose to act. A
full elaboration and defense of such an interpretation of allegedly akratic action
in non-moral cases is not possible here, though it seems to me at least plausible
on its face, and wholly consonant with Kant’s positions on the analyticity of the
instrumental principle, and on the necessary role of freedom in setting ourselves
ends through the incorporation of desires into our maxims.19
Moreover, such a position need not be defended to counter the above objec-
tion. For even if we granted the possibility of culpable irrationality, it would clearly
mark a profound departure from Kant’s view and from common moral sense to
assimilate all forms of evil to this kind of irrationality. By denying the normative
independence of humanity and personality, the objector could only admit the
possibility of imputably evil acts in so far as they were akratic. Setting to one side
the question of whether or not irrationality voids responsibility, it seems clear that
the ethical character of evil, and the character of our condemnation of it, is grossly
distorted if evil is denied rationality. Is it at all plausible that the scheming heir
who smothers his sick mother with a pillow in order to receive his inheritance is
acting akratically? Is his failure really failure to be rational? Kant’s humanity/per-
sonality distinction saves one from venturing such unlikely interpretations. While

Although he doesn’t seem to find the position particularly plausible, Pippen suggests that Kant
was committed to offering just this sort of interpretation of prudential akrasia in his “Kant’s Theory
of Value: On Allen Wood’s Kant’s Ethical Thought” [“Kant’s Theory of Value”], Inquiry 43 (2000),
humanity and evil 645
rational (though not purely so), the murderous heir is guilty of a moral failure;
that is, he is evil.20
That Kant does not consider evil a form of irrationality is strongly suggested
by the text of the Religion. For Kant, evil’s hallmark is the eminently rational activ-
ity of “self-deception” (Rel, 6:38). He characterizes our self-willed susceptibility
to temptation not as a kind of failure to be fully rational, but as a “secret league”
on the part of our reason with temptations—a rational choice to subordinate
moral demands to their satisfaction (Rel, 6:60). Moreover, understanding evil as
irrationality could be seen as a form of the fundamental error of Kant attributes
to the Stoics: they “mistook their enemy,” in so far as they held that evil was a
kind of “folly,” rather than a “malice (of the human heart)” which “hides behind
reason” (Rel, 6:57). Finally, the rationality of evil is brought to the fore in Kant’s
basic gloss of the theory of radical evil as the claim that “every man has his price,
for which he sells himself” (Rel, 6:38); for here Kant portrays evil in terms of an
economic negotiation and exchange–– that is, in terms of a paradigmatic activity
of non-moral reasoning.
Kant’s position on the rationality of evil may be illuminated by means of a
brief digression, considering the resources Kant’s theory of evil makes available
for responding to the penetrating objection to Kantian ethical thought advanced
by Henry Sidgwick, namely, that Kant’s identification of reason as the source of
moral principle renders the imputation of evil impossible.21 Sidgwick’s argument
may be parsed as follows:
1) Kant holds that rationality is a necessary condition for free agency
2) Because Kant understands moral goodness to be conformity to a principle
of practical reason, he identifies moral worth as rationality (512).
3) But freedom is a necessary condition of moral responsibility (514–15).
4) Therefore, a morally unworthy or evil agent is not morally responsible for
his or her evil (516).
Sidgwick is certainly correct that
the life of the saint must be just as much subject—in a 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. (516)

So the conclusion of the argument suggests that something is seriously faulty with
Kant’s basic moral theory.
Kant’s treatment of moral weakness of the will may be compared with the position I am ascrib-
ing to him regarding prudential akrasia. Weakness in our pursuit of moral ends is counted as a moral
failure, and as such must be grounded in an agent’s freedom, namely, in his “secret league” with
temptations towards immorality, which first renders them genuinely motivational by freely incorporat-
ing them into his maxim. Moral weakness, in other words, is self-imposed. That is why it comes in for
moral condemnation, and is taken by Kant as a species of evil (Rel, 6:29). As is the case for all forms
of evil, moral weakness cannot (on Kant’s view) be understood in terms of a failure to be rational. See
Allison’s discussion of moral weakness in his Kant’s Theory of Freedom 158–60.
Henry Sidgwick, “The Kantian Conception of Free Will” in his The Methods of Ethics, 7th ed. (In-
dianapolis: Hackett Publishing Company, 1981), 511–16. The following references are to this work.
646 journal of the history of philosophy 44:4 october 2006
Sidgwick frames his objection as part of a critique of what he believes is a
Kantian confusion over two distinct senses of freedom: “Rational” or “Good”
freedom (expressed by determination of the will by moral law) and “Neutral”
freedom (expressed in a will’s responsible choice between good and evil) (511).
But we underestimate the objection if we believe it can be eluded by claiming that
Kant was already aware of this distinction, insofar as he drew a roughly equivalent
distinction in terms of the “positive” freedom of autonomy and the “negative”
freedom from natural causation.22 For Kant’s difficulty, according to Sidgwick’s
argument (at least in the form extracted above), is not that he fails to distinguish
between these two senses of freedom, but rather that both are present in his
theory at the same time. In other words, making explicit Sidgwick’s own distinc-
tion doesn’t help Kant, since it does nothing to rebut the argument presented
above, according to which the Kantian cannot impute evil to an agent. What is
needed is rather a critique of Sidgwick’s distinction: namely, we must emphasize
that “positive freedom” entails only that an agent is subject to moral law, not that
he or she acts in accordance with it. That is, for Kant both the good and the evil
agent are positively free (autonomous), as well as negatively free.
It is helpful here to draw attention to a distinction that Kant makes, not between
senses of freedom, but between senses of practical reason. For Kant rejects premise
(2), since he denies the identification of mere rationality with moral worth. While
pure practical reason legislates moral principle, failure to act according to such
principle in no way entails that an agent is not acting in accordance with other,
non-moral (and empirically conditioned) principles of practical reason. This
distinction can be found in several areas of Kant’s ethical doctrine, but it is given
an especially clear presentation in the theory of the original predisposition, in
Kant’s distinction between humanity and personality. Indeed, Kant foregrounds
this theory in his investigation of evil, where his most basic commitment is to
denying the conclusion of Sidgwick’s argument.
Sidgwick’s objection may remain a difficulty for a reading of Kant that denies
the normative independence of humanity and personality. For it is difficult to
see how Kant could reject either premise (1)—that rational agency is a necessary
condition of freedom—or premise (3)—that freedom is a necessary condition
of responsibility. But the denial of the humanity/personality distinction seems to
entail the acceptance of premise (2)—that moral worth can be identified with
rationality. Therefore, we would appear to be compelled either to reject, with
Kant, the conception of the normative dependence of morality upon practical
reason underlying premise (2), or to accept the unsettling conclusion (4) that
we are responsible only for morally worthy action and thereby forsake an account
of evil as culpable.

Humanity as an End in Itself in the Groundwork

With his theory of the original predisposition, Kant makes clear that he excludes
the possibility of a deduction of moral norms from the concept of humanity, or

This is Korsgaard’s response to Sidgwick’s objection in Creating the Kingdom of Ends (172, 162),
which, for the reasons given here, I do not believe is successful.
humanity and evil 647
of end-setting practical reason. This view may appear to diverge from the position
defended in the Groundwork so favored by the value of humanity interpreters:
that is, the famous claim that we are obliged to treat humanity in ourselves and
in others as an end in itself, and never merely as a means, in accordance with
humanity’s “dignity” or unconditional worth. Kant presents this notion in his
second major formulation of the categorical imperative, the so-called ‘formula
of humanity.’ Hypothetical imperatives, Kant argues, must always presuppose an
end or purpose, which would be furthered by conformity to their dictates. But the
value or “worth” (Werth) of these ends is always “relative” and “conditioned”—that
is, relative to and conditioned by the empirical constitution of the subject’s faculty
of desire. In the absence of the relevant desires for a given hypothetical end, its
worth vanishes. Categorical imperatives, on the other hand, cannot function in
this way. They cannot presuppose any conditioned worth of some end, but rather
command that we take some end as possessing an “absolute” or “unconditioned”
worth. Whatever deserves to be valued in this way must be an “end in itself,” and
Kant argues that the only possible candidate for such a status is “human being and
every rational being as such” (Gr, 4:427–28). Is this endorsement of the absolute
value of humanity compatible with the theory of the Anlage? In the Groundwork
passages, Kant makes no reference to the distinction between mere humanity and
personality. Rather, he seems here to conflate them: the all-important distinction
between “things” and “persons” is made between “non-rational beings” and “ratio-
nal beings”, which certainly suggests that all rational beings are persons (Gr, 4:428).
Thus, he claims that “Man is not a thing,” perhaps suggesting that personality is
included in the concept of humanity (Gr, 4:429), and that “violators of the rights
of humanity” are those who “use another’s person as a mere means” (Gr, 4:430).
In these passages Kant also tends to attribute personality to every rational being,
or to mere “rational nature” as such. From the Religion passages, we would expect
Kant to require that rational beings also be morally responsible, if they are to
hold the rights of personhood, leaving open at least the conceptual space for a
non-moral, merely rational nature.
However, in the second part of the Groundwork, it is far from clear that Kant is
either able to derive personhood from or identify personhood with mere human-
ity or the idea of an end-setting practical reason—or even that he intends to do
so. Rather, Kant makes the conditional assumption that a categorical imperative
exists and then proceeds to identify what could qualify as the end in itself corre-
sponding to that imperative. His argument first establishes a connection between
the concept of a categorical imperative (which binds unconditionally) and the
concept of what he refers to here and later as ‘personhood,’ or of a being who
exists as an end in itself.23 This connection is analytic and is thoroughly consistent
with Religion’s account of the original predisposition to personality. The difficulty
comes with the next step in the argument, when Kant asserts that humanity (or
rational nature as such) is the sole viable candidate for personhood, or thing-in-
itself status. Is there a necessary connection between the concept of personality
and that of rational nature as such? If we presuppose that we are in fact persons,

Kant consistently uses of the term ‘person’ and its related forms to refer to full-fledged moral
being throughout the corpus. See KprV, 5:87; MdS, 6:380; MV, 27:627, 543.
648 journal of the history of philosophy 44:4 october 2006
then of course it follows that we ought not treat our rational nature merely as
a means. And in fact this is just Kant’s presupposition in the second part of the
Groundwork: that we are indeed moral persons to whom a categorical imperative
applies. But it does not follow from the fact that we are rational beings that we are
for that reason persons. Kant does assert here that a human agent must take both
his or her own existence (and that of “every other rational being”) as an end in
itself. But he notes at the same time that this is merely a “postulate” or conditional
assumption which will not be grounded until the Groundwork’s third part, where
Kant will mount a full-fledged deduction of our status as transcendentally free,
moral agents (Gr, 4:429).24 That is, without assuming that we are indeed entitled
to take ourselves as “free” in the strong sense—and thus subject to a categorical
imperative—we cannot validly attribute absolute worth or dignity to our human-
ity. Even in the Groundwork, Kant appears to be aware that the capacity for mere
rational agency—what he will later define as ‘humanity’—does not itself amount
either to freedom or to moral status, since he is usually careful to attach the quali-
fications of autonomy or personhood to his references to humanity. That is, in
the Groundwork, the concept of humanity is already moralized; when Kant writes
of the “idea of humanity as an end in itself,” this should be taken as already the
idea of human personhood.25 Therefore, the formula of humanity should be seen

The text of the Groundwork is ambiguous regarding what precisely Kant holds to require a con-
ditionally assumed “postulate.” I have read Kant as claiming that the attribution of end-in-itself status
as such requires the postulation of freedom and moral status. But the text can also be read as claiming
that such a postulate is necessary to move from a “subjective” principle (according to which I take my
own agency as an end in itself) to an “objective” principle (according to which such an attribution of
absolute worth is universalized over all rational agents). However, it is unclear that end-in-itself status
in the fully moral, unconditionally binding sense could be attributed even to the subject of agency
without already presupposing freedom.
Outside of the Groundwork, Kant is usually clear that it is only through our capacity as morally
legislative—as persons—that absolute worth attaches to our humanity. In the second Critique, Kant
writes that “the human being must regard his own nature in reference to his second and highest vocation
[his status as free moral legislator and member of an intelligible world] only with reverence, and its
laws with the deepest respect” (KprV, 5:87).
In the Metaphysics of Morals, Kant makes extensive use of the concept of the unconditional worth
of “humanity in one’s own person.” But he makes clear that this worth is given only through “pure
practical reason” whose legislation makes us “moral beings” (MdS, 6:380). In discussing the duties of
self-development in the Doctrine of Virtue, Kant writes: “the capacity to set oneself an end—any end
whatsoever—is what characterizes humanity (as distinguished from animality). Hence there is also
bound up with the end of humanity in our own person the rational will and hence a duty” to develop
humanity in the sense of mere rational agency (MdS, 6:392). The moral worth of our humanity here
arises only on the condition that said humanity belongs to persons, or full-fledged moral beings. The
Metaphysics of Morals does contain one more troublesome passage from the standpoint of the present
interpretation. Kant writes:

No human being is entirely without moral feeling, for were completely lacking in receptiv-
ity to it he would be morally dead; and if (to speak in medical terms) the moral vital force
could no longer excite this feeling, then humanity would dissolve (by chemical laws, as it
were) into mere animality and be mixed irretrievably with the mass of other natural beings.
(MdS 6:400)

According to my view, Kant should have said “personality would dissolve . . . into mere humanity.”
Nevertheless, Kant may be read as relying on a broader, less technical meaning of “humanity”—as he
tends to in this text—to refer to our whole characteristic as free, morally bound persons, or our already mor-
alized rational agency. (See Anth, 7:277–78). According to the moral psychology of the Religion,
humanity and evil 649
as fundamentally just a formula of personality. It is not here Kant’s claim that the
unconditional value of humanity “is a presupposition of rational choice”,26 or of
value as such. Rather, his claim is that, if our humanity were not taken as an end in
itself—if humanity were not accompanied in us with personality—“nothing would
be found to have absolute worth . . . and thus no supreme practical principle for
reason would be found” (Gr, 4:428). That is, the denial of the absolute worth of
humanity entails the denial of morality, not of practical agency in general.
Kant may have believed that a necessary link could be found between ratio-
nal nature and personality, although he does not explicitly argue for this in the
Groundwork’s second part. After he first nominates humankind and “every rational
being as such” as things in themselves, or persons, he goes on to discuss how relative
ends—which are not ends in themselves—are always conditioned by inclination.
Kant then turns to the exhaustive dualism of reason and inclination: if all condi-
tioned ends are determined by inclination, and if we have a power of determining
ourselves through reason, then we have a power of determining ourselves with
respect not just to conditioned, but also to unconditioned ends. Thus, rationality
entails personality. But if this is Kant’s thinking here, he is guilty of the basic error
of conflating the practical use of reason with its pure practical use. It does not
follow from the fact that we are beings with practical reason that we are capable
of respecting ends set by reason alone. Although our conduct may be determined
by practical reason, this reason itself may in the end be wholly subservient to
inclination. As Kant puts the matter in the Religion, “the most rational being of
this world might still need certain incentives, coming to him from the objects of
inclination, to determine his power of choice” (Rel, 6:26 n.). In this case, while we
would be rational beings, our reason would not be practical of itself—it would not
allow of a pure use.27 Of course, if our reason were always ultimately determined
by inclination—if we could only act on the basis of empirical incentives—then we
could not consider ourselves transcendentally free.28
We noted above that the normative independence of personality and humanity
is a necessary consequence of the view that the fact of reason is the only possible
deduction of freedom and justification of morality. Therefore, perhaps Kant’s
failure to distinguish in a systematic manner between humanity and personality
in the second part of the Groundwork can be explained by the fact that the theory

if we were to lose our moral capacity (personality), we would indeed join the ranks of other natural
beings, morally speaking. And yet according the account given in the Religion, we should be able to
distinguish such unmoralized agents from animals as such, on the basis of the former’s rationality.
Korsgaard, Creating the Kingdom of Ends, ix.
See Pippen’s critique of the value of humanity interpretation, where he points out that even
if end-setting as such requires “in general reasoning,” this does not entail that “reason itself” establishes
an end (“The ‘New Kant’,” 324).
This argument can also be put in terms of the conception of autonomy. While Kant is certainly
correct in claiming that “autonomy is the ground of the dignity of humanity and every other rational
being” (Gr, 4:436), he would be wrong in assuming that practical rationality as such entails autonomy,
for there is nothing internally contradictory or incoherent about the conception of a merely heter-
onomous (and both non-moral and un-free) reason. A merely heteronomous being would indeed
set his or her own ends, but would always require an incentive of inclination—however mediated by
reflection—to do so. That is, such a being’s reason would be practical, and thus legislative in a limited
sense, but not practical of itself, and thus not genuinely self-legislative, or autonomous.
650 journal of the history of philosophy 44:4 october 2006
of the fact of reason would not be developed until the writing of the Critique of
Practical Reason three years later. In the third section of the Groundwork Kant does
indeed address the issue of the deduction of morality, arguing that our status as
reflective, spontaneous minds entails that we have a transcendentally free, intel-
ligible existence (Gr, 4:452). This is of course a deduction of freedom—which can
then be used to deduce moral principle—from non-moral conditions, namely,
those of thought as such. It is not possible here to evaluate whether our freedom
through autonomy over our faculty of desire is indeed entailed by the spontaneity
of our intellect. Nevertheless, it is clear that, in the second Critique, Kant replaces
such a deduction with the theory of the fact of reason. But as long as Kant was
still wedded to a full deduction of morality along the lines of the third part of the
Groundwork, he would understandably tend to underemphasize the conceptual
independence of humanity and personality.29 That is, if Kant argues in the third
part of the Groundwork that mere rationality can be used to deduce the validity
of moral principle (later denied by the theory of the fact of reason), he would
therefore be tempted in the second part to suggest that mere finite rationality
(humanity) suffices to qualify us as moral persons (later denied by the theory of
the Anlage).
If we set aside the attempted deduction of the third section, nothing Kant says
in the Groundwork stands in direct conflict with the view of the Religion. Admittedly,
Kant seems to have a too facile connection between rationality and personality
in mind in his discussion of the formula of humanity, but neither his conception
of that formula itself, nor his broader conception of categorical imperative, com-
mits him to such a position. To the contrary, Kant’s views (clearly presented in the
Groundwork) that morality is a matter of strict obligation to formal principle and
that transcendental freedom must be taken as the condition of the possibility of
such obligation are the cornerstones of his views on evil found in the Religion, and
indeed are thoroughly presupposed by his arguments for them. Nevertheless, if
we take the theory of the fact of reason and the correlated theory of the Anlage
to be the more adequate expression of Kant’s view, then we must part with the
doctrine of the unconditional value of rational nature so often ascribed to Kant.
Given Kant’s theory of the normative independence of humanity and personality,
we must acknowledge that, for Kant, rational nature as such is not absolutely valu-
able, and therefore cannot serve as the ground for a “theory of value.”30 We have
seen that one of the chief casualties of an alternate conception of Kant’s ethical
theory grounded in the alleged absolute value of mere humanity would be the
conception of morally accountable evil.

Kain suggests that the rejection of the Groundwork III deduction of freedom from the nature
of theoretical “judgment” does not exclude the possibility of a deduction from “the recognition of
prudential norms” (“Prudential Reason in Kant’s Anthropology,” 258). Both options, however, are
denied by the theory of the fact of reason.
In the Critique of Judgment, Kant’s denial of the absolute value of mere humanity finds expression
in his distinction between man as the “ultimate end of nature” and man as the “final end of nature”
(KU, 5:429, 434). The former refers to the last or highest member in nature’s teleological hierarchy,
while the latter refers to the end—necessarily outside of nature­­—for the sake of which creation as a
whole exists. Kant identifies human culture, or the development of our merely rational, non-moral
agency, as nature’s ultimate end, while arguing that “in man, but only in him as the being to whom moral
law applies, do we find unconditional legislation . . . which alone qualifies him to be a final end” (KU,
5:435–36, emphasis added).
humanity and evil 651
2. the value of humanity and the conception
of evil in guyer, wood, and korsgaard
I turn now to some of the work of the value of humanity reinterpretation’s three
most influential leading spokespeople: Paul Guyer, Allen Wood, and Christine
Korsgaard. While the disagreements among these authors are extensive, they
nevertheless share a commitment to re-reading Kant’s ethics as grounded by the
unconditional value of humanity, and as we shall see, each of their accounts seri-
ously undermine Kant’s conception of evil as genuinely morally accountable.

For Paul Guyer, “the foundation of Kant’s entire moral philosophy is his belief in
the absolute value of the freedom of rational beings” (155).31 By “freedom,” Guyer
does not refer to our alleged transcendental freedom from natural causation, but
to our ability to choose, set, and pursue our own ends—in other words, our mere
humanity.32 It is in fact the value Kant places on this kind of freedom of choice or
liberty which, according to Guyer, first gives duty (or adherence to moral principle)
its moral worth. That is, it is only because liberty is “preserved, enhanced, and pro-
moted” by conformity to moral law that that law is unconditionally binding. Kant’s
argument must work this way, because otherwise it would leave the “underlying
motivation for taking up moral obligation in the first place” a complete mystery
(143). Indeed, here is precisely where Guyer faults the “traditional” or “classical”
reading of Kant as “profoundly unsatisfying”: without a reason explaining why an
agent would accept moral norms as unconditionally binding “in the first place,”
Kant’s moral theory collapses into mere assertion and question-begging (138).
Guyer proposes that the required basic justification of morality is just that morality
secures and furthers our interest in the supreme value of human freedom.
On Guyer’s reading, the concept of “personality” is derived from the concept of
rational nature in general; thus, our capacity of “humanity” (or mere finite practi-
cal reason as such) suffices to make us persons (192). By eliding the capacities
of personality and humanity, Guyer effectively takes human beings to be nothing
more than human­­; that is, they are not endowed with a “special” moral predisposi-
tion to personality.33 Kant points out that such a being would reason “without even
suspecting the possibility of such a thing as the absolutely imperative moral law
which announces to be itself an incentive, and, indeed, the highest incentive” (Rel,
6:26 n.). Accordingly, human agents in Guyer’s view do not recognize the moral
law as “itself an incentive;” rather, they stand in need of a “supreme value” which
would justify or motivate their adoption of morality in general. Guyer is certainly
correct that merely human agents “would have no reason to adhere to a principle
unless it advanced an end” (146) and that they could not “determine their wills to

Guyer, Kant on Freedom, Law, and Happiness, 155. The following references are to this work.
Transcendental freedom corresponds to what Guyer refers to as “metaphysical” “freedom of
the will” (5, 135, 54–57). On his reading, this notion does not—or at least ought not—play a decisive
role in Kant’s argument.
Guyer does refer to the Religion distinction between humanity and personality, holding that
these terms “connote self-love and rationality respectively” (192 n.16). But this surely cannot be the
case, since Kant is quite clear that the end-setting self-love of humanity already is rational.
652 journal of the history of philosophy 44:4 october 2006
action except by the representation of some particular ends” (225). But according
to Kant, such an inability to determine our will in the absence of determining ends
would render morality impossible.34 It is for this reason that Kant holds that in so
far as we are obligated by moral law, we are justified and indeed required to take
ourselves not merely as human, but at the same time as moral persons.
On Guyer’s reading, Kant was unable to work out an adequate demonstration of
why “freedom of thought and action” should be taken as “intrinsically” valuable.35
This result is not surprising, since, as we have seen, according to Kant no conception
of absolute value is entailed by the capacity for mere humanity. Guyer’s criticisms
of the deduction of moral norms from the concept of humanity—especially as
seen in Korsgaard’s reconstruction of part two of the Groundwork—are well taken
(150–51). However, Guyer wrongly assumes that Kant’s theory requires such a
deduction.36 On the contrary, Kant’s mature conception of the fact of reason and
the correlated theory of the Anlage explicitly deny such a deduction’s possibility. In
so far as it is a free-standing, unconditional constraint, the “value” of morality—the
obligation to which entails our independence from natural determination—can
only be established by a fact of reason, or by a “special” predisposition for the
determination of our will from sheerly moral considerations.37
In Freedom, Law, and Happiness, Guyer offers instructive discussions of much of
the theory of radical evil (223–24, 424). But the compatibility of the Kantian con-
ception of evil with Guyer’s grounding of morality in an alleged absolute value of
humanity is doubtful. Because we are no longer obligated by a purely formal moral
principle, but rather by a material imperative (the only other kind) prescribing how
the supreme end of rational choice ought to be furthered, we have no grounds for
conceiving of our agency as transcendentally free, and thus no grounds for taking
ourselves to be strictly responsible for our moral transgressions. Aside from this
objection, however, it is worth considering what manner of evil Guyer’s reading
could acknowledge as possible. Immorality could arise in two ways, given Guyer’s
understanding of moral principle. First, an agent standing under the supreme com-
mand to further rational choice by acting morally might accept the principle’s end
without taking the necessary means it prescribes. In such a case, the agent would
act irrationally, in the familiar sense of contradicting his or her own intentions.
Alternately, an agent might reject the supreme end altogether. But according to
Guyer, by failing to make rational nature one’s ultimate value, the immoral agent’s
“judgments of value” and correlated conduct would lack a “coherent basis”; one’s
evil would be (in that sense) an expression of irrationality (171). In either case,
Guyer would seem to be committed to the strategy of explaining immorality in
terms of a kind of rational incoherence. Such an explanation of evil—that it is at

That our will must always set some end is a necessary and nearly trivial consequence of its
finitude. It does not follow from this that our will is always determined (and thus motivated) by an end,
at least not if we presuppose that our will is free (KprV, 5:29).
Nevertheless, Guyer deems Kant’s conception of freedom as perhaps “the most coherent basis
for all our judgments of value” (171).
See also Pippen’s critique of Guyer on this point in his “A Mandatory Reading of Kant’s Eth-
ics?,” Philosophical Quarterly 51 (2001): 386–93; esp. 389–90.
Guyer is implicitly demanding an answer to the question, ‘Why be moral?,’ which (as I argue
below) cannot be given from a Kantian perspective.
humanity and evil 653
its root a form of irrationality—is at the same time an excuse, in so far as it must
forsake holding agents strictly accountable for their evil, as we saw above.38 Is the
loss of the ethically rich Kantian conception of evil unavoidable for an interpreta-
tion that grounds morality in the absolute value of humanity? In order to answer
this question, we would do well to turn to the recent work of Allen Wood, a value
of humanity interpreter who devotes significant attention not only to the theory
of evil, but even to Kant’s crucial humanity/personality distinction.

Allen Wood’s discussion of Kant’s ethics shares with Guyer’s the following basic
concern: since “an obligation to follow a rule or principle makes sense only if there
is some value or end that provides a reason for following the rule,” only in the case
that Kant is able to establish such an end will Kantian moral theory be possible
(114).39 On Wood’s reading, this end is humanity or rational nature. He writes:
“perhaps the most fundamental proposition in Kant’s entire ethical theory is that
rational nature is the supreme value and the ground of whatever value anything
else might possess” (121). Rational nature is the capacity of a being in the world
to select its own ends—“to value and to judge” through reflective comparison.
Regarding the ultimate justification for the supreme value of rational nature,
Wood is sympathetic to Korsgaard’s position that Kant’s argument “takes the form
of a ‘regress on conditions’.” In so far as it “confers” value on ends by choosing
them, rational nature “is the source of all such value” and therefore “is regarded
as absolutely and unconditionally valuable, and as an end in itself” (127). A closer
consideration of the Korsgaardian regress will be given in the next section, where
Korsgaard’s own views will be examined.
Unlike Guyer, Wood addresses the issue of Kant’s distinction between humanity
and personality at some length.40 In addressing the thesis defended in the Ground-
work that humanity is an end in itself, Wood poses the question of why—given Kant’s
apparent distinction between humanity and personality—Kant selects the former,
and not the latter, as the end in itself (120). He offers two reasons. First, he points
out that we are bound to respect rational nature “in all of its functions”, not just
in its morally legislative function given through the predisposition to personality.
Second, rational beings ought not to be considered ends in themselves only in so
far as they are virtuous or obedient to moral laws. Thus, it is not a “good will” which
gives a being dignity, but merely “the capacity to set ends from reason, irrespective
of whether its will is good or evil” (121). Neither of these reasons is compelling,
however. To the first point, we should grant that the “whole” of human rational
nature ought to be respected. This does not exclude, however, that our qualifi-
cation for such respect is grounded by one particular feature of human reason,
namely, the predisposition to personality. Note also that at least for worldly beings,
personality presupposes humanity, since the ability to determine one’s maxims

At the very least, it must venture a drastically impoverished conception of evil by assimilating
immorality as such to a form of weakness.
Wood, Kant’s Ethical Thought, 155. The following references are to this work.
Pippen’s critique of Wood’s reading of Kant on this matter is instructive. See his “Kant’s Theory
of Value,” 252–54.
654 journal of the history of philosophy 44:4 october 2006
according to moral principle presupposes that one is capable of setting maxims at
all. Therefore, to respect a person is necessarily to respect a human person.41
The problems with Wood’s second point are especially revealing. Contrary to
Wood’s position, the claim that personality is what gives beings the status as ends
in themselves in no way requires that this status depends on how virtuous they are.
The predisposition to personality refers not to an agent’s virtue, but to an agent’s
capacity for virtue. Wood argues that the dignity-bestowing feature of our agency
is “the capacity to set ends from reason, irrespective of whether its will is good or
evil.” But the crucial point is that a will with the mere “capacity to set ends”—absent
the predisposition to personality—can be neither good nor evil, since it neither is
bound by, nor can act from moral principle. Wood likely takes the predisposition
to personality to refer to actual compliance with moral norms, rather than with
moral capacity, because he takes the latter to be already included in the predisposi-
tion to humanity.42 As we have seen, Kant firmly denies that possibility.
Wood’s most extensive engagement with the humanity/personality distinction
is found in a long footnote to the passages discussed above. There, Wood asks,
“are humanity and personality necessarily coextensive?” (364). He accurately
notes that although the two concepts are treated as more or less equivalent in
the Groundwork, the theory of the Anlage—especially Kant’s important footnote
quoted above—seems to claim that human non-persons are at least a conceptual
possibility. While granting that the two concepts are in some sense “distinct,”
Wood doubts that “there could be a reason that reflects on objects of desire and
combines them into a sum of satisfaction but fails to recognize morality as an incen-
tive.” The heart of Wood’s position, and indeed of the general value of humanity
approach to Kant’s ethics, is found in the argument that follows this expression
of doubt. Wood writes:
For if humanity involves the rational capacity not only to set ends but also to compare
their objects and fashion a whole based on priorities among these values, then reason
would seem to include the capacity to make comparative judgments of value, hence
an (at least implicit) awareness of the rational standards of value­­–– whose ultimate
foundation, Kant claims, is the dignity of humanity or rational nature as an end in
itself. (365)

In fact, Kant never claims that the dignity of humanity as an end in itself (i.e.,
personality) is the necessary foundation of any judgment of rational value. If it
were, moral norms could indeed be deduced from norms of practical reason as
such. Above, we saw the disastrous consequences of such a view for Kant’s theory
of the fact of reason, for his conception of evil as accountable, and for the doctrine

However, what about non-human moral beings, such as God? While God may be considered
a person, he is not someone to whom we owe respect in the sense of having duties towards him (Rel,
6:154). Indeed, we cannot treat God in any manner, neither as a mere means, nor as an end in himself.
(See MdS, 6:443–44; Rel, 6:175–82.) Although I cannot defend the claim here, I believe that the “finite
holy beings” discussed in the Doctrine of Virtue are indeed owed respect (MdS, 6: 383), but that they are
also human (according to the technical sense given by the theory of the original predisposition).
Wood may have shifted positions in his commentary to the Groundwork, which appears to
conclude that only morally legislative personality—and not the mere capacity to set ends—serves as
humanity’s qualification as an end in itself. See Allen Wood and Dieter Schönecker, Kants “Grundle-
gung zur Metaphysik der Sitten”: ein einführender Kommentar (Paderborn: Verlag Ferdinand Schöningh,
2002), 143.
humanity and evil 655
of transcendental freedom as entailed by (and only by) obligation to morality.
Wood considers the latter issue in this same footnote. “Perhaps Kant’s thought
is that a being might be instrumentally or prudentially rational without being
transcendentally free,” he writes. But, Wood argues, any being who can reason
prudentially must be “practically free” to set ends, and is therefore “capable of
resisting sensuous impulses.” Thus, “if transcendental freedom is necessary for
practical freedom, then such a being must already be transcendentally free.” In fact,
transcendental freedom is not necessary for practical freedom in Wood’s sense.
Just because a being can use practical reason to order and prioritize inclinations,
and therefore can resist a given “sensuous impulse” in pursuing some end, this
does not entail that the choice of that end—and indeed of the entire system of
ends—is not itself determined by inclination. In order to conclude that we can
determine our will free from all determination by inclination, we must be capable
of acting, not just rationally, but morally. That is, our practical reason must have
not just a practical use, in which case it might still be ultimately a slave of the pas-
sions, but a pure practical use.
Wood closes his footnote with a discussion of the following “serious problem”
that would be generated by the possibility of a merely human non-person. Given
that possibility, Kant’s
doctrine of the theoretical unprovablity of transcendental freedom would commit
him to the admission that we have no way of empirically distinguishing human be-
ings from such creatures. That would entail that there could be no way in principle
of answering a moral skeptic who argues that people are prudentially rational but
not morally accountable. (365–66)

I fail to see why this is a “serious problem” for Kant’s theory. As Wood concludes,
“for practical purposes, then, Kant must assume that personality is found wher-
ever humanity is found” (366). This is certainly the case, but such an assumption
on practical grounds does nothing to answer the moral skeptic, who asks for a
theoretical demonstration that empirically given people are in fact moral persons.
According to Kant’s theory, there is no way of giving such a demonstration, and
thus no way of answering the skeptic.43
I have argued that the denial of the normative independence of humanity
and personality undermines Kant’s conception of evil as morally accountable.
Kant’s Ethical Thought, in which the denial of this normative independence is
defended at length, contains an extensive discussion of Kant’s theory of radical
evil. Wood sees the theory of radical evil as a central element in Kant’s neglected
conception of anthropology. Wood’s emphasis on the role of anthropology in
Kant’s ethical view is well-taken, and his discussion illuminates the rich and com-
plex understanding of human nature and the nature of human society at work
in Kant’s writings. Indeed, Wood is certainly correct that much of Kant’s ethical
doctrine could not be properly understood without taking account of Kant’s views
regarding the empirical anthropological context in which our practical reason
operates. However, Wood’s anthropological approach to the question of evil goes
beyond the claim that the development and expression of human evil is shaped

Of course, the skeptic himself is unable to prove that people are not moral persons.
656 journal of the history of philosophy 44:4 october 2006
by certain predominant features of our empirically given nature; he insists that
the very “source” of evil in human beings must be understood as empirically
given (288).44 Wood argues that radical evil is rooted in our natural tendency to
associate with and compete against one another; it is no more or less than our
“unsociable sociability” (287–89). The difficulty with this reading is that it threat-
ens to undermine the conception of evil as a product of freedom.45 For, unlike an
evil disposition, which must be freely chosen, unsociable sociability is part of our
given nature. Indeed, the latter is already attributed in the Anlage to humanity, in
so far as it involves not only “comparison with others,” but also a “competitiveness”
that serves as “an incentive to culture” (Rel, 6:27)—the constituent elements of
what Kant elsewhere calls ‘unsociable sociability’ (IaG, 8:20). As Kant makes clear,
such a predisposition cannot be considered a source of evil; on the contrary, he
claims that our humanity—which would include our unsociable sociability—is a
given predisposition to the good. Our competitive nature can only be considered
evil, and therefore culpable, if we freely incorporate our natural drive towards
manipulation and domination of others into our maxim as dominant over moral
requirements. Therefore, it cannot be the case, as Wood argues, that radical evil “is
to be accounted for by natural purposiveness (in the same way we account for our
animal instincts)” (401–02). Such a “naturalistic” understanding of evil removes
the dimension of transcendentally free accountability, and thus ultimately removes
evil’s ethical dimension. Of course, if we did not have a distinctive predisposition
to personality, but were merely human, Wood’s reading of radical evil as a conse-
quence of our given nature would be quite natural, since we would have already
given up on attributing genuinely moral good or evil to human agents. Wood’s
study of Kant’s ethics thus provides suggestive evidence that the value of human-
ity interpretation leads to a loss of the conception of evil as strictly imputable. If
Wood’s emphasis on the systematic significance of the theory of radical evil is well
placed, it is even more crucial that evil not be drained of ethical meaning in the
manner of his naturalizing value of humanity interpretation.
Is Wood’s naturalistic conception of evil the only option available to a value of
humanity interpretation? In the work of Christine Korsgaard, another conception
of evil emerges: one that is, in many ways, more philosophically interesting, and

Wood holds that the claim that human nature is radically evil is empirical, although it is not
an inductive generalization. While a full critique of this position is not possible here, I note that the
thesis of radical evil is better understood as a synthetic a priori proposition, in so far as it predicates a
certain (negative) relation between freely adopted human subjective practical principle (which is not
empirically observable) and a priori moral law. Accordingly, this thesis can only be justified through
(a priori) conceptual analysis of the conditions of the possibility of certain elements of moral experi-
ence—on my reading, most crucially, the phenomenon of temptation. Here I follow Allison’s critique of
Wood in his “Ethics, Evil, and Anthropology in Kant: Remarks on Allen Wood’s Kant’s Ethical Thought,”
Ethics 111 (2001), 605–10.
Wood is aware of this difficulty and insists both that evil “is an act of our freedom” (401 n.4)
and that individual agents “bear the entire moral responsibility” for their evil (289). I do not see,
however, how it is possible to impute radical evil to an agent’s free will and at the same time claim
that the source of his or her radical evil lies in human social nature. Wood’s strategy for reconciling
these positions—namely, by construing the attribution of radical evil as a merely reflective teleological
judgment rather than as a genuine imputation or moral judgment (289–90)—appears to be out of
step with Kant’s own characterization of radical evil as an ethical imputation, indeed, as the condition
of the possibility of the imputation of evil as such (see, e.g., Rel, 6:37).
humanity and evil 657
yet no less removed from Kant’s own view as Wood’s de-ethicized conception. As
we shall see, her “negative” conception of evil emerges directly from her view of
the kind of deduction possible of Kantian moral principle.

Korsgaard offers perhaps the most robust defense of the view that moral norms are
derivable from a conception of rational agency in general. While other value of
humanity interpreters have expressed doubt about the success of such a deduction,
her basic argumentative strategy can nevertheless be found submerged in their
work. What I refer to as Korsgaard’s ‘basic argumentative strategy’ is her effort
to mount an immanent deduction of the validity of moral principle by means of
a regressive reflection on the conditions of practical rationality as such.46 Rather
than directly challenge the validity of the deduction on its own terms, I want to
consider a basic problem with her deductive approach in general.47 As we shall
see, as long as evil is construed as strictly imputable, such a deduction will remain
Korsgaard argues that unconditional value, and indeed the unconditional value
of humanity as the power of practical reason, is the necessary condition of value
as such, and therefore of any rational choice whatsoever.48 She writes: “if there is
such a thing as a reason for action, then humanity, as the source of all reasons and
values, must be valued for its own sake” (Sources of Normativity, 122). Of course, to
act according to the unconditional value of humanity is precisely to act morally, or
under the categorical imperative in its formulation of humanity. On Korsgaard’s
reading, a “fully rational” practical reason is both objective and unconditional—or
at least is conditioned by an objective, unconditional rational norm. Korsgaard’s

The deduction by regression appears in several forms in Korsgaard’s works. In Creating the
Kingdom of Ends, the deduction trades on the nature of practical reasons and the “rational concept”
of the good (120–23). The Sources of Normativity makes the notion of “practical identity” central, such
that a commitment to any practical identity has moral identity as its necessary condition (101–05). In
“Self Constitution in the Ethics of Plato and Kant” [“Self-Constitution”] (The Journal of Ethics 3 [1999]:
1–29), the argument is that moral principle provides the only adequate ground for constituting one’s
self as a unified, integrated agent.
For critiques of the regressive deduction, see discussions in Guyer (Kant on Freedom, Law, and
Happiness, 150–51); Pippen (“The ‘New Kant’,” 323); Samuel Kerstein (“Korsgaard’s Kantian Argu-
ments for the Value of Humanity,” Canadian Journal of Philosophy 31 [2001], 30–35; 45–51). Hannah
Ginsborg’s valuable critique (“Korsgaard on Choosing Nonmoral Ends,” Ethics 109 [1998]: 5–21) draws
attention to Kant’s humanity/personality distinction (19). Donald Regan’s critique is instructive, in so
far as his realist approach shares a key assumption of the Kantians he criticizes, namely, that rational
choice as such presupposes a conception of unconditional value (“The Value of Rational Nature,” Eth-
ics, 112 [2002]: 267-91). As we have seen, Kant denies this, holding to the contrary that the existence
of unconditional value is not a condition of the possibility of rational choice—that is, that humanity
and personality are normatively independent. Regan considers briefly the possibility of establishing
the unconditional value of humanity on the grounds of our morally legislative reason, but rejects this
approach (which he calls “‘classical’ Kantianism”) both because he believes a formal principle cannot
generate duties and because he believes (like the value of humanity interpreters) that moral value re-
quires a non-circular grounding (287–88). In this sense, Regan and the value of humanity interpreters
are also united in their rejection of the theory of the fact of reason. See also David Sussman’s limited
defense of the regress deduction in response to Regan’s critique, which in sum tends to support the
present interpretation (“The Authority of Humanity,” Ethics 113 [2003]).
Compare Herman’s value of humanity interpretation, which claims that “if anything is good . . .
there must be something that is unconditionally good” (The Practice of Moral Judgment, 209).
658 journal of the history of philosophy 44:4 october 2006
view that an agency in the absence of objective norms could not be considered
rational is rather plausible. After all, practical reason as such must operate within
the “space of reasons”; that is, it must by guided and justified by norms which hold
generally for all particular wills, and thus are not merely subjective. But Korsgaard
claims more than this. Korsgaard claims that the space of reasons required by ra-
tional agency necessarily contains absolute or unconditional norms, holding that
the objectivity of practical reasons entails the unconditionality of practical reasons
(Creating the Kingdom of Ends, 122). Kant, however, does not hold that objective
norms or imperatives must legislate an unconditional value to function as genuine
reasons.49 Hypothetical imperatives are objective (they hold for all rational agents),
and yet prescribe action as only conditionally good. Thus, it is not Kant’s view
that “the price of denying that humanity is of [unconditional] value is complete
normative skepticism” (Sources of Normativity, 163). Such a denial would entail the
non-existence not of rational choice as such, but only of morality.
Korsgaard’s view that the objectivity of practical norms presupposes an uncon-
ditional practical principle—or, as she argues in Sources of Normativity (101–05),
that practical identity as such presupposes moral identity—is equivalent to the
denial of the normative independence of humanity and personality. For accord-
ing to the theory of the Anlage, humanity is just the capacity to act according to
objective, and yet conditional principles, while only the “special” predisposition to
personality gives us the capacity to act according to unconditional moral norms.
Korsgaard writes that “humanity”—or the power of rational choice—is not fully
realized unless choice is fully rational. On her view, this occurs when humanity is
“completed and perfected” in the realization of “personality,” or when its choice
is determined by moral principle. Thus, on Korsgaard’s account, the predisposi-
tion to personality (or the “capacity for the good will”) is nothing but humanity,
or the capacity for rational choice as such (Creating the Kingdom of Ends, 122–23).
Moral norms, specifically the concept of the objective, unconditional good, are
already entailed by general norms of practical reason. When we act according to
our subjective, conditional good without heeding the objective, unconditional
good, our action is “contradictory,” and we are to that extent not acting as rational
agents. That is, violation of the status of humanity as the ultimate value disqualifies
a choice from being rational (or at least “fully rational”). All of this is denied by
Kant in the theory of the Anlage, where Kant points out that no amount of “subtle
reasoning” or regressive reflection upon the conditions of practical rationality
would ever generate moral principle, or the concept of an unconditional good,
unless obligation to the latter were given as a fact of reason—unless, that is, humans
had a separate, “special” predisposition to personality. Korsgaard’s basic strategy,
on the other hand, rejects the theory of the fact of reason; she prefers to deduce
moral principle by way of a “transcendental argument,” thereby showing that the

The claim that objectivity of practical norms entails unconditionality can also be found in Donald
Regan’s critique of Korsgaard, in so far as he shares her view that practical value as such must have an
unconditional ground (“The Value of Rational Nature,” 272–74). See David Sussman’s response to
Regan, where he argues that an ambiguity in the sense of ‘objective’ in the second part of the Ground-
work leads to misunderstanding of Kant’s position (“The Authority of Humanity,” 7).
humanity and evil 659
condition of the possibility of “rational action” is the unconditioned moral value
of humanity (Sources of Normativity, 123–24).50
The ethical consequences of Korsgaard’s denial of the normative independence
of humanity and personality—especially regarding the conception of evil—are
significant. Since Korsgaardian moral agency is just action that is rational in the
fullest sense, evil action must be seen as irrational. The evil agent, on her reading,
becomes a failure as an agent. But since the free will is just practical reason, the
evil agent must be seen as less free than the good agent. For Korsgaard, moral
worth, freedom, and rationality are all directly proportionate; the most moral
action is the most free, as well as the most rational. So long as moral principle is
taken to be necessarily entailed by the conception of rational agency as such, this
proportionate relationship must hold. Korsgaard writes: “By acting morally, we make
ourselves free” (Creating the Kingdom of Ends, 176). She claims that “morality is the
natural condition of a free will. The free will that puts inclination above moral-
ity [i.e., the evil will] sacrifices its freedom for nothing” (Creating the Kingdom of
Ends, 167). However, for Kant, we must already be free when we face the choice
between good and evil.51 The evil will cannot be one that sacrifices its freedom,
since it must retain its freedom in order to be held morally accountable for its
evil. This sort of freedom is not, in Kant’s view, something we are capable of los-
ing. On the contrary, for Kant the evil will is one that—in full possession of its
reason—subordinates the unconditional norms of morality for the sake of the
more-or-less prudential interest in one’s own, selfish (conditional) good. On Kant’s
view, therefore, it is quite false that “a free person, as such, follows the moral law”
(Creating the Kingdom of Ends, 160).52

If my reading of Kant is correct, then, despite Korsgaard’s hope, Kantian practical philosophy
cannot offer the “reply to the moral skeptic” (Sources of Normativity, 163). The doubt that there is such
a thing as morality cannot be dispelled (according to Kant) because unconditional obligation as such
cannot be justified by reference to some principle external to morality. For the same reason, the demand
that a sensible answer be provided for the question, ‘Why be moral?,’ must be judged (from a Kantian
perspective) as already betraying a misunderstanding of the kind of obligation morality imposes.
See also Herman’s critique of “canonical deontology,” which according to her is unsatisfactory in
so far as it leaves “moral skepticism” in place as a “reasonable response” (The Practice of Moral Judgment,
210). To answer the skeptic, on her view, Kantian theory must be able to explain the “rationale” for
moral action, namely by an appeal to the claim that immoral action is “inconsistent with principles
that describe the nature of rational agency per se” (The Practice of Moral Judgment, 226–27).
For an instructive account that is sympathetic to the “temptation to try to demonstrate a priori
that reason [as such] must prescribe . . . the fundamental principle of morality” (237), see Donagan
(The Theory of Morality, 237–43).
This way of putting the matter can be misleading, since the claim of the theory of radical evil
is precisely that we never occupy a position of moral neutrality from which we evaluate our ethical
choices; rather, human moral life “begins in sin.” We find that we have already committed ourselves
to evil, and therefore we must take up a positive struggle to overturn our anti-moral orientation. (Rel,
6:38). Nevertheless, our freedom in the strong sense is conceptually prior to our ethical orientation.
Note, however, that this view does not commit Kant to a conception of “morally neutral autonomy.”
Kant still holds that the autonomous will necessarily binds itself to moral law, even if an autonomous
will is not by that token a will oriented towards fulfilling its obligations. This consideration also explains
why Sidgwick’s characterization of transcendental freedom as “Neutral freedom” is inaccurate (“The
Kantian Conception of Free Will,” 512).
So expressed, it becomes especially clear that Korsgaard’s reading of Kant is vulnerable to the
Sidgwick objection. Sidgwick alleges that freedom for Kant is not “a faculty of laying down laws which
may or may not be obeyed; [Kant] must mean that the will, qua free, acts in accordance with these
660 journal of the history of philosophy 44:4 october 2006
Of particular interest is Korsgaard’s gloss of evil as “unintelligible” (Creating
the Kingdom of Ends, 171). This is apparently her translation of ‘unerforschlich’: a
term Kant does apply to evil, which is perhaps better rendered as ‘inscrutable’
(Rel, 6:21). By characterizing the choice of evil as inscrutable, Kant means that it
belongs to our ultimate maxim, which by definition cannot be justified or explained
by a more fundamental maxim. For precisely the same reason, a fundamental
choice of virtue in our ultimate maxim would also be inscrutable. For Korsgaard,
however, the choice of evil is “unintelligible” in the sense that it makes no sense
(since it violates the intrinsic norms of practical reason as such), and is a “wholly
unmotivated abandonment of freedom” (Creating the Kingdom of Ends, 171). But
evil for Kant is motivated, namely by inclination. Moreover, it is not an abandon-
ment of freedom, but a free subordination of moral requirements to self-love.
Korsgaard notes that her conception of evil as unintelligible may appear to raise
problems for evil’s imputability, observing that “if the moral law is the unique
positive conception of freedom, then it seems as if only morally good actions
are really free” (Creating the Kingdom of Ends, 172). Her solution appears to be to
consign the imputation of evil action to the “explanatory standpoint of theoreti-
cal reason,” admitting that on her account, such an imputation is “completely
unintelligible” from a practical point of view (Creating the Kingdom of Ends, 173).
But to construe the imputation of evil as a theoretical explanation is to radically
undermine its ethical dimension; explanation here would function as excuse.
This solution clearly indicates that Korsgaard’s reconstruction of Kant sacrifices
his ethical conception of evil as genuinely imputable in so far as it renders the
imputation of evil impossible from a practical point of view.
Korsgaard’s view of evil is given a far more expansive expression in her article
“Self-Constitution in the Ethics of Plato and Kant,” whose basic ideas are developed
further in her Locke Lectures of 2002.53 The general thesis of these works is that
good action is a matter of “self-constitution” (“Self-Constitution,” 3), by which an
agent gives unified structure to his or her will, constituting him- or her- self as an
agent. In this context, Korsgaard addresses the question of how bad action is even
possible, since by her hypothesis only good action is attributable to a unified, effec-
tive, whole agent. In fact, bad or evil action, according to Korsgaard, is “defective
as action” and arises from a person’s failure to properly constitute their agency
(“Self-Constitution,” 16).54 But who then is held accountable for the decision to

laws” (“The Kantian Conception of Free Will,” 515). Korsgaard, I take it, is prepared to defend such a
reading of Kant’s view. My contention is that such a reading of Kant condemns his theory to precisely
the failing Sidgwick saw, namely, the inability to account for evil as imputable.
“Self-Constitution in the Ethics of Plato and Kant” [“Self-Constitution”], The Journal of Ethics 3
(1999): 1–29, and “Self-Constitution: Actions, Identity, & Integrity” [Locke Lectures], Locke Lectures,
University of Oxford, 22 May 2002; Harvard University, 8 April 2003, <http://www.people.fas.harvard.
Sources of Normativity contains no elaborated discussion of evil. However, Korsgaard’s reply to G.
A. Cohen’s objection that fulfillment of “practical identity” does not require compliance with morality
is noteworthy. Cohen imagines an “idealized Mafioso,” who is a determined, principled agent, although
one with evil principles (G. A. Cohen, “Reason, Humanity, and the Moral Law”, in Sources of Normativ-
ity, 183–84). Korsgaard responds to the possibility of the Mafioso’s evil practical identity by claiming
that there is “no coherent point of view from which it can be endorsed in the full light of reflection”
(Sources of Normativity, 256). Such a claim about evil is not only false according to Kant’s theory—
there is nothing “incoherent” for him about an evil practical identity—but also it misconstrues
humanity and evil 661
constitute themselves in such an improper way? Korsgaard suggests that, according
to Kant, no one really decides to be evil; rather the evil person “follows the lead of
inclinations, without sufficient reflection” (Locke Lecture V, 16). But on Kant’s
view, in order to count as evil in the required, culpable sense, the choice of evil
must be a free “decision” of the agent. Indeed, contrary to Korsgaard’s theory of
self-constitution, Kant’s conception requires that a person’s autonomous agency
be taken as conceptually prior to their commitment to or against morality—that
their agency be already constituted when they subordinate morality to self-love.
Korsgaard’s distinction between “negative” and “positive” theories of evil is espe-
cially revealing, in so far as it is incapable of accommodating the Kantian view. On
the negative conception (which she ascribes to Kant and Plato), “evil is a privation,”
a “lack,” and “a psychological failure,” while good is taken to be a substance (Locke
Lecture V, 23).55 On the positive conception (represented in her account by Plato’s
Thrasymachus), evil is seen as a real power, while good is seen as a form of “weak-
ness” (Locke Lecture V, 24). While Plato’s view of evil may plausibly be read as
“negative” in this sense, Kant in fact rejects both sides of Korsgaard’s distinction.
It is not the case that “the structure of the problem of evil” in Kant and Plato “is
exactly the same,” as Korsgaard claims (“Self-Constitution,” 13). For Kant, both evil
and good are positive forces of a free will. Most fundamentally, this is required by
his commitment to holding both evil and good to be accountable products of a will
free in the fullest sense. The comparison with Plato is helpful here because Kant
viewed his conception of evil as a crucial point of disagreement between himself
and the ancients (Rel, 6:60 n.). Kant belongs to a modernity deeply influenced
by Christianity (especially in the form developed by St. Augustine) in which the
human moral condition is marked by the predicament of a free will which must
decide between two possible allegiances: to good or to evil.
Finally, Korsgaard’s distinction between the “Combat model” of the will (in
which reason battles with sensibility for the upper hand in controlling the agent)
and the “Constitutional model” (in which reason carries out a self-constitution of
the agent in the first place) must also be judged as failing to make room for the
Kantian view (“Self-Constitution,” 1–3; Locke Lecture V, 2–4). Human character,
for Kant, is indeed a matter of combat—in Kant’s terminology, a Kampf (Rel,
6:57)—not between reason and sense, but between reason and itself. While Kant
would agree with Korsgaard that there is no real agency at all in the “Combat
model,” he would insist that, unless the source of moral goodness is normatively

the very ethical nature of evil, reducing it to a matter of insufficient reflection. According to Kant, the
moral responsibility of the evil agent is not to reflect more extensively and abstractly on the rational
conditions of his practical identity, but rather to effect an ethical “revolution” or “rebirth” in the basic
orientation of his character (Rel, 6:47).
Korsgaard’s preferred models of evil agency are “the drunk in the gutter, the junkie, the stupid
hothead” (Sources of Normativity, 23), as well as the “dominating obsession” of the “serial sex killer”
(Sources of Normativity, 25). Note that each of these cases lends itself to an interpretation involving
some significant element of diminished freedom, and therefore diminished responsibility. The actions
of the drunk and the junkie are determined by chemical influence, or perhaps by the disease that is
their addiction; the hothead’s crimes can be attributed to his unintelligence, which isn’t his fault; the
sex killer, at least in the popular understanding, is typically someone whose agency is overtaken by a
pathological disorder. Kant’s interest in strictly imputable evil is ill served by an emphasis on these
less-than-free characters.
662 journal of the history of philosophy 44:4 october 2006
independent from the constitutive standards of agency as such—that is, unless the
“Constitutional model” is rejected—no sense can be made of the ethical Kampf
which dominates our moral experience.

3. conclusion
The inability to account for evil as genuinely imputable is an endemic problem
for the three value of humanity interpretations discussed above. Moreover, it is
precisely the privileging of humanity or mere rational nature as the ground of a
theory of value that leads to the loss of an ethically rich conception of accountable
evil, since the latter necessarily requires the thesis of the normative independence
of humanity and personality. Thus, when Korsgaard claims that “a Kantian does
not believe in the split between rationality and morality” (Creating the Kingdom of
Ends, 395 n.40), it must be pointed out that, while this may be an apt characteriza-
tion of Guyer, Wood, and herself, any Kantian who remains committed to giving a
meaningful account of human moral life—marked as it is by the ethical struggle
against evil—must insist on the necessity of such a split. As we have seen, one such
Kantian was Kant himself.
In the value of humanity reconstruction of Kantian theory, the available con-
ception of evil is either “naturalistic” (Wood), or “negative” (Korsgaard). Thus,
evil in this moral-theoretical context is seen either as an absurd (while perhaps
purposive) instinct, or else as ranging from insanity to obtuseness, from incoher-
ence to incompetence. Evil as “malevolence,” “perversity,” “corruption,” “pride,”
“depravity,” “frailty,” and indeed as “sin”—that is, evil in the senses that interested
Kant—is especially difficult to understand from the perspective of the value of
humanity’s naturalistic or negative conceptions (Rel, 6:29–30). Moreover, con-
ceptions such as these would require us to subscribe to doubtful claims about
an agent’s guilt over his or her incompetence or obtuseness, struggling against
the temptation to be ineffective, and requiring virtue or strength of will to resist
such temptation. The value of humanity interpretation may find itself required to
forsake an account of guilt, temptation, and virtue; or at least it will be forced to
endorse rather violent reconfigurations of these familiar ethical notions.56 More
fundamentally, the basic picture of what is at stake in moral life according to this
interpretation—namely, success or failure in achieving fully effective, fully rational
choice—is rather distant from Kant’s conception of what is at stake, which is the
“victory” or defeat of “the good principle over the evil” (Rel, 6:93).
The present critique of the value of humanity interpretation on the grounds
of its incompatibility with the conception of accountable evil raises the following
question: Is an accounting of culpable evil—together with the related phenom-
ena of temptation, guilt, struggle, and progress—a decisive desideratum for an
ethical theory? Many current and historical theories, it must be admitted, do not
construe it as such. Nevertheless, that Kant so clearly did constitutes one of his

In Sources of Normativity, Korsgaard presents a brief account of the “negative moral emotions,”
according to which guilt is understood as a painful perception of an unfulfilled obligation. Note,
however, that what is offered is a general account of both “regret and remorse,” and indeed that Kors-
gaard is unable to distinguish between (prudential) regret and (moral) remorse, except in terms of
degree (150–51).
humanity and evil 663
major contributions to moral thought, and any reinterpretation that sacrifices
his conception of the imputability of evil renders this contribution unintelligible.
Moreover, although I can only offer the following series of bald suggestions, it is
likely the case that Kant’s conception of evil has systematic relationships to other
areas of his critical project. The conception of virtue as requiring the pursuit of
moral perfection, the doctrine of the ethical commonwealth as a morally neces-
sary ideal of a public moral community,57 the view of the indirect moral function
of aesthetic judgment,58 the doctrine of the obligatory status of the highest good
defined as a union of perfect virtue and complete happiness,59 the teleological
conception of history, and the conception of moral religion and rational faith all
may prove to have deep connections to Kant’s conception of evil as the central
problematic of the human moral condition. Were we to choose the path of the
value of humanity interpretation, we would close off a far richer understanding
of what Kant’s ethical thought has to offer us.

Sharon Anderson-Gold argues that the problem of radical evil gives rise to a novel conception
of virtue as requiring the collective response of the ethical commonwealth; “Kant’s Ethical Common-
wealth: the Highest Good as a Social Goal,” International Philosophical Quarterly 26 (1986): 23–32.
Allison argues that the need for “counterweights” against our radically evil tendency grounds
the moral function of taste (Theory of Taste, 229–35).
I argue that the duty to promote the highest good and the theory of radical evil reciprocally
entail one another in “Kant’s Conception of the Highest Good, the Gesinnung, and Radical Evil,”
Kant-Studien 97 (2006): 184–209.