<<

. 2
( 8)



>>

footnote in the Groundwork Kant offers a dense de¬nition of inclination:
The dependence of the capacity of desire on sensations is called inclination, and
inclination always indicates a need. The dependence of a contingently determinable
will on principles of reason, however, is called an interest. An interest is present only
in a dependent will, which is not of itself always in conformity with reason; in the
divine will we cannot conceive of an interest. But even the human will can take an
interest in something without therefore acting from interest. The former signi¬es the
practical interest in the action; the latter, the pathological interest in the object of the
action. The former indicates only the dependence of the will on principles of reason
in themselves, while the latter indicates the dependence of the will on principles of
reason for the sake of inclination, since reason gives only the practical rule by which
the needs of inclination are to be aided. In the former case the action interests me,
and in the latter the object of the action (so far as [sofern] it is agreeable to me)
interests me. (GMS 413, note)

We need to go carefully in order to understand the note™s main points.
As a ¬rst step, let us focus on Kant™s notion of the capacity of desire
(Begehrungsverm¨gen ). Although it has largely been neglected, this notion is
o
one of the most fundamental in Kant™s theory of agency.29 An agent™s capac-
ity of desire, says Kant, is her capacity to cause, through her representations
Kant™s Theory of Agency 25

of objects, the reality of these objects (KpV 9, note; KU 177, note; MS 211).
The term “representation” (Vorstellung) refers here to a mental representa-
tion, that is, an idea; “object” refers to a state of affairs or to an event. An agent
who had an idea of an object and who brought about the object through
this idea would count as having exercised her capacity of desire with respect
to this object. For example, a person who had an idea of catching a butter¬‚y
and who, guided by this idea, caught one would count as having exercised
her capacity of desire with respect to catching the butter¬‚y.30 It is crucial to
recognize that by Kant™s de¬nition the capacity of desire is not a capacity to
have or to acquire a desire. Rather, it is a capacity to try to realize a desired
object. It is a capacity to act on a desire.
In the Groundwork footnote, Kant says that inclination is the dependence
of the capacity of desire on sensations. When an agent acts from inclination,
suggests Kant, his capacity to realize an object through his idea of it is de-
pendent on sensations. Kant gives only an indirect answer to the question of
how this capacity is dependent on sensations, an answer that emerges from
his discussion of the concept of interest. Kant de¬nes an interest as the de-
pendence of a contingently determinable will “ for example, the human
will “ on principles of reason. The human will is by de¬nition dependent on
principles of reason. For whenever an agent exercises her will, she does so
on at least one such principle (GMS 412). Thus, whenever an agent acts, she
has some interest. Kant distinguishes practical from pathological interest. He
identi¬es a practical interest as an interest in an action itself. An agent, he
says, takes a practical interest in an action when she acts from duty. A patho-
logical interest is an interest in the object (i.e., end or aim) of an action,
rather than in the action itself.31
Kant claims that when an agent acts from a pathological interest in the end
of an action, the end interests him “so far [sofern] as it is agreeable” to him. In
other words, to act from pathological interest is to act to realize an end that
one is interested in realizing so far as he expects that its realization would
give him pleasure.32 Yet what does it mean to be interested in realizing an
end, so far as one believes that its realization would give him pleasure? On my
view, Kant™s text permits two different readings of this notion: the ¬rst leads
us to the traditional interpretation of acting from inclination; the second,
to the alternative interpretation. According to the ¬rst, to be interested in
realizing an end so far as one believes that its realization would give him
pleasure amounts to being interested in the end to the extent that one expects
to gain pleasure from its realization. The more pleasure one expects to gain
from realizing the end, the more interested one is. Since, according to this
reading, one™s pathological interest in an end is directly proportional to the
pleasure one expects from it, it is natural to assume that when one acts from
pathological interest in the end, pleasure from it is one™s only motive.
According to the second reading “ the one that leads to the alternative
interpretation “ Kant holds that a necessary condition for the agent™s interest
Search for the Supreme Principle of Morality
26

in the end is that he believe that its realization would give him pleasure.33
Kant conceives of acting from pathological interest in an end as trying to
realize the end on condition that one expect pleasure from doing so. For
example, when, from pathological interest, someone attempts to master a
piano sonata, her attempt is conditional on her expectation that mastering
it would give her pleasure (see also GMS 442). As Kant makes clear in the
note, acting from pathological interest amounts to acting “for the sake of
inclination.” In effect, Kant equates acting from pathological interest with
acting from inclination. Therefore, according to our second reading, Kant
conceives of an agent™s acting from inclination as her trying to realize an end
only if she expects the end™s realization would give her pleasure.34 Strictly
speaking, that an agent performs certain actions only on the condition that
she expects pleasure from doing so does not entail that she has hedonic
motivation in performing them. After all, the pleasure the agent necessarily
expects when she acts from inclination might be instrumental to, or serve
merely as a sign for the attainment of, some further end she has. However,
Kant does not seem to have these possibilities in view. He seems to embrace
the notion that in acting from inclination an agent always has some hedonic
motivation. In the second Critique, for example, Kant (as we will see) clearly
suggests that when an agent acts on material practical principles (i.e., from
inclination), his expectation of gaining pleasure constitutes a determining
ground of his acting.
Kant™s account of inclination in the Groundwork note weighs against
Reath™s interpretation. Reath asserts that Kant holds pleasure to play a role
in the development of inclinations. This assertion seems true (see KU 207).
But, as his remarks regarding Kant™s famous example of the “philanthropist”
(or “friend of humanity”) will soon reveal, Reath also suggests that once an
agent has developed a Kantian inclination, it is not the case that he acts from
it only on condition that he expect pleasure from his action. This suggestion
seems misguided. In the note, Kant strongly implies that an agent™s expecta-
tion of experiencing pleasure plays a role each time he acts from inclination:
a role as a motive for acting on the alternative interpretation; a role as the
agent™s only motive for acting on the traditional interpretation.
In his interpretation of Kant™s account of inclination, Reath does not men-
tion the Groundwork note. Nevertheless, he does appeal to the Groundwork
to bolster his rejection of the traditional interpretation of acting from incli-
nation. In particular, Reath appeals to the example of the philanthropist, a
person who helps others not from duty but rather from inclination. Accord-
ing to Reath, Kant holds the following: “The object of [the philanthropist™s]
concern and the motive of his actions is their [others™] happiness.”35 The
philanthropist, Reath unambiguously suggests, does not have the expecta-
tion of his own pleasure as a motive in helping others. On Reath™s inter-
pretation, Kant rejects the notion that an agent™s expectation of his own
pleasure constitutes a motive in all acting from inclination.
Kant™s Theory of Agency 27

But a close look at Kant™s remarks regarding the philanthropist will, I
believe, show this interpretation to be ¬‚awed. Kant says: “To be bene¬cent
where one can is a duty; and besides this, there are many souls so sympathet-
ically constituted that, without any further motive of vanity or self-interest,
they ¬nd an inner pleasure in spreading joy around them and can rejoice
in the satisfaction of others, so far as it is their own work” (GMS 398). Here
Kant speaks of sympathetically constituted persons, of whom the philan-
thropist is one, who ¬nd an “inner pleasure” (inneres Vergn¨ gen) in spread-
u
ing joy around them. According to Reath, Kant holds that, in their helping
actions, such persons have improving the lot of others as their only mo-
tive. The “inner pleasure” they experience stems from their belief that they
have actually managed to spread joy to others. The attractiveness of this
interpretation is evident. It suggests that Kant understood the motives of
sympathetically constituted persons much as many of us do. But I ¬nd this
interpretation questionable. Kant does not state here that these persons fail
to have their own pleasure as a motive. He does not say that without any
motive of vanity or self-interest they try to help others. If Kant did assert
this, then Reath™s interpretation would obviously gain support. What Kant
does say here is that without any further (anderen) motive of vanity or self-
interest, the sympathetically constituted ¬nd pleasure in spreading joy to
others. This statement leaves open the possibility that, on Kant™s view, these
persons do have a motive of self-interest: the pleasure they expect to gain
from spreading joy to others. But they have no further motive of self-interest:
they are not, for example, prompted to act by the expectation that those
they help will render them some service in the future. Kant™s discussion
of sympathetically constituted persons, of whom the philanthropist is one,
does not seem to justify Reath™s rejection of the traditional (and presumably
the alternative) reading of acting from inclination.
Reath bases his interpretation of inclination mostly on Kant™s Metaphysics
of Morals de¬nition. But I argue that this de¬nition, like the Groundwork one,
fails to support his view. Instead, it lends credibility to the view that Kant must
have embraced either the traditional or the alternative interpretation.
In his discussion of agency in the Introduction to the Metaphysics of Morals,
Kant offers another dense and dif¬cult de¬nition of inclination:

As for practical pleasure, that determination of the capacity of desire which must be
preceded by this pleasure as cause is called desire [Begierde] in the narrow sense; habitual
desire in this narrow sense is called inclination [Neigung]; and the connection of
pleasure with the capacity of desire, provided that the understanding judges this
connection to hold as a general rule (though only for the subject), is called interest.
So if a pleasure necessarily precedes the determination of the capacity of desire, the
practical pleasure must be called an interest of inclination. (MS 212)

The ¬rst aspect of this de¬nition to notice is that Kant is employing the
term “inclination” in a slightly different way than he does in the Groundwork.
Search for the Supreme Principle of Morality
28

Here Kant suggests a distinction between inclination and whim. We may say
that a person has an inclination to begin her mornings with a cup of coffee.
She habitually desires to begin her mornings this way. But suppose a person
experiences a never-before-entertained desire to eat asparagus sauteed in
raspberry jam. If we employ the sense of inclination contained in this def-
inition, we may not say that she has an inclination for the dish.36 We may,
however, say that she has a “desire in the narrow sense” for it. Both inclina-
tions and what I have called whims count as such desires. In his Groundwork
de¬nition of inclination as the dependence of the capacity of desire on sen-
sation, Kant does not distinguish between inclination and whim. Since we
are interested in Kant™s general account of action not performed from duty,
we can safely bracket this distinction. Important to us is what Kant says about
desires in the “narrow sense,” which we, following Kant™s own Groundwork
usage, call “inclinations.”37
For our purposes, the central assertion in the Metaphysics of Morals passage
is D: “That determination [Bestimmung] of the capacity of desire which must
be preceded by pleasure as cause is called inclination.” Reath argues that
D amounts to the following: an inclination is a desire for an object such that
before an agent can come to have it, she must at some point have determined
that the realization of the object would give her pleasure.38 So, for example,
before I can count as having an inclination to play basketball, I must come
to the view that playing would give me pleasure. Moreover, suggests Reath,
D does not imply that once an agent has an inclination, whenever he tries
to satisfy it, he must do so on the basis of his expectation that its satisfaction
would give him pleasure. D does not imply that once I have an inclination
to play basketball, every time I try to satisfy it I do so on the basis of my
expectation that playing would give me pleasure.
Reath™s interpretation is, I believe, based on a misunderstanding of Kant™s
notion of the capacity of desire. In D, claims Reath, Kant is merely pointing
out a condition that must be ful¬lled in order for an agent to come to have
an inclination. Apparently, Reath takes the truth of this claim to be obvious.
It would indeed seem obvious, if one made, as Reath apparently does, the
following assumption: the capacity of desire is a capacity to have or to develop
desires, including inclinations. Under this assumption, D seems to set out a
necessary condition for the development of an inclination, namely that feel-
ings of pleasure play a causal role in this development. Recall that D reads:
“That determination of the capacity of desire which must be preceded by
pleasure as cause is called inclination.” The “determination” of this capacity
would, under this assumption, presumably amount to the acquiring of a de-
sire. D seems to specify that an inclination is a desire that an agent acquires
in a certain way: by being prompted by feelings of pleasure (either experi-
enced or expected) to do so. As we have noted, however, the assumption in
question is false. Although in light of its name it is tempting to think other-
wise, the capacity of desire is not a capacity to come to have a desire. Rather,
Kant™s Theory of Agency 29

it is the capacity to realize an object through one™s representation of it.
What, then, is the “determination” of this capacity? To my knowledge, Kant
never answers this question explicitly. Nevertheless, it is natural to suppose
that determining the capacity of desire amounts to choosing to realize an
object. It amounts to setting oneself to bring the object about. In effect, for
an agent to determine her capacity of desire is for her to choose to realize
the object of a desire.39
We can now see that, according to D, acting from inclination involves
making a choice to realize an object, which choice is “preceded by pleasure
as cause.” D asserts: that choice to realize an object, which must be “preceded
by pleasure as cause,” is called inclination. But what would it mean for an
agent™s choice to realize an object to be “preceded by pleasure as cause”?
We ¬nd an important clue for interpreting D in the Critique of Practical
Reason. There Kant suggests how pleasure can determine an agent to choose
to realize an object. It can do so only in the sense that her expectation of
gaining pleasure from the object™s realization determines her to choose to
realize it (KpV 22). In light of this suggestion, it makes sense to think of an
agent™s choice to realize an object being “preceded by pleasure as cause”
when the agent makes her choice because she expects to gain pleasure
from the object™s realization. For example, if someone™s choice to master a
piano sonata is preceded by pleasure as cause, then she chooses to master
it because she expects pleasure from mastering it.
On this interpretation, Kant™s Metaphysics of Morals account of inclination
coheres well with his Groundwork account. Like its predecessor, it invokes the
notion of an interest: “If a pleasure necessarily precedes the determination
of the capacity of desire, the practical pleasure must be called an interest
of inclination.” Even when we act from inclination, we act on a “general
rule” (e.g., a maxim). Inclinations do not bring about our action alone, but
when incorporated into practical rules. Moreover, like Kant™s Groundwork
account, his Metaphysics of Morals account is amenable to two readings. Kant
speaks of the determination of an agent™s capacity of desire being preceded
by “pleasure as cause.” On our interpretation, he is indicating that for an
agent to act from inclination is for her to do something because she expects
that it will enable her to gain pleasure. His account permits both a reading
on which her expectation of pleasure is her only motive and a reading on
which it is a motive but not necessarily her only one.


1.8 Material Practical Principles: Acting from Inclination
in the Critique of Practical Reason
No examination of Kant™s account of nonmoral action would be complete
without taking stock of his remarks on the topic in the Analytic of the Critique
of Practical Reason. These remarks support a rejection of Reath™s interpreta-
tion. In my view, they also permit both the traditional and the alternative
Search for the Supreme Principle of Morality
30

readings, though, as we will see, it is reasonable to contend that they ¬t
better with the traditional reading.
Before analyzing Kant™s account in the second Critique, it will be helpful
to review some of the terminological background against which it takes
shape. First, Kant says that practical principles are “propositions that contain
a general determination of the will” (KpV 19). This remark is somewhat
obscure. But I take Kant to be suggesting that practical principles “contain” a
“determination” of the will in the sense that they are rules that some agent(s)
have suf¬cient motive to act on. Second, a material practical principle is a
rule such that an agent™s having suf¬cient motive to act on it is conditional
on her view that doing so will enable her to realize some object she desires
(KpV 21). Take the rule: “During your free time, you ought to exercise.”
To say that it is a material practical principle is to say that an agent™s having
suf¬cient motive to act on it (i.e., to exercise,) is contingent on her belief
that doing so will enable her to realize some object she desires (e.g., her
staying in shape). Third, for Kant if an agent acts on a material practical
principle, then she is not acting from duty.40 Therefore, it seems, she must
be acting from inclination: to act on a material practical principle is to act
from inclination. As this book unfolds, we will have many occasions to refer
to Kant™s concept of a material practical principle.
With these points in mind, we can see that Kant™s remarks in the Analytic
of the second Critique clash with Reath™s reading of Kant. For example,
under Theorem II of “On the Principles of Pure Practical Reason,” Kant
states that all material principles “place the determining ground of the will
in the pleasure or displeasure to be felt in the reality of some object” (KpV
22). As we just noted, if a rule is a practical principle, then someone has
suf¬cient motive to act on it, and her having suf¬cient motive to act on it
is conditional on her believing that acting on it will enable her to realize
some object she desires. But, as Kant™s statement suggests, this is not the
end of the story. The agent™s having suf¬cient motive to act on the rule is
also conditional on her expectation that realizing the object she desires will
enable her to gain pleasure or avoid displeasure. Therefore, whenever an
agent acts on a material practical principle “ that is, follows the principle™s
prescription for trying to realize an object “ she has hedonic motivation. Or,
what amounts to the same thing: whenever an agent acts from inclination
(i.e., nonmorally), she has hedonic motivation.
In our discussion, the most serious question posed by Kant™s remarks in
the second Critique is not whether he held all nonmoral actions to have
hedonic motivation but whether he held them ultimately to have hedonic
motivation alone. The texts permit this reading but, in my view, do not re-
quire it. Take Kant™s claim that all material principles place the determining
ground of the will in the pleasure or displeasure to be received from an ob-
ject (KpV 22). We might read him to be saying that all such principles place
the one and only motive for willing in the agent™s expectation of pleasure.
Kant™s Theory of Agency 31

We are not, however, compelled to read him in this way. After all, Kant does
not say that all material practical principles place the determining ground
of the will exclusively in expected pleasure. What he does say permits a read-
ing according to which he acknowledges that a particular material principle
places the determining ground of the will not only in expected pleasure but
in something else as well “ for example, simply a desire to realize the object.
Here one might point out that, according to Kant, all material practi-
cal principles belong under the general principle of self-love or one™s own
happiness (KpV 22). Since Kant de¬nes happiness as the uninterrupted ex-
perience of pleasure (KpV 22), does it not follow that on his view all material
practical principles place the motive of action solely in the expectation of
pleasure?41 I think it is plausible to answer this question negatively. It is
unclear what Kant means by “the principle of happiness,” as well as by the
claim that all material practical principles belong under this principle. But
let us assume, as it seems reasonable to do, that the principle of happiness
is a principle prescribing that in order to attain maximum experience of
pleasure, an agent ought to perform those actions he believes will enable
him to do so. We may then plausibly interpret Kant to hold that all mate-
rial principles belong under the principle of happiness in the sense that an
agent™s acting on a material principle always has a feature in common with
her acting on the principle of happiness: in both cases, she has the prospect
of her own pleasure as a motive. In the former case, she has her own plea-
sure as one, though not necessarily her only, motive; in the latter, her sole
motive is presumably her own pleasure.
Admittedly, with respect to Kant™s discussion of “Theorem II,” the tra-
ditional interpretation seems to have more textual plausibility than the al-
ternative interpretation. In particular, it is more natural to interpret along
the lines of the traditional interpretation Kant™s statement that all material
principles place the determining ground of the will in the pleasure or dis-
pleasure to be received from an object. Consider a similar English usage.
A critic says: “In speaking thus, the author assumes that the motive for any-
one™s getting married lies in the desire for companionship.” In my view, the
critic might be describing the author as someone who takes the desire for
companionship always to be one motive, but not necessarily the exclusive
motive, for getting married. The critic™s statement is consistent with this in-
terpretation. But it is, I acknowledge, more natural to hold that the critic is
describing the author as someone who takes the desire for companionship
to be the only (real) motive for getting married.42

I have argued that Kant™s texts permit either a radically hedonistic or a
moderately hedonistic account of all actions that are not done from duty. It
is more charitable to attribute the latter account to him, since, unlike the
former, it does not have the highly questionable implication that, ultimately,
all actions not done from duty are done solely from hedonic motives. But, as
Search for the Supreme Principle of Morality
32

we just noted, some of Kant™s claims in the Critique of Practical Reason ¬t more
naturally with the notion that he embraced a radically hedonistic account.
I leave it to the reader to decide which of these two accounts to attribute
to Kant (though I favor attributing to him the moderately hedonistic one).
For our purposes, the important point to emerge from this discussion is
the following. Regardless of whether he upholds a radically hedonistic or
moderately hedonistic account of them, Kant maintains that all actions not
done from duty “ that is, all actions done on material practical principles “
are hedonically conditioned. That means (contrary to Reath) that an agent™s
having suf¬cient motive to perform each of these actions is conditional on
his expectation that doing so will have some hedonic payoff for himself. We
might not agree with this position, but we need to recognize that it is Kant™s.
2

Transcendental Freedom and the Derivation
of the Formula of Universal Law




2.1 Derivation in the Critique of Practical Reason:
Allison™s Reconstruction
Having settled on interpretations of some key concepts in Kant™s theory
of agency, we are ready to focus on the main topic of this book: Kant™s
derivations of the Formula of Universal Law and the Formula of Humanity.
On the traditional reading, Kant™s Groundwork derivation of the Formula of
Universal Law contains a conspicuous gap. Before I construct and defend
a new reading of this derivation, one according to which it is much more
forceful and philosophically rich than the traditional construal implies, I
¬rst consider how some contemporary philosophers have responded to the
traditional reading.
Kant offers a derivation of the Formula of Universal Law not only in
the Groundwork, but in the Critique of Practical Reason as well. At the end
of section 8 in the Analytic of Pure Practical Reason, Kant concludes that
the Formula of Universal Law “is the sole principle that can possibly be
¬t . . . for the principle of morality, whether in appraisals or in application
to the human will in determining it” (KpV 41). Kant clearly devotes parts of
sections 1“8 to defending this conclusion.1 Therefore, in light of the appar-
ent gap in Kant™s Groundwork derivation of the Formula of Universal Law, it
makes sense to look to the second Critique for a means to ¬ll it.
According to Henry Allison, Kant™s second Critique derivation relies ex-
plicitly on the premise that we have transcendental freedom. If we accept
this premise, we can close the gap that has traditionally been found between
the claim that the supreme principle of morality would have to require con-
formity to universal law and the claim that the supreme principle could
only be Kant™s Formula of Universal Law: “The problematic notion of tran-
scendental freedom must be presupposed, if Kant is to arrive at a content-
ful, action guiding moral principle; but given such freedom, the derivation
succeeds.”2 Allison defends this striking claim in a sophisticated fashion.

33
Search for the Supreme Principle of Morality
34

However, I argue that, in the end, embracing the notion of transcenden-
tal freedom helps us not at all to rescue the derivation of the Formula of
Universal Law.
This chapter is not intended as a commentary to sections 1“8 of the
Analytic of Pure Practical Reason but rather as a critical discussion of a
particular derivation Allison ¬nds there. Therefore, I discuss passages in
Kant™s text only when they are relevant to Allison™s reconstruction. Later
(7.2), I discuss (an aspect of ) a different derivation argument that, I believe,
Kant offers in the second Critique.
In outline, the argument Allison reconstructs contains four main steps.
First, assuming that we, Kantian rational agents, take ourselves to be tran-
scendentally free, we must hold that our inclinations and desires in them-
selves fail to constitute suf¬cient reasons for acting. Second, as Kantian
rational agents, we require some nonsensuously based justi¬cation for our
acting (or adopting a maxim on which to act). The idea seems to be that,
as such agents, our adopting a maxim must have some justi¬cation. Since
the mere presence of a desire cannot itself provide it, something else must
contribute to doing so. Third, this other source of justi¬cation must be the
maxim™s conformity to unconditional, universal law. For us to have suf¬cient
reason for adopting a particular maxim, our adopting it must be justi¬ed by
its conformity to a universally and unconditionally valid practical principle.
Fourth, only the Formula of Universal Law (or its equivalents) could be
this principle. Therefore, we must hold the Formula of Universal Law to be
binding on us.
Actually Allison™s argument would do more than ¬ll the gap in the deriva-
tion. Filling the gap would amount to showing that if there is a supreme
principle of morality, then it is the Formula of Universal Law. However, sup-
pose we accept the background assumptions of Allison™s argument, that we
are transcendentally free Kantian rational agents. If successful, his argument
would then establish that the Formula of Universal Law is actually binding
on us.
The chapter begins by exploring the basis of the argument, which is
not merely the assumption of transcendental freedom, but a thick Kantian
account of rational agency (section 2.2). It then considers the ¬rst two steps
of the argument, especially the notion that, as rational beings, we require
some nonsensuously based justi¬cation of our maxims (2.3). The bulk of
the chapter (2.4“5) concerns steps 3 and 4 of the argument, namely the
plausibility of Allison™s claims that our maxims can be justi¬ed only by appeal
to some universally and unconditionally valid practical principle, and that
only the Formula of Universal Law could stand as such a principle.


2.2 A Thick Account of Kantian Rational Agency
Allison bases his argument in a thick account of Kantian rational agency,
three features of which require our attention. The ¬rst is what Allison calls
Transcendental Freedom to the Rescue? 35

the “justi¬cation requirement,” which we have already mentioned (section
1.5). On Kant™s view, says Allison, rational agents must hold that their maxims
are subject to “criteria of reasonableness.”3 At the time they act, the agents
must regard their maxims as “in some sense rationally justi¬able,” whether
it be morally or prudentially.4
Second, the “central insight” of Kant™s theory of rational agency lies in
what Allison calls the “Incorporation Thesis,” a thesis we mentioned earlier
(1.5).5 In Allison™s words, this thesis implies that “inclinations or desires do
not of themselves constitute a suf¬cient reason to act but do so only insofar
as they are “taken up” or “incorporated” into a maxim by the agent.”6 So,
in effect, rational agents never act on brute desires, but they sometimes do
act on self-given rules in which they take account of their desires. To revisit
one of Allison™s examples, the mere presence of a strong desire to eat an
ice cream cone cannot itself give me (a rational agent) suf¬cient reason
to eat one. I can have suf¬cient reason to eat one only if I commit myself
to a rule (a maxim), for example, one permitting me to indulge myself
in this way under certain circumstances.7 Actually, when fully expressed,
the Incorporation Thesis says more, namely that no incentive, including the
moral law, itself constitutes a suf¬cient reason to act unless it is incorporated
into a maxim by the agent.8
In the context of the Incorporation Thesis, “reason” is a translation of
“determining ground,” which, I suggested (1.5), should be understood as
motivating reason or, simply, motive. Allison seems to agree with this view.9
So the Incorporation Thesis says that no incentive can itself constitute a
suf¬cient motive for an agent to act. According to the thesis, a rational agent
simply cannot act on inclinations alone. Whenever an inclination constitutes
a motive of her action, she has incorporated it into some self-given rule.
Returning to the Incorporation Thesis itself, Allison argues that in Kant™s
account an agent must view his incorporating inclinations into self-given
rules as an act of spontaneity on his part.10 The agent must view his “taking
up” a desire into his maxim (or, presumably, his refraining from doing so) as
an act done independently of the causality of nature, that is, as an exercise
of transcendental freedom.11
Transcendental freedom is “independence from everything empirical
and so from nature generally” (KpV 97). According to Allison, a transcen-
dentally free act would be one that was not causally necessitated by preceding
events in time. However, he argues that there is more to transcendental free-
dom than that. If this were all there were to it, then an agent would have
such freedom simply by virtue of possessing a capacity to make causally un-
necessitated choices of means to ends that were themselves foisted upon
her by nature. For example, suppose that the end of experiencing plea-
sure was one she, by nature, always necessarily pursued. She would count as
transcendentally free if she could, without being causally necessitated to do
so, choose between different means of pursuing pleasure.12 To block such
possibilities, in which the scope of transcendental freedom is restricted to
Search for the Supreme Principle of Morality
36

acts of choosing means to ends given by nature, Allison insists that transcen-
dental freedom also involves what he calls “motivational independence.” He
de¬nes this independence as “a capacity to recognize and be motivated by
reasons to act that do not stem, even indirectly, from the agent™s sensuous
nature.”13 For Allison, an agent sees herself as transcendentally free only
if she takes herself to be able to act for reasons that do not stem from her
inclinations or desires at all.
In sum, at the basis of Allison™s argument lies a threefold conception of
rational agency. A rational agent must (1) view the maxims on which he acts
as justi¬ed, (2) hold that he can act on an incentive only if he incorporates
it into a maxim, and (since he holds 2) (3) regard his act of incorporation
as an exercise of transcendental freedom.
Of course, one might question each of these claims. For example, one
might ¬nd problematic the notion that accepting the second requires ac-
cepting the third. Suppose I agree that it belongs to the core of my rational
agency that I cannot act on my inclinations directly but only through maxims
in which I take account of them. Why must I look upon my act of incorpo-
rating an inclination into a maxim as an exercise of transcendental freedom
instead of an event necessitated by natural causes?14 At bottom, Allison re-
sponds that, if I looked upon my act of incorporation as an event necessitated
by natural causes or always motivated by sensuously based reasons, then I
would not be considering myself as a rational agent. In other words, there is
a conceptual connection between seeing oneself as exercising the capacity
to take up desires into self-given rules and seeing oneself as transcenden-
tally free.15 Allison himself acknowledges that this response is not likely to
satisfy his naturalizing critics. In any case, I do not explore this issue here.
My aim is to determine whether the argument succeeds in light of the as-
sumption of transcendental freedom and a robust Kantian view of rational
agency.


2.3 Desire and Justi¬cation of Action
Let us now consider Allison™s argument. According to step 1, assuming that
we (i.e., rational agents) view ourselves as transcendentally free, we must
hold that our inclinations and desires in themselves fail to constitute suf¬-
cient reasons (in the sense of motives) for acting. Since we are assuming that
the Incorporation Thesis is true, this step is unproblematic. For, according
to this thesis, it is essential to being a rational agent that one™s inclinations
or desires do not of themselves constitute a suf¬cient motive to act. A ratio-
nal agent can act on her desires only when she has incorporated them into
some self-given rule.
As we have noted, Allison attributes to Kant a “justi¬cation requirement”
on an agent™s adopting a maxim on which to act (and thus on his acting).
This is the requirement that the agent regard his maxim as, in some sense,
Transcendental Freedom to the Rescue? 37

rationally justi¬ed. Step 2 of the argument applies this requirement to
desire-based action. According to this step, given that we take ourselves
to be transcendentally free rational agents, we must acknowledge that we
require some nonsensuously based justi¬cation for our adopting a maxim
on which to act (and thus for our acting). “[D]esire-based action requires a
desire-independent warrant.”16 In other words, an appeal to desires alone
does not constitute a suf¬cient justi¬cation for our maxims.
Step 2 does not follow trivially from step 1. Suppose an agent adopts a
maxim in which she takes account of a strong inclination she has. Step 1
does not itself rule out the legitimacy of her justifying her adopting this
maxim simply by appealing to the fact that she has this strong inclination.
To return to the earlier example, an agent might have a strong inclination
to eat ice cream. Granted, as step 1 indicates, her having this desire could
constitute a motivating reason for her to act only if she committed herself
to some rule allowing herself to indulge an inclination for ice cream under
certain circumstances. However, the question remains as to whether she
could justify her committing herself to such a rule simply by appealing to
the notion that she has a strong desire for ice cream. In other words, step 1
leaves it an open question whether an agent would be able to justify her act
of incorporation simply on a sensuous basis.
Allison himself seems to recognize this, for he offers an argument that is
supposed to lead us from step 1 to step 2:

[T]his conception of transcendental freedom has important implications for the
justi¬cation of maxims. This is because, assuming motivational independence, the
ground of the selection of a maxim can never be located in an impulse, instinct or
anything “natural”; rather, it must always be sought in a higher-order maxim and,
therefore, in an act of freedom. Consequently, even if one assumes the existence of
a natural drive such as self-preservation, a transcendentally free agent is capable of
selecting maxims that run directly counter to its dictates. And from this it follows that
the mere presence of a drive or inclination does not of itself constitute a suf¬cient
or justifying reason for acting on the basis of it.17

In the second sentence, Allison claims that, if an agent has motivational
independence, then his justi¬cation of his choice of maxim can never be
located in any impulse or inclination. It may be unclear, however, why one
should accept this claim. Allison de¬nes motivational independence as a
capacity to recognize and be motivated by reasons to act that do not stem,
even indirectly, from the agent™s sensuous nature. Why does an agent™s hav-
ing this capacity entail that, rationally speaking, he cannot justify his choice
of maxim simply with an appeal to sensuously based reasons?
Allison suggests a response in the latter half of the quoted passage:
“[E]ven if one assumes the existence of a natural drive such as self-preserva-
tion, a transcendentally free agent is capable of selecting maxims that run
directly counter to its dictates. And from this it follows that the mere presence
Search for the Supreme Principle of Morality
38

of a drive or inclination does not of itself constitute a suf¬cient or justifying
reason for acting on the basis of it.” Yet I ¬nd his reasoning puzzling. First,
Allison™s argument threatens to prove too much. Allison seems to reason
that it is because we can act counter to our drives that these drives alone
are not justifying reasons for our actions. But we can also act counter to the
moral law. By the same reasoning, we should, it seems, conclude that the
moral law is not a justifying reason for our action. Second, it is not obvious
that a (transcendentally free) agent™s being capable of choosing a maxim
that goes directly against an inclination precludes him from justifying his
action by an appeal to his having this inclination. Suppose someone is in
excruciating physical pain and has an inclination for it to stop. Granted,
as a transcendentally free agent, he is capable of choosing and acting on a
maxim of enduring all pain as long as it lasts, a maxim that would preclude
him from acting on his inclination by, let™s say, taking morphine. But why
does the agent™s being capable of acting on such a maxim prevent him from
acting on a different one, that is, a maxim of trying to relieve excruciating
pain when it occurs, and, moreover, justifying his choice of this other maxim
simply by appealing to the fact that he wants this excruciating pain to stop?
Even if the particular argument Allison suggests is problematic, there
are familiar Kantian grounds for embracing step 2 of Allison™s argument.
Following Thomas Nagel, for example, we might argue that in undertaking
to justify an action, an agent must remove herself from the purely inter-
nal perspective and consider her action from an “external” point of view.18
From this point of view, her action is the action of “this person.” To justify
the action, she would have to justify the action of anyone in the same circum-
stances. Yet to justify what anyone would do in the same circumstances, an
appeal to the fact that the agent had a certain desire would not be enough.
She would need to appeal to some principle, for example, the principle
that it is morally permissible for anyone to act on a certain kind of desire
in certain circumstances. For example, suppose I do have an inclination to
be rid of my excruciating pain, and I believe I can rid myself of it by taking
morphine. Asking myself whether I am justi¬ed in acting on my inclina-
tion for my pain to stop requires me to distance myself from my particular
inclination. It makes me see my inclination as that of a person in a certain
situation. To justify my taking the morphine, I would have to justify any
person™s doing so in the same circumstances. Yet appealing to the fact that
I now have this inclination to get out of pain would not accomplish this.
To justify anyone™s taking the morphine in the same circumstances, I would
have to appeal to some principle, one such as: it is morally permissible for
anyone in excruciating pain to take measures to stop it, if these measures
do not cause anyone else comparable pain.
Of course, this argument, especially when sketched so broadly, does
not leave an opponent of step 2 without recourse. He might, for exam-
ple, demand an explanation of why, precisely, he must agree that to justify
Transcendental Freedom to the Rescue? 39

his action, an agent must always take an “external” perspective toward it.
However, pursuing this discussion would lead us far a¬eld, and I hope I may
be excused for not doing so here. I propose that we simply accept step 2
of Allison™s argument. An appeal to desires alone cannot justify an agent™s
maxims. Principles are required not only for the motivation, but for the
justi¬cation, of our acting.


2.4 Practical Law and Justi¬cation of Action
According to step 3, the justi¬cation of our maxims must lie in their con-
formity to some “universally and unconditionally valid practical principle” “
some practical law.19 Only our maxims™ conformity to a practical law could
justify them. Acknowledging that Kant does not explicitly defend this
premise, Allison constructs a Kantian argument for it.20
The argument unfolds in two stages. In the ¬rst, Allison claims that a suf-
¬cient condition for a maxim™s being justi¬ed is that it be adopted because
it is required by some practical law. As Allison suggests, this claim seems rel-
atively unproblematic: “[I]f my reason for x-ing is that it is dictated by such
a law . . . then I have all justi¬cation I could conceivably need for x-ing.”21 In
the second stage of the argument, Allison defends the view that a necessary
condition for a maxim™s being justi¬ed is that the maxim conform to some
practical law. This far more controversial stage of the argument deserves a
careful look.
Allison begins by observing that a transcendentally free rational agent,
as we are assuming ourselves to be, cannot take his maxims to be justi¬ed
unless he holds them to be permissible “ not contramanded by whatever
principle serves as their standard of justi¬cation. Allison then tries to show
that such an agent must hold the following: a necessary condition of his
maxims being permissible is that they conform to some practical law. In
other words, only one kind of standard for the permissibility of maxims will
do: a universally and unconditionally valid practical principle.
This latter move requires scrutiny. In defense of it, Allison points out
that a standard by which we could determine the permissibility of any of
our maxims would have to be “governing the pursuit of any end at all,
including desire- or interest-based ends.”22 If a standard of permissibility
applied only to certain ends, then the standard would not apply to some
(possible) maxims. For every maxim contains, if only implicitly, a description
of an end.23 So, for example, a standard such as “If you want to maximize
your life-span, you ought to do x, y, and z” would not do. For it would not
be a basis on which to determine the permissibility of a maxim that does
not (even implicitly) have maximizing one™s life-span as its end. And we, as
Kantian rational agents, can, of course, have such a maxim. Next, Allison
argues that, since an adequate principle for determining the permissibility
of our maxims would have to be one governing the pursuit of any end at all,
Search for the Supreme Principle of Morality
40

“it must not only apply to all transcendentally free rational agents, it must
also apply to them regardless of what desires or interests they may happen
to have.”24 And, concludes Allison, such a principle is precisely what Kant
means by a practical law.
This argument is, I believe, unsatisfactory. As Allison suggests, Kant does
conceive of a practical law as a principle that applies to all transcendentally
free rational agents. However, Allison™s argument falls short of showing that
the standard for the permissibility of our maxims “ that is, the maxims
of human rational agents “ must apply to all transcendentally free rational
agents. Therefore, the argument does not establish that this standard must
be a practical law.
An example helps to illustrate this point. Consider the following prin-
ciple of happiness, PH: “Always do what you believe will maximize your
own pleasure.”25 Let us suppose that PH is a categorical imperative in the
following sense: it is a principle that commands us to act in a certain way re-
gardless of what we do or might desire. PH prescribes that an agent do what
he believes will maximize his pleasure even if he does not want to maximize
it. Strictly speaking, PH is not a viable candidate for a practical law. There
might be rational agents who, because of their constitution, cannot have
pleasure. Speaking of the pleasure (or pain) experienced by these agents
would be akin to speaking of the pleasure (or pain) experienced by the
number 3: non-sense. PH could not be a practical law, for it could not serve
as the principle by which to determine the permissibility of these agents™
maxims.
Allison™s argument, however, does not close the possibility that PH could
serve as the principle by which to determine the permissibility of our maxims.
As Allison points out, this principle would have to govern the pursuit of any
end we could have, including desire- or interest-based ends. But PH could do
so. According to PH, if an agent believes that adopting a particular maxim
would enable him to maximize his pleasure, then his adopting it will be
permissible; if he does not believe this, then his adopting it will be imper-
missible. This standard would be in effect no matter what the ends described
in the agent™s maxims might be, even if they include minimizing his own plea-
sure. Even though PH is not a practical law, it could serve as the principle
by which the permissibility of our maxims is determined.
Allison might respond that we ¬nd in step 2 of his argument the key to
seeing why PH could not serve as this principle. In step 2 he has established
that we require a nonsensuously based justi¬cation of our maxims. For the
same reasons we require this, we also require a nonsensuously based stan-
dard of our maxims™ permissibility. Since PH is sensuously based, concludes
the response, it could not be such a standard. Once again, “desire-based
action requires a desire-independent warrant.”26
This response is ineffective. What step 2 actually establishes is that an
agent cannot justify his maxims simply by appealing to the notion that he has
Transcendental Freedom to the Rescue? 41

some particular desire or impulse. To justify them, he must appeal to some
principle. Likewise, let us grant, the standard of whether an agent™s maxims
themselves are permissible cannot simply be whether they further some
particular desire or impulse he has. To determine his maxims™ permissibility,
he must appeal to some principle. The important point here is that PH is
not some desire or impulse (or the mere notion that one has some desire
or impulse). It is a principle. It gives some maxims a “desire-independent
warrant” “ a warrant no matter what an agent actually desires. PH does not
prescribe that we maximize our pleasure on condition that we want to. It is,
in the sense already described, a categorical imperative.
Nevertheless, Allison might insist that PH is sensuously based in that it
invokes a concept of something sensuous, namely pleasure. I agree; if one
wants to use the designation “sensuously based” in this way, then PH is
sensuously based. However, Allison has not shown that no principle that
invokes the concept of something sensuous could serve as the standard of
the permissibility of our (rational human agents™) maxims.
Let me put the general point I am making in a different way. Allison is
correct in thinking that a principle cannot serve as the standard for the
permissibility of each and every one of our ( human rational agents™) max-
ims unless it applies to us regardless of which desires and interests we may
happen to have. A principle cannot serve as such a standard for our maxims
unless it applies to us unconditionally. However, Allison holds (in my view
wrongly) that if a principle applies to us unconditionally, it must therefore
apply to all rational agents as well. Following Kant, he makes an unwarranted
move from the requirement that a principle that serves as the standard
for the permissibility of maxims be unconditional ( binding on us regard-
less of what we might desire) to the further requirement that such a prin-
ciple™s scope extend to all rational beings.27 Since PH fails to ful¬ll the
¬rst requirement, it could not be a practical law. For all Allison has said,
however, it still might serve as the principle by reference to which we, tran-
scendentally free human agents, are to determine the permissibility of our
maxims.
My point here is not to defend PH as a candidate for the supreme principle
of morality. I mention it merely as an illustration of a principle that Allison™s
argument here does not rule out as a possible justi¬catory basis for our
maxims. Just as his defense of step 3 does not exclude PH as a justi¬catory
basis for our ( human beings™) maxims, it also fails to bar a principle such
as “Perfect your human rational and physical capacities,” if we understand
the principle as an unconditional command. Although this perfectionist
principle is not a viable candidate for a practical law “ it would obviously
not provide a basis for nonhuman rational agents to judge the permissibility
of their maxims “ it might, nevertheless, serve as a basis for us to judge the
permissibility of our maxims. In general, if I am correct, Allison has not
established that the scope of whatever principle serves as the basis for the
Search for the Supreme Principle of Morality
42

justi¬cation of our maxims must extend to all rational beings. Therefore,
he has not proved step 3 “ that this principle needs to be a practical law.


2.5 Practical Law and the Formula of Universal Law
In step 3, Allison tries to establish that our maxims are justi¬ed only when we
adopt them on the basis of their conformity to some practical law. According
to step 4, this practical law must be the “moral law,” speci¬cally Kant™s For-
mula of Universal Law. To establish this, Allison must obviously show that no
principle besides the Formula of Universal Law is capable of justifying our
maxims. Allison™s argument for step 4 has two main parts, which we need to
examine in turn.
The ¬rst part is relatively uncontroversial. For a practical law to serve
an agent as a justi¬cation of her adopting a particular maxim, he suggests,
it does not suf¬ce that the maxim conform to the law.28 Suppose, for a
moment, that the Formula of Universal Law is the only practical law. An
egoist might act on a maxim of keeping her promises, yet do so in order
to secure a good reputation and, ultimately, simply to promote her own
happiness.29 She might contend that her maxim was justi¬ed simply by
virtue of its conforming to the Formula of Universal Law. But this contention
would not withstand scrutiny. For a maxim to be justi¬ed by reference to
a practical law, it must noncontingently conform to the law. Yet, Allison
seems to suggest, the only way a maxim can noncontingently conform to a
practical law is if it is adopted at least in part because it conforms to the law.
Although the egoist™s maxim conforms to a practical law, it does so merely
as a result of a coincidence of the law™s dictates and her ultimate end. If
in different circumstances the egoist believes that the best way to promote
her happiness is to break her promises, then she will do that. According to
Allison, a justi¬ed maxim is one that an agent has adopted ultimately at least
in part because it conforms to a practical law.
But when does a maxim count as having been adopted ultimately at least
in part because it conforms to a practical law? Allison, it seems, recognizes
two main possibilities. First, if I treat my maxim™s conformity to a practical
law as a suf¬cient reason for my adopting it, then I have adopted the maxim
at least in part because it conforms to a practical law. Second, if I treat my
maxim™s conformity to a practical law as “an ineliminable component in a
jointly suf¬cient reason,” then I have adopted it in part because it conforms
to a practical law.30 This second kind of scenario arises when my maxim is
based on inclination. Allison appears to be suggesting the following kind of
case. Suppose I adopt a maxim of exercising regularly and that if my maxim
violated the law, then, for that reason, I would refrain from adopting it. In
this case, I count as adopting my maxim ultimately (at least in part) because
of its conformity to a practical law. I count as doing so even if I don™t take the
exercising maxim™s conformity to practical law as in itself suf¬cient for my
Transcendental Freedom to the Rescue? 43

adopting it, that is, even if I would not adopt it unless I had an inclination
to stay in shape.
Now the question arises: why must a justi¬ed maxim be adopted by virtue
of its conformity to the Formula of Universal Law, instead of some other
principle? If Allison answers this question, it will be in the second part of his
argument. He moves very quickly:

[I]f I am required to adopt a maxim at least in part because of its conformity to
universal law or, equivalently, an unconditional practical law, then, clearly, this maxim
must be able to include itself as a “principle establishing universal law,” which is just
to say that the maxim must have what Kant terms “legislative form.” In other words,
the intent expressed in the maxim must be compatible with the putative universal
law produced by the generalization of that maxim. Otherwise, its conformity to such
a law could not possibly be the reason (or even a reason) for adopting the maxim in
the ¬rst place.31

Allison claims that only if an agent™s maxim has legislative form can she
adopt it even in part because of its conformity to practical law. On Allison™s
reading, it seems, a maxim has legislative form just in case an agent can act
on it and will that it become a universal law. In other words, a maxim has
legislative form just in case it passes the test contained in Kant™s Formula of
Universal Law. A rational agent who assumes she is transcendentally free,
says Allison, will ¬nd on re¬‚ection that only if her maxim passes the For-
mula of Universal Law test can she adopt it even in part because of its
conformity to practical law. A necessary condition of her maxim™s being
justi¬ed is thus that the maxim conform to the Formula of Universal Law.
Therefore, Allison seems to hold, the agent must conclude that the law on
the basis of which she adopted any justi¬ed maxim would have to be this
formula.
But why should we agree that only if an agent™s maxim passes the Formula
of Universal Law test can she adopt it even in part because of its conformity
to practical law? In my view, Allison leaves this question unanswered. Let
me illustrate this point with the help of a few examples. First, consider the
principle PRU: “Act only on maxims such that you believe that your acting
on them will maximize the general happiness.” It seems that someone™s
reason for adopting a maxim could be that it conformed with PRU, even if
his maxim failed the Formula of Universal Law test. Take a maxim such as,
“In order to maximize the general happiness, I will devote myself entirely to
the study of ethics.” This maxim fails the Formula of Universal Law test, at
least on a common reading. It is not possible for me, as a rational being, to
act on it and at the same time will that it become a universal law. In willing
the universalization of my maxim, I would be willing a state of affairs in
which no one would develop the skills necessary for maximizing the general
happiness, for example, skills in food production or medicine. I would, in
effect, be willing it to become impossible for me to attain the end described
Search for the Supreme Principle of Morality
44

in my maxim. Despite the maxim™s lack of legislative form, it seems that I
could adopt it because it conforms to PRU.
In response, perhaps Allison would suggest that at this point in his ar-
gument it is already clear that an agent could not justify his adoption of a
maxim by appealing to a principle such as PRU. According to his step 3,
the justi¬cation of a transcendentally free agent™s maxims must lie in their
conformity to some practical law. But, Allison might argue, PRU is simply
not a viable candidate for a practical law. Such a law must apply to all rational
agents “ something that PRU, like PH, fails to do, since presumably there
might be some rational agents (e.g., angels) to whom the concept of happi-
ness fails to apply.
Actually, it is not obvious that PRU fails to apply to all rational agents.
Granted, we cannot presuppose that all rational agents can be happy. Per-
haps angels cannot be. However, it would be mistaken to conclude simply
from this observation that PRU could not apply to all rational agents. That
angels cannot themselves be happy does not entail that they are incapable
of acting on maxims that, they believe, would maximize the general welfare.
They presumably hold that what they do has an impact on agents capable of
various degrees of happiness “ that is, agents like us. If PRU fails to apply to
any rational agents, it fails to apply only to those who both cannot themselves
be happy and who can in no way in¬‚uence the happiness of others. Since it
is possible that such isolated agents exist, it seems that, after all, PRU does
not necessarily apply to all rational agents, and that, therefore, PRU is not a
viable candidate for a practical law. Of course, if, as I have claimed, Allison
fails to establish step 3, and we (transcendentally free human agents) could
justify our maxims with reference to a principle that does not quite have the
scope of a practical law, then this point is moot.
But even if Allison succeeds in showing that a maxim based on PRU could
not both fail to have legislative form and be adopted on the basis of a prac-
tical law, other challenges are not dif¬cult to generate. I will consider two.
First consider the weak principle of universalization WU: “Act only on that
maxim which, when generalized, could be a universal law.” Notice that WU
is not equivalent to the Formula of Universal Law. As Kant suggests in the
Groundwork, a maxim of nonbene¬cence could, when generalized, consti-
tute a universal law (GMS 423). Since a world where no one acted bene¬-
cently is indeed a coherent possibility, acting on a maxim of nonbene¬cence
does not violate WU. On Kant™s view, of course, acting on such a maxim runs
afoul of the Formula of Universal Law. It does so, he thinks, because as a
rational agent it is not possible to act on it and, at the same time, will that its
generalization be a universal law. A maxim of nonbene¬cence does not have
“legislative form.” However, Allison does not explain why an agent could not
adopt a maxim of nonbene¬cence on the basis of its conformity with WU.
He does nothing to rule out the possibility that WU is the practical law on
the basis of which we can justify our maxims. There remains a gap between
Transcendental Freedom to the Rescue? 45

the notion that our maxims must be justi¬ed on the basis of some practical
law and the notion that this law must be the Formula of Universal Law.
A rather bizarre principle can serve as a second illustration of this point,
namely BP: “Act only on that maxim that you cannot, at the same time, will
that it become a universal law.” It is, of course, not hard to ¬nd maxims that
would be in accordance with BP, yet that would not have legislative form.
Once again, the maxim of nonbene¬cence Kant discusses in Groundwork I
¬ts the bill. If the maxim of nonbene¬cence fails the Formula of Universal
Law test (as Kant holds it does), then it passes the BP test. According to
Allison, an agent™s reason for adopting a maxim could not be its conformity
to BP. No one, Allison claims, could adopt a maxim of nonbene¬cence
and have as his reason for doing so the maxim™s conformity with BP. Yet why
couldn™t a transcendentally free agent™s reason for adopting a maxim be that
it conform to BP, rather than to the Formula of Universal Law? BP, we might
specify, has the form of a practical law. It is an unconditional imperative,
commanding that we act in a certain way regardless of what we desire to do.
Moreover, like WU, it is not sensuously based, not even in the minimal sense
of making use of sensuous concepts such as happiness or pleasure.
But surely we need not take BP seriously, one might here interject. For
running maxims through the BP test has obviously counterintuitive impli-
cations. Just to name one, if we rely on BP as a standard, we ¬nd that we
have no duty of bene¬cence, which goes strongly against ordinary moral
thinking. I agree wholeheartedly that BP has counterintuitive implications
(as does WU) and that, therefore, we need not take it seriously as a candi-
date for the supreme principle of morality. My point is simply that Allison™s
argument does not exclude this principle (or WU ) as a basis on which tran-
scendentally free rational agents justify their maxims. And since it does not,
the argument is not a successful derivation.
One might see it as an advantage of Allison™s approach that it does not
rely on Kant™s appeals to ordinary moral reasoning in Groundwork I, for ex-
ample, on his discussion of which actions we take to have moral worth. These
appeals generate considerable controversy “ controversy that it might seem
desirable to avoid. But shortcomings in Allison™s argument suggest that an
examination of ordinary moral reasoning must play a role in any effective
derivation of the Formula of Universal Law. To put it bluntly, Allison™s argu-
ment remains at too abstract a level to be successful. Even if we begin with a
very robust notion of a Kantian agent and, contrary to my objection, grant
him that such an agent (even if he is also human) must justify his maxims
by an appeal to a practical law, we ¬nd that Allison fails to show that this
practical law must be the Formula of Universal Law. There is a gap in this
derivation: one perhaps just as wide as that in Kant™s Groundwork I derivation
(traditionally interpreted).
3

The Derivation of the Formula of Humanity




3.1 Outline of the Derivation
On the received view, Kant™s derivation of the Formula of Universal Law
fails, and, if I am correct, Allison™s attempt to rescue it also falls short. But
the Formula of Universal Law is not the only principle Kant defends. He
also advocates the Formula of Humanity: “So act that you treat humanity,
whether in your own person or in the person of any other, always at the same
time as an end, never merely as a means” (GMS 429, emphasis omitted).
Perhaps the Groundwork derivation of this principle is a success. This chapter
explores whether Kant shows that if there is a supreme principle of morality,
then it is the Formula of Humanity.
From the outset, we should keep in mind that Kant employs “humanity”
in a somewhat technical sense. The term does not refer to the class of human
beings but rather to a set of capacities. In the Metaphysics of Morals, Kant tells
us that “the capacity to set oneself an end “ any end whatsoever “ is what
characterizes humanity (as distinguished from animality)” (MS 392). So
at the very least, to have humanity involves having the capacity to set ends.
Kant, it seems, uses “humanity” interchangeably with “rational nature” (e.g.,
GMS 439). In doing so, he suggests that to have humanity is to have certain
rational capacities. Indeed, for Kant the capacity to set ends is a rational
capacity. An agent sets herself an end through adopting a rule that speci¬es
the end, as well as means to take to it in certain circumstances; and adopting
such a rule is an exercise of reason (see MS 395). Unfortunately, Kant does
not offer a list of precisely which rational capacities belong to humanity
(rational nature). But we will, I believe, not go astray if we take the central
ones to be the capacity to set oneself ends and the capacity to adopt and act
on rules, including rules of prudence (hypothetical imperatives) and rules
of morality (categorical imperatives).1 This set of capacities is neither one
that is (necessarily) possessed only by human beings, nor is it one that is
possessed by all human beings. Nonhuman beings (e.g., extraterrestrials)

46
The Formula of Humanity 47

could have humanity and some human beings (e.g., ones in a persistent
vegetative state) presumably lack it.
Let me now offer a brief sketch of Kant™s derivation of the Formula of
Humanity (GMS 428“429). On Kant™s basic concept, the supreme principle
of morality would have to be a categorical imperative, that is, a principle
binding on us no matter what our particular inclinations might be. The
derivation takes shape against the background of this fundamental tenet.
First, Kant contends that if there is a supreme principle of morality (and thus
a categorical imperative), then there is an objective end, something that is
unconditionally good. Second, Kant claims that this unconditionally good
thing would have to be humanity. In his view, therefore, if there is a supreme
principle of morality, then humanity is unconditionally good. But, third, if
humanity is unconditionally good, then we must always treat it not merely
as a means but also as an end. Therefore, if there is a supreme principle of
morality, then we ought to do just what the Formula of Humanity says. So
the supreme principle of morality, if there is one, must be this formula, or
at least something equivalent to it.
This chapter focuses on the derivation™s ¬rst two steps. The chapter begins
(section 3.2) by examining (and criticizing) Kant™s attempt to show that
assuming there to be a categorical imperative requires us to take something
to be unconditionally good. Next (3.3) it explores Kant™s discussion of the
claim that this something must be humanity. Since, I suggest, this discussion
is not an adequate basis for embracing the claim, we need to reconstruct
Kant™s argument for it. The most forceful reconstruction has been offered
by Christine Korsgaard, and the bulk of this chapter (3.4“6) concerns it.
In my view, Korsgaard™s reconstruction does not succeed. I explain why in
section 3.7.
Since I argue that the ¬rst two steps of the derivation fail, I do not here
examine the third. I do not criticize the claim that if humanity is uncondi-
tionally good, then we must always treat it not merely as a means but also as
an end. Yet I do not wish to suggest that this claim is unproblematic. In 3.8,
I mention a possible dif¬culty concerning it.
As I understand it, Kant™s derivation of the Formula of Humanity is not in
any obvious way predicated on the success of his derivation of the Formula
of Universal Law. However, I argue that the ¬rst step of the former derivation
falls prey to a serious objection unless the success of the latter is assumed.


3.2 The Supreme Principle of Morality and Unconditional Value
In his initial step in the derivation of the Formula of Humanity, Kant claims
that if there is a supreme principle of morality (and thus a categorical imper-
ative), then there is something of absolute worth. In something of absolute
worth alone “would lie the ground of a possible categorical imperative”
(GMS 428), and “if all worth were conditional and therefore contingent,
Search for the Supreme Principle of Morality
48

then no supreme practical principle for reason could be found anywhere”
(GMS 428). But why couldn™t a principle be unconditionally binding on
us if nothing was unconditionally good? How does Kant tie the notion of
unconditional bindingness to that of absolute goodness?
This question is not easy to answer, but this much seems clear. According
to Kant, an agent sets himself to do something “ that is, he determines his will,
on the basis of his idea that doing this thing will enable him to secure some
end. In Kant™s view, all acting has an end (see, e.g., KpV 34). This should
come as no surprise since Kant holds that all acting is acting on a maxim
and that, when fully described, a maxim will contain a description of an
end (see section 1.2). Kant distinguishes between subjective and objective
ends. Objective ends, if there are any, would hold for all rational beings.
The idea of securing them would make available to all rational beings a
suf¬cient ground (motive) for acting. But subjective ends do not give all
rational beings grounds for securing them. These ends are such that their
“mere relation to a specially constituted capacity of desire on the part of the
subject gives them their worth” (GMS 428). Suppose a particular object is
a subjective end. If an agent does not value this object, either in itself or as
a means to something else, then it has no worth to him. And if the object
has no worth to him, intimates Kant, then he does not have a ground to
secure it. For him, it is not an end. Apparently, Kant has the following view:
an agent has a suf¬cient ground to secure an object only if he values it “ or
at least is rationally compelled to value it. In the latter case, the agent is
presumably able, through rational re¬‚ection, to come to value the object,
thereby gaining a suf¬cient ground to secure it.
Against the background of this view, we can reconstruct the basis of Kant™s
claim that if there is a categorical imperative, then there must be an objec-
tive end “ something absolutely valuable. A categorical imperative would be
necessarily binding on all rational agents. But a principle could not be nec-
essarily binding on all rational agents unless each of them necessarily had
a suf¬cient ground (motive) at his or her disposal for obeying it.2 Take us,
human rational agents. To say that a principle is binding on us is to say that
we ought to (i.e., have an obligation to) conform to it. Kant, of course, holds
that, if an agent ought to do something, then she must be able to do it (e.g.,
KrV A 807/B 835; KpV 125, 159). But if an agent did not have a suf¬cient
ground available to her for conforming to a rule, then she might not be able
to conform to it. Thus, if not all rational agents necessarily have a ground for
obeying a principle, then it cannot be a categorical imperative. For Kant,
as we noted, to have a ground for doing something an agent must hold
(or be rationally compelled to hold) the action or its effects to be valuable.
Therefore, Kant seems to conclude, if there is a categorical imperative, then
there must be something that everyone holds (or must hold) to be valuable:
an objective end. There must be something that everyone, in every context,
is rationally committed to valuing: something that is absolutely valuable or,
equivalently, unconditionally good.
The Formula of Humanity 49

A brief discussion of how Kant conceives of unconditional goodness will
put us in position to see that this argument is problematic. For something to
be unconditionally good, it must (obviously) be good under every possible
condition, in every possible context. Moreover, if something is uncondition-
ally good, then it is good from the perspective of an impartial rational spec-
tator, or so Kant makes clear at the beginning of Groundwork I. He dismisses
the notion that happiness is unconditionally good thus: “[A]n impartial ra-
tional spectator can take no delight in seeing the uninterrupted prosperity
of a being graced with no feature of a pure and good will, so that a good
will seems to constitute the indispensable condition even of worthiness to
be happy” (GMS 393). In contemporary terms, we might say that for Kant
unconditional value is agent-neutral value. But now the question arises: to
hold a principle to be a categorical imperative, must an agent really main-
tain that there is something unconditionally good in Kant™s sense? Suppose
that Fred holds the following to be a categorical imperative: PW, that is,
“Maximize your power over rational beings.” In Fred™s view, PW is binding
on all agents, no matter what they might desire. Given that Fred holds PW
to be a categorical imperative, he must hold that each agent is rationally
committed to the view that obeying PW is always good for her (or necessar-
ily enables her to promote something good for her). Fred might conclude
the following: he always has available a suf¬cient ground for obeying PW by
virtue of being rationally compelled to take his having maximum power over
rational beings to be good for him, whereas another person, b, always has
available a suf¬cient ground for obeying PW by virtue of being rationally
compelled to take b™s having maximum power over rational beings to be
good for b, while yet another, c, always has such a ground by virtue of being
thus compelled to take c™s having such power to be good for c, and so forth.
There does not seem to be anything self-contradictory in Fred™s conception
of PW as a categorical imperative. Yet Fred does not hold his own power
(or anything else) to be unconditionally valuable, at least on Kant™s con-
ception of unconditional value as agent-neutral value. He has no particular
commitment regarding whether an impartial spectator would take his hav-
ing maximum power to be good; for all he knows, such a spectator would
be totally indifferent to this possibility. Although a categorical imperative
requires a ground, it does not seem as if this ground must be something
unconditionally good.
Perhaps this objection would not worry Kant. At the point in the Ground-
work at which Kant derives the Formula of Humanity, he has already com-
pleted his derivation of the Formula of Universal Law. He has, he believes,
proved that, if there is a supreme principle of morality, then it is the Formula
of Universal Law (or something equivalent to it). So, from his perspective,
any principle conformity to which would require violating the Formula of
Universal Law is thereby eliminated as a candidate for the supreme principle
of morality. (If conforming to a principle requires violating the Formula of
Universal Law, then the principle is obviously not equivalent to the Formula
Search for the Supreme Principle of Morality
50

of Universal Law.) In response to the objection, Kant might say that, even if
he granted that the ground of a categorical imperative need not be some-
thing unconditionally good, it would be the burden of his opponent to show
that a principle that had a different ground could be compatible with the
Formula of Universal Law. Perhaps one could successfully assume this bur-
den, but I have yet to discover how. In acting on PW as Fred would do, one
obviously might be violating the Formula of Universal Law, for example, by
making a false promise. Given the context of the derivation of the Formula
of Humanity, it makes sense that Kant would not address the objection we
have raised.
However, we need to take the objection seriously. The response that, I
have suggested, he might offer turns on the assumption that the derivation of
the Formula of Universal Law has been successful. But this is obviously not an
assumption that we are in a position to make. Does Kant have the resources
to meet the objection without relying on such a robust assumption?
To reply to this objection, one might appeal to some of Kant™s remarks
in the second Critique. Kant distinguishes between well-being (das Wohl )
and good (das Gute), and ill-being and evil. Well-being and ill-being refer
to a person™s “state of feeling” (KpV 60). A scoundrel who provokes an
innocent person and gets a thrashing for it experiences an ill. But, says
Kant, “everyone would approve of it and take it as good in itself even if
nothing further resulted from it” (KpV 61). He goes on to say: “What we
are to call good must be an object of the capacity of desire in the judgment
of every rational human being, and evil an object of aversion in the eyes of
everyone; hence for this appraisal reason is needed, in addition to sense”
(KpV 60“61). For us legitimately to hold an object to be good (as opposed to
conducive to our own well-being), it must be something that each rational
agent judges, or at least should judge, to be worth bringing about “ to be
desirable. If this is correct, then Fred™s conception of the “ground” of PW
is ¬‚awed. According to Fred, each agent is rationally compelled to take her
own power to be good for her, and thereby always has available to her a
suf¬cient motive to conform to PW. However, there is obviously no reason
to assume that each agent does in fact judge other agents™ having maximum
power over rational agents to be desirable. Nor does it seem that each agent
should judge this to be desirable. After all, another agent™s maximizing her
power might prevent him from maximizing his own power, that is, from
securing what he must (in his view) take to be valuable. From this we can
see that Fred cannot, rationally speaking, take his maximum power over
rational beings to be good. Since he cannot, he is obviously not rationally
compelled always to take it to be good. Therefore, Fred does not necessarily
have a motive available to him for conforming to PW. (Although Fred might
have an inclination to maximize his power now, there is no guarantee that he
will want to do so at other times.) Despite initial appearances, Fred cannot
really hold PW to be a categorical imperative.
The Formula of Humanity 51

Another passage in the second Critique suggests a related response to the
objection. According to Kant, “a law, as objective, must contain the very same
determining ground of the will in all cases and for all rational beings” (KpV 25).
Kant appeals to this notion in order to show that a principle of happiness
cannot be a practical law. But he might just as well appeal to it to show that
each agent™s notion that his own power is necessarily good could not serve as
a ground for PW to be a practical law. For Kant here suggests that each agent
must have the very same motivating reason available to him for conforming
to a rule, if this rule is to be a practical law. And, on Fred™s own conception,
not every agent has the very same motivating reason available to him for
maximizing his own power. Fred has a ground to maximize his power in
that he is rationally compelled to take his power to be good; another agent
has a ground to maximize her power in that she is rationally compelled to
take her power to be good, and so forth. These are obviously not the very
same grounds. (Fred™s ground for maximizing his power does not lie, for
example, in his being rationally compelled to take someone else™s power to
be good.) If grounds such as these are all the ones agents have available for
conforming to PW, they do not suf¬ce to ground it as a practical law.
These two replies are convincing “ if one accepts their premises. But
I do not think we can credit Kant with proving the premises to be true.
The ¬rst argument rests on the notion that for us legitimately to hold an
object to be good (as opposed to conducive to our own well-being), the
object must be something that each rational agent judges, or at least should
judge, to be desirable. However, consider the stage in Kant™s dialectic at
which he has not successfully derived any of his candidates for the supreme
principle of morality. At this stage, which is after all our stage, I do not ¬nd
in Kant a demonstration of the notion in question. (Kant himself makes
the claim that “what we are to call good must be an object of the capacity
of desire in the judgment of every rational human being” long after he has
completed his second Critique derivation of the Formula of Universal Law.)3
What, precisely, would be irrational in an agent™s calling an object good
(as opposed to conducive to his well-being), even if, in the agent™s view,
not everyone was rationally compelled to judge the object desirable? The
agent might reasonably believe that the object “ for example, her having
maximum power over rational beings “ would actually diminish her well-
being. To put the point in contemporary terms: is it always irrational for an
agent to take an object to be good in an agent-relative sense, that is, good
from her standpoint (though not in terms of her happiness), but not good
in an agent-neutral sense, that is, desirable from an impartial perspective?
On the face of it, this sort of view does not always seem to be irrational.
Suppose that you sacri¬ce your professional ambitions and your ties to your
lover for the sake of your child, who has a painful, though not debilitating,
disease. You take a high-paying job you despise in a new city. Dif¬cult as it is
for you to acknowledge, it turns out that you are even more miserable than
Search for the Supreme Principle of Morality
52

you would have been had you not made the sacri¬ces and your child re-
mained sick. But your extra income allows you to obtain effective treatment
for him. Your child thrives. As it stands, though, if you had donated to relief
organizations the money it took to treat your child, a dozen other children
would have been saved from intense suffering and death. Now the state of
affairs that results from your action is obviously not good in terms of your
well-being. Moreover, it might not be good from an impartial standpoint,
for example, if the impartial standpoint is a utilitarian one. Nevertheless,
it seems that you might, without irrationality, hold the state of affairs to be
good from your standpoint.
According to Amartya Sen, there would not necessarily be anything ir-
rational in your holding this view. He defends the coherence of “evaluator
relativity,” according to which, ultimately, the goodness of a particular state
of affairs depends intrinsically on the position of the evaluator in relation
to the state.4 For Sen, there need be nothing self-contradictory in saying
that the state of affairs we have been discussing is good from the position
you hold but not good from an impartial standpoint. Of course, much dis-
cussion would be needed to defend the notion of evaluator relativity. But
my aim here is modest. I simply want to point out the following. The ¬rst
Kantian reply relies on the view that for us legitimately to hold an object to
be good (as opposed to conducive to our own well-being), the object must
be something that each rational agent judges (or ought to judge) to be
desirable. However, this view is controversial and, so far as I can tell, not one
that Kant bolsters with arguments.
The second reply rests on the premise that each agent must have the very
same motivating reason available to him for conforming to a rule, if this
rule is to be a practical law. But this is not obvious. Let us grant for now
that no rule could be unconditionally binding (and thus a candidate for the
supreme principle of morality) unless everyone necessarily had a suf¬cient
motive available to him for abiding by it. It does not follow from this that
everyone would have to have the very same motive for abiding by it.
One might construe Kant to be suggesting an argument for this view in
the second Critique (KpV 28): suppose that everyone has a motive, but not
the very same motive, for conforming to a principle. In this case, conformity
to the principle would not produce harmony. Consider what would occur if
each agent was motivated by the notion that his own happiness was good to
conform to the principle that one ought to maximize one™s own happiness.
Everyone™s conforming to this principle would surely result in disharmony,
if only because some agent™s promoting his happiness by securing an object
would preclude others from promoting theirs by securing this object. For
another example, look what would happen if each agent had a motive to
conform to PW, and that motive was the notion that maximizing his own
power was good. Each agent™s conforming to PW would bring about dishar-
mony, a vast competition to gain the upper hand. However, the argument
The Formula of Humanity 53

continues, a law, whether it be practical or natural, must produce harmony.
Hence for a practical principle to count as a law, conformity to it must pro-
mote harmony. Yet unless everyone has the very same motive for conforming
to a principle, conformity to it might not promote harmony. Therefore, to
be a practical law, a principle must be such that everyone has the very same
motive for conforming to it.
This argument (whether or not it is really Kant™s) suffers from several
dif¬culties. First, it is questionable whether natural laws necessarily make
everything harmonious. On Kant™s view, such laws presumably govern the
occurrence of earthquakes, droughts, and ¬‚oods. In so doing, are they pro-
moting harmony? From the perspective of those struggling to survive these
events, the answer would seem to be negative. In response, one might claim
that from a god™s-eye view, in governing these events natural laws are promot-
ing (or at least maintaining) harmony, speci¬cally the harmony constituted
by all events being governed by such laws, instead, say, of being random. But
this answer to the ¬rst dif¬culty yields a second one. For it seems that the
scenario in which everyone acted on the principle of maximizing one™s own
happiness (or power) would also produce harmony from a god™s-eye view.
Although, to an individual battling to advance his happiness (or power)
amid others battling to advance theirs, the world might not seem a harmo-
nious place, it would appear as such to someone who stepped back from
the fray to consider that each agent was acting on the same practical prin-
ciple, namely one commanding him always to promote his own happiness
(or power), not on whatever principle he happened to stumble upon.
Let us agree that unless everyone has the very same motive for conform-
ing to a practical principle, universal conformity to it might not, from the
perspective of those doing the conforming, yield harmony. But why should
we accept the notion that to count as a practical law, a principle would have
to yield harmony in this sense? Why isn™t the kind of harmony that would
be manifest from a god™s-eye view enough? One might observe that it would
be unfortunate for us if universal obedience to a practical law failed to yield
harmony from the perspective of those conforming to it. That it would be
unfortunate, however, does not entail that it would be impossible.
Finally, let us grant for the sake of argument that natural laws yield a
kind of harmony that everyone™s obeying the principle of happiness or of
power would fail to yield. For Kant natural laws determine what is; practical
laws determine what ought to be (GMS 387“388, KpV 19“20). Given that
natural laws differ in kind from practical ones, why should we assume that
the latter (when universally obeyed) must share the harmony-promoting
characteristic of the former?5 Kant fails to give us grounds for accepting
his claim that no practical law could promote discord. He fails to prove
the relevance of the question of whether a practical principle, when acted
on universally, would lead to disorder to the question of whether such a
principle has (or could have) status as a practical law.
Search for the Supreme Principle of Morality
54

As far as I can tell, Kant does not offer a convincing argument for the claim
that a practical law (in the sense of a categorical imperative) must be such
that everyone has the very same motive for conforming to it.6 Since he does
not, we are not in position to reject a principle such as PW as a possible
categorical imperative on the basis that it does not “contain the very same
determining ground of the will in all cases and for all rational beings.”
To sum up this section, Kant claims that if there is a categorical impera-
tive, then there is something unconditionally good. His argument for this
claim seems to go as follows. Suppose some principle is a categorical im-
perative. By de¬nition, this principle would be unconditionally binding on
all rational agents. But now suppose that there is nothing unconditionally
good. Some rational agents might ¬nd themselves with insuf¬cient motive
to conform to the principle, and thus might be unable to do so. If some
agents were unable to conform to the principle, then it would not be bind-
ing on them; for an agent does not have a duty to do something that she
cannot do. Contradicting our initial supposition, the principle would not be
a categorical imperative. Kant™s argument turns on the notion that only if an
agent holds there to be something unconditionally good can she maintain
that every agent always has available to her a suf¬cient motive to conform
to a given principle. I have suggested that Kant fails to establish this notion.
He does not successfully block the possibility that an agent can deny there
to be anything unconditionally good (i.e., good in every possible context,
according to an impartial rational spectator), yet at the same time coher-
ently maintain that every agent always has at her disposal a suf¬cient motive
to conform to a given principle.


3.3 The Unconditional Value of Humanity: Kant™s Argument
In the ¬rst step of his argument, Kant tries to show that if there is a categorical
imperative, then there must be some object (or objects) that all rational
agents must hold to be unconditionally good. In the second step, Kant
claims that this something is humanity. If we hold that there is a categorical
imperative, then we must conclude that humanity is unconditionally good.
Kant packs his defense of this claim into one dense and dif¬cult paragraph.7
On its face, the defense seems inadequate. Kant needs to demonstrate
that only humanity could be the “ground” of a categorical imperative. What
Kant does is dismiss three candidates for unconditional goodness: objects
of inclinations; inclinations themselves; and beings “the existence of which
rests not on our will but on nature” (GMS 428), for example, animals. Then,
without further ado, he announces that humanity could be unconditionally
good.
His dismissal of rival candidates for unconditional goodness seems pre-
cipitous. For example, regarding inclinations, Kant merely says: “But the in-
clinations themselves, as sources of needs, are so far from having an absolute
The Formula of Humanity 55

worth, so as to make one wish to have them, that it must instead be the uni-
versal wish of every rational being to be altogether free from them” (GMS
428). Most of us, I venture, are not tempted to the view that our desires
themselves (as opposed, perhaps, to their objects) have an absolute value.
It seems strange, however, to dismiss this view on the grounds that all of us
wish to be altogether free from our desires.
Even if the remarks Kant here makes do eliminate these candidates for
absolute goodness, the question arises as to whether he is entitled to con-
clude that it is humanity he is looking for. After all, might not Kant have
overlooked some other candidate for absolute goodness? What about the
state of affairs of all rational agents being happy? How does Kant dismiss
this possibility? This candidate is not itself an inclination. It need not be
considered an object of an inclination. (We can easily envisage a world in
which no one desires everyone “ including his enemies “ to be happy.) And
everyone™s happiness is not obviously something the existence of which
would rest on nature rather than on our will.
In sum, Kant™s argument that only humanity could be the absolutely
valuable thing required if there is to be a categorical imperative (supreme
principle of morality) appears to suffer from two shortcomings. First, his
arguments against other candidates seem too quick; second, he offers no
grounds for the conclusion that he has considered all other candidates.
On the face of it, Kant™s argument that if we take there to be a categorical
imperative, then we must take humanity to be unconditionally good does
not seem very promising.


3.4 Korsgaard™s Reconstruction: Preliminaries
In suggesting an opposing view, Christine Korsgaard offers an ambitious
and in¬‚uential account of Kant™s argument.8 Since this account is the most
forceful one I am familiar with, I consider it in detail.9 But I do not explore
the extent (if any) to which Korsgaard™s account departs from the letter of
Kant™s text in the Groundwork. My view, which I do not here try to defend, is
that, though the argument Korsgaard presents is available to Kant, it is not in
all of its details one that Kant actually gives.10 It is, I think, a reconstruction
of Groundwork 428“429.
The key concept in Korsgaard™s reconstruction is that of a good end.11
The reconstruction can be divided into two claims, both of which invoke
this concept. First, if we (rational agents) take there to be a categorical
imperative and thus something unconditionally good, then we must hold
ourselves to have good ends. Second, if we take ourselves to have good ends,
then we must hold humanity to be unconditionally good. If these two claims
are successful, then, in effect, a big part of Kant™s argument goes through.
He manages to establish that if we take there to be a categorical imperative,
then we must hold that humanity is unconditionally good.
Search for the Supreme Principle of Morality
56

To evaluate the ¬rst claim, we need to have in view what Korsgaard means
by a good end. She attributes to Kant a robust conception of one. To count as
good, an end must meet the following criteria. First, it must be the object of
rational choice. By this, Korsgaard apparently means that reason must play a
role in the process through which the agent comes to have the end.12 A good
end is not one simply laid down by instinct. It is, rather, one an agent sets
himself after some re¬‚ection on whether it is worthy of pursuit.13 Second,
a good end must provide “reasons for action that apply to every rational
being.”14 This is Korsgaard™s gloss on Kant™s claim that the good “must be
an object of the capacity of desire in the judgment of every rational human
being.” According to Korsgaard, the claim entails that, for Kant, to be good
an end must be one that we can share.15 Third, a good end must be fully
justi¬ed.16 It appears that, for Korsgaard, a fully justi¬ed end would be one
that was either itself unconditionally good or that derived its goodness from
something unconditionally good.
Since the notion of a fully justi¬ed end is crucial to the argument, let us
view in detail Korsgaard™s explication of it:
An end provides the justi¬cation of the means; the means are good if the end is good.
If the end is only conditionally good, it in turn must be justi¬ed. Justi¬cation, like
explanation, seems to give rise to an inde¬nite regress; for any reason offered, we can
always ask why. If complete justi¬cation of an end is to be possible, something must
bring this regress to a stop; there must be something about which it is impossible or
unnecessary to ask why. This will be something unconditionally good. Since what is
unconditionally good will serve as the condition of the value of other good things,
it will be the source of value.17
Justifying a claim that an end is good involves answering the question of
why it is good. Suppose someone asserts that his jogging is good. To justify
this assertion, the person would need to explain why it is good, perhaps by
pointing out that it keeps him in shape. Yet then the question arises: why is it
good for him to be in shape? Perhaps he answers that he feels better when he
is ¬t, an assertion that might, in turn, give rise to a question of why his feeling
better is good, and so forth. In Korsgaard™s view, a full justi¬cation of the
goodness of an end would bring this sort of regress to a close. It would show
us that the goodness of the end depended on the goodness of something
regarding which it would be either impossible or unnecessary to pose the
question: why is this good? In Korsgaard™s view, this special something will
be unconditionally good.
In sum, for an agent to count his end as good, the end must be the object
of his rational choice; provide reasons for action to every rational agent; and
be fully justi¬ed.

3.5 The Supreme Principle of Morality and Good Ends
If an agent holds there to be a categorical imperative, must he hold that he
has at least one end that meets all three criteria? Korsgaard herself explains
The Formula of Humanity 57

what she (and presumably Kant) mean by a good end, and she suggests
that the notion that we have good ends serves as an assumption in Kant™s
derivation of the Formula of Humanity.18 But she does not focus on the
question of whether taking there to be a categorical imperative rationally
compels one to believe that he has some good end(s): at least one end that
meets all of the criteria.
So let us consider this question. Clearly, an agent must have at least one
end that ful¬lls the ¬rst criterion. In Kant™s view, being a rational agent
involves setting ends for oneself. And any end an agent sets for himself
would be an object of his rational choice. For the power of rational choice
(humanity) is the power to set ends (MS 392). Any rational agent, let alone
any agent who holds there to be a categorical imperative, would have to
hold that he has ends that are objects of his rational choice.
Next, does assuming there to be a categorical imperative rationally com-
pel an agent to hold that he has an end that everyone has a reason to
promote? If holding there to be such an imperative required each of us to
maintain that some particular object was unconditionally good, then the
answer would be yes. For in this case, all of us would be rationally compelled
to promote (or perhaps preserve) this unconditionally good thing. Sup-
pose this unconditionally good thing were everyone™s happiness. I would
have reason to further your end of promoting the general welfare, and you
would have reason to further my end of promoting the general welfare and
so forth. However, I have argued that maintaining there to be a categorical
imperative is consistent with denying that there is anything unconditionally
good. By virtue of holding PW to be a categorical imperative, Fred does not
rationally commit himself to the view that everyone has a reason to promote
his (Fred™s) having maximum power over rational beings. But Fred is com-
mitted to the view that everyone has a reason to promote his own power
over such beings. An agent™s assuming that there is an unconditionally and
universally binding practical principle does not entail that he must af¬rm
that there is something unconditionally good that everyone has reason to

<<

. 2
( 8)



>>