<< . .

. 9
( : 27)

. . >>

it, and vice versa. In a similar way (as we have already noted in section 3.2),
all agents pursuing their own happiness would not have a consistent object;
each agent wants his own happiness to be promoted, which, in Kant™s view,
would prevent (at least some) other agent from promoting his. For example,
part of Pete™s happiness would be winning this year™s tournament. Yet if he
wins, then Boris couldn™t be happy, since he was counting on victory as well.
It seems that if an object is consistent and harmonious, then one agent™s
promoting it would not itself preclude any other agent from doing so.
Korsgaard concludes that everyone™s happiness “does not form a consis-
tent harmonious object.”37 This conclusion, however, does not follow from
the understanding we have thus far attained of what it means to form one.
For it is not clear that one agent™s securing everyone™s happiness (if we as-
sume for a moment that it would be practically possible for one agent to do
this) would itself preclude another agent from doing so. One agent (angel 1)
Search for the Supreme Principle of Morality

might initially bring it about that everyone was happy, while another agent
(angel 2) might thwart a threat to everyone™s happiness, for example, a
threat from a natural disaster. So far, everyone™s happiness does seem to be
a consistent, harmonious object. Yet I suspect that there is a further condi-
tion on being such an object that we have not yet captured; a consistent,
harmonious object must be realizable. And, according to Korsgaard, every-
one™s happiness is not. For each person to be happy, each person would
have to have all of his desires satis¬ed. But the satisfaction of some agents™
desires necessarily precludes the satisfaction of some other agents™ desires.
Not everyone can be happy. In short, Korsgaard argues that since uncondi-
tionally good objects must be harmonious, and everyone™s happiness is not
harmonious, everyone™s happiness is not unconditionally good.
Moving forward in the argument, we have assumed that there are good
ends “ good in the very robust sense Korsgaard has speci¬ed. Given that
there are good ends in this sense, we must be able to pinpoint the suf¬cient
condition of their goodness. Yet the question remains: what is it? It will be
helpful to cite the passage in which Korsgaard answers this question:

Now comes the crucial step. Kant™s answer, as I understand him, is that what makes
the object of your rational choice good is that it is the object of a rational choice.
That is, since we still do make choices and have the attitude that what we choose is
good in spite of our incapacity to ¬nd the unconditioned condition of the object™s
goodness in this (empirical) regress upon the conditions, it must be that we are
supposing that rational choice itself makes its object good. His idea is that rational
choice has what I will call a value-conferring status.38

Here Korsgaard seems to be saying: a good end derives its value from being
the object of an agent™s exercise of a certain capacity, namely his power of
rational choice. Suppose an agent takes one of his ends to be good. Upon
re¬‚ection, suggests Korsgaard, he will conclude that it has this status by
virtue of his having exercised his power of rational choice with respect to it.
That the agent, not driven by impulse but rather guided by reason, chose
this end suf¬ces, in the agent™s considered view, to make it good. “We act as
if our own choice were the suf¬cient condition of the goodness of its object:
this attitude is built into (a subjective principle of) rational action.”39 An
agent holds that a suf¬cient condition of the goodness of his good ends is
that they be the object of his rational choice. This is the sixth step of the
regress argument.
How does Korsgaard move from step vi to vii? Suppose an agent embraces
the idea that a suf¬cient condition of the goodness of his good ends is that
they be objects of his rational choice (vi). According to ii, he is then commit-
ted to the view that his rational choice is either itself unconditionally good or
derives its goodness from something unconditionally good. The regressive
argument has the agent af¬rm the latter. He af¬rms that his exercising his
power of rational choice derives its goodness from this power itself, which is
The Formula of Humanity 65

unconditionally good: “[R]egressing upon the conditions, we ¬nd that the
unconditioned condition of the goodness of anything is rational nature, or
the power of rational choice.”40 It is not rational choosing, but the power
of rational choice that the agent holds to be unconditionally good. The ar-
gument™s seventh step ¬nishes the regress on conditions of the goodness of
the agent™s ends. It maintains that an agent must view his power of rational
choice to be unconditionally good.
Korsgaard™s transition from this step to the conclusion that you must hold
everyone™s power of rational choice to be unconditionally good appears to
go as follows. According to step vi, you (an agent who has af¬rmed that she
has good ends) are rationally compelled to view yourself as having “value-
conferring status” in virtue of your power of rational choice. But “[i]f you
view yourself as having a value-conferring status in virtue of your power of
rational choice, you must view anyone who has the power of rational choice
as having, in virtue of that power, a value-conferring status.”41 In short, you
must embrace viii. Moreover, just as your holding yourself to have value-
conferring status requires you to hold your power of rational choice to be
unconditionally good (the move from vi to vii), so your holding others to
have value-conferring status requires you to hold their power of choice to be
unconditionally good. In effect, as step ix states, you must hold everyone™s
power of rational choice (humanity) to be unconditionally good.

3.7 The Failure of the Regressive Argument
I do not believe that this argument succeeds in showing that, if an agent
assumes that he has good ends, then he must hold that humanity is uncon-
ditionally valuable. I try to highlight two problems with the argument.
The ¬rst dif¬culty concerns step v, speci¬cally the denial that a suf¬cient
condition of the goodness of your good ends is that they contribute to every-
one™s happiness. As a basis for this denial, Korsgaard appeals to the notion
that such a condition would have to be unconditionally good. However, she
argues, if something is unconditionally good, then it is a “consistent, har-
monious object,” which entails that it is realizable. But everyone™s happiness
is not realizable, since making some people happy necessarily involves pre-
cluding others from being happy. Therefore, everyone™s happiness is not
unconditionally good.
For the sake of argument, let us grant that if we characterize happiness as
the complete satisfaction of all inclinations, as Kant sometimes does, then
happiness is not a harmonious object.42 Given the con¬‚icting set of desires
people have (and, let™s say, necessarily will have), the happiness of some
would always prevent the happiness of others.43 Yet why should we embrace
this desire-satisfaction account of happiness in the ¬rst place? Philosophers
who argue that everyone™s happiness is unconditionally good need not em-
ploy such an account. They might, rather, invoke a conception according to
Search for the Supreme Principle of Morality

which happiness is a harmonious object. These philosophers might contend,
for example, that being happy amounts to having a number of goods “ for
example, loving relationships, a sense of self-respect, security “ such that one
person™s having them would not preclude anyone else from having them.
In order for her rejection of happiness as unconditionally good to be effec-
tive, Korsgaard must, it seems, show not only that happiness on a Kantian
conception fails to be a harmonious object but also that happiness on other
(plausible) conceptions fails as well.
Korsgaard might here appeal to Kant™s view that the only thing we can
conceive of as good without quali¬cation is a good will. Everyone™s happi-
ness, no matter how we de¬ne it, would not be good without quali¬cation,
Korsgaard might argue. Kant suggests a thought experiment for determin-
ing that something fails to be good without quali¬cation. In it, we ask our-
selves whether, in some possible context, an “impartial rational spectator”
would ¬nd that the thing was not good (GMS 393). If, in our view, there is
such a possible context, then we conclude that the thing is not good without
quali¬cation. According to Kant, the notion that only a good will is uncon-
ditionally good is to be found in the “moral cognition of common human
reason” (GMS 403). Kant defends the notion with an appeal to our every-
day moral intuitions. Korsgaard might claim that there is a possible context
in which an impartial rational spectator would ¬nd that everyone™s being
happy is not good, namely when some happy individuals did not have a good
But what is a good will? Interpreting Kant™s notion (or notions) of a good
will is a challenging task, and I do not attempt to do so thoroughly here.
For our purposes, we can take note of two ways in which Kant seems to
employ “good will,” as it applies to us, agents who can be tempted by their
inclinations to act contrary to the moral law. According to the ¬rst way, a good
will is a particular sort of willing or, what for him amounts to the same thing,
of acting (section 1.4). Kant writes of “the unquali¬ed [uneingeschr¨ nkten]
worth of actions” (GMS 411), presumably of actions done from duty, which
he has previously stated to have “unconditional and moral worth” (GMS
400). Since, according to Kant, the only thing good without quali¬cation
(ohne Einschr¨ nkung) is a good will, it appears that sometimes “good will”
refers to a certain kind of action, that is, that done from duty.45 I call this
usage the “particular action” understanding of a good will.
According to a second way in which Kant employs “good will,” it refers
not to a particular kind of action an agent might perform but rather to a
kind of character she might have. An agent has a good will on this usage just
in case she is committed to doing what duty requires, not just in this or that
particular action, but overall. Presumably if an agent has this commitment,
then she will sometimes act from duty. (For example, she will invoke duty as
her incentive to do what is morally required in cases where she is tempted by
her inclinations to act contrary to what morality demands.) Kant intimates
The Formula of Humanity 67

that having a good will amounts to having a certain kind of character in the
¬rst paragraph of Groundwork I. Right after suggesting that the only thing
good without quali¬cation is a good will, he tells us that certain qualities
of temperament (e.g., courage or resolution) “are undoubtedly good and
desirable for many purposes, but they can also be extremely evil and harmful
if the will which is to make use of these gifts of nature, and whose distinctive
constitution is therefore called character, is not good” (GMS 393). Later Kant
is discussing a man who is by temperament cold and indifferent to others,
but who, from duty, acts bene¬cently. “It is just then,” says Kant, “that the
worth of character comes out, which is moral and incomparably the highest”
(GMS 398“399). These passages suggest that “good will” refers not merely
to a particular kind of action, but to a kind of character that can be expressed
in action. Sometimes Kant employs what I (following Karl Ameriks) call the
“whole character” conception of a good will.46
It appears that Kant employs (at least) two different notions of a good
will. For it seems that one could have a good will on the particular action
understanding, yet not have a good will on the whole character conception.
After all, why could one not act from duty in a particular case, yet not be
committed overall to doing what morality requires? I do not pursue this
question here. For our purposes, it suf¬ces to make clear which of these
notions of a good will we are employing at a given point, leaving aside the
issue of whether, ultimately, they coincide.
Let us now return to the argument we were considering before our brief
discussion of a good will. Although I am not entirely sure, I believe that when
Korsgaard invokes the notion of a good will, she has in view the whole char-
acter conception.47 An impartial rational spectator, Korsgaard might claim,
would not ¬nd everyone™s happiness to be good if some happy individuals
did not have a good will in the sense of an overall commitment to doing what
morality requires. On Kant™s view, of course, if an agent does not have a good
will, then she might not only stray from duty sometimes, but actually make a
habit of doing so. But do we really hold that an impartial rational spectator
would not approve of everyone™s being happy if some did not have a good
will? Some of us might imagine such a spectator reacting to this scenario
as follows: “The agents who do not have a good will do not morally deserve
their happiness. However, this does not mean that everyone™s happiness is
not good. For in the scenario in question, the actions of those without a
good will “ their lying, cheating, and so forth “ do not prevent others from
being happy. Since they do not, the scenario is actually still good. Granted, a
scenario in which everyone is happy but some are without a good will is not
as good as one in which everyone is both happy and has a good will. Yet the
former scenario is still good.” In Kant™s view, of course, this reaction does
not conform to ordinary moral reason. But this view seems dubious.
In any case, an appeal to the thought experiment in question would
be a dangerous tactic for a defender of the regressive argument to take.
Search for the Supreme Principle of Morality

For employing it might show that the power of rational choice is itself not
good without quali¬cation. Recall that the power of rational choice is the
capacity to set all sorts of ends, including, but not limited to, morally good
ones. Consider an agent who has the power of rational choice, but employs
it with no concern for whether he is conforming to moral requirements.
Whenever this agent is inclined to realize a morally bad end, he does all he
can to do so. In this context, which certainly seems to be a possible one,
would an impartial rational spectator judge the power of rational choice to
be good? That she would is, I think, doubtful.48 For it is his power of rational
choice that enables the agent to choose the morally bad ends. If he did not
have this power, yet instead sought merely instinctual grati¬cations, then he
might cause much less harm. It would not be helpful to respond here that
what makes the agent™s power of rational choice unconditionally good is
that by virtue of having it, he has a further capacity, namely that to develop
a good will. For we are imagining a case in which the agent never exercises
his capacity to develop a good will. And in this case, why should we hold this
capacity to be good, rather than, say, indifferent?
This discussion allows us to see in Kant™s doctrine an apparent tension
that I am unsure how to resolve. Kant holds that the good will alone is
good without quali¬cation (GMS 393). He also holds that rational nature is
unconditionally good (GMS 428). So unless I am overlooking some subtle
distinction between being good without quali¬cation and being uncondi-
tionally good, Kant seems to be identifying the good will and rational nature.
But on the conceptions of the good will I sketched above “ that is, the par-
ticular action and whole character conceptions “ it seems that a being could
possess rational nature and yet not have a good will. As we have just seen,
that a being has rational nature does not entail that he ever acts from duty,
let alone that he has committed himself to an overall policy of doing what
duty requires. As Ameriks notes, some philosophers, perhaps based on such
considerations, have attributed to Kant the view that the good will simply
is rational nature.49 But if we substitute this understanding into the begin-
ning of Groundwork I, we get nonsense. Since Kant™s notion of the good will
is problematic, so is appealing to this notion in an effort to rescue step v of
the regressive argument.
The second dif¬culty with the regressive argument concerns step iii. In
steps i and ii, you have assumed that some of your ends are good and that
their goodness must derive from something unconditionally good. Step iii
aims to rule out the possibility that your ends themselves count as this un-
conditionally good thing. Recall our example of the kind of position iii dis-
claims. An environmentalist who has the end of preserving the maximum
number of living species on earth might hold that this end not only meets
all of Korsgaard™s criteria for goodness but is itself unconditionally good.
Korsgaard suggests a Kantian response to this position. The environmen-
talist is mistaken about the source of his end™s goodness, believing that he
The Formula of Humanity 69

wants to maximize species preservation because such preservation is intrinsi-
cally good, when, in reality, any goodness had by species preservation would
actually stem from his desiring it. But given that the goodness of species
preservation derives from the agent™s desire for it, it is not unconditionally
good, since the agent may well cease to desire it.
This argument does not threaten value realists who hold goodness to be
inherent in their ends themselves.50 For Kant does not here establish that
these realists must accede that the goodness of what they take to be un-
conditionally good derives simply from their desiring it. Why, for example,
must the environmentalist agree that the goodness of species preservation
depends on his wanting it? Why can he not maintain that species preserva-
tion is unconditionally good, and thus good regardless of whether he (or
anyone else) desires it?
It is once again open to Kant to appeal to his claim that the only thing good
without quali¬cation is a good will. Such an appeal might be more effective
here than it was in eliminating the possibility that everyone™s happiness was
unconditionally good. Would an impartial rational spectator hold that the
maximum number of currently existing species being preserved was good
in every context?
There are some contexts in which such preservation would have what (we
might plausibly think) the spectator would take to be bad effects. For ex-
ample, in environmentally sensitive areas, preserving species might require
closing businesses and thus causing hardship to workers and their families.
However, that species being preserved would in some contexts have bad ef-
fects does not itself preclude it from being unconditionally good. As many
commentators have remarked, a good will can also have what (we might
plausibly think) a rational spectator would consider to be bad effects. Some-
one with a good will might be, as it were, cursed. When she acts from duty,
she might not only fail to realize her ends but, by a “special disfavor of for-
tune,” bring about the opposite of her aim. For example, her effort to save
a choking victim might actually result in his death. Or her large donation to
an emergency relief fund might end up in the hands of terrorists, ¬nancing
their destruction of innocent civilians. If Kant ruled out something™s being
good without quali¬cation on the grounds that in some contexts it had bad
effects, then he would be compelled to rule out a good will itself.
But Kant suggests another basis for ruling things out: if an object, disre-
garding its effects, is good in all contexts, then it is good without quali¬cation.
Qualities such as “moderation in affects and passions,” “self-control,” and
“calm re¬‚ection” are, Kant acknowledges, helpful in attaining all sorts of
ends. Yet he denies that they are good without quali¬cation, “for, without
the basic principles of a good will they can become extremely evil, and the
coolness of a scoundrel makes him not only far more dangerous but also
immediately more abominable in our eyes than we would have taken him
to be without it” (GMS 394). Kant seems to be suggesting here that when
Search for the Supreme Principle of Morality

“coolness” belongs to a scoundrel, it undergoes a “value reversal,” to bor-
row a phrase from Berys Gaut.51 What we (presumably imagining ourselves
to be impartial rational spectators) often take to be good (e.g., coolness
in an astronaut) becomes bad in some contexts. And it is bad considered
independently of its effects. The coolness of a scoundrel is presumably bad
even if, with its help, the scoundrel never manages to do anyone any real
harm. Kant, of course, claims that a good will is the only thing that never
undergoes a value reversal; an impartial rational spectator would hold that
in every context it is good.
Would an appeal to this Kantian argument force value realists “ those
who take their ends themselves to be unconditionally good “ to abandon
their positions? In returning to our environmentalist, do we ¬nd that a max-
imum number of (currently existing) species being preserved undergoes a
value reversal? Considered independently of its effects, is there a context
in which an impartial rational spectator would not take this to be good? I
suspect that answers to this question will differ. Those who see no inherent
value in biodiversity will be drawn to the view that in many contexts an im-
partial rational spectator would take maximum species preservation to be
indifferent rather than good. They might, for example, ask us to imagine the
following world. The human species is fully ¬‚ourishing and a maximum num-
ber of species have been preserved. Moreover (in the imagined world), if
it comes to pass that a maximum number of species is no longer preserved
(e.g., if thousands go extinct) humans would fully ¬‚ourish just the same.
Since maximum species preservation is important only insofar as it affects
human ¬‚ourishing, they might conclude, in the imagined world maximum
species preservation would have no value to an impartial rational spectator.
Others would disagree with this view, however, contending that even in that
world the existence of the maximum variety of life would itself be valuable.
To deny this would, in effect, be to embrace the appallingly prideful view
that human beings are all that really matters, the others might say; and this
is surely not a view that an impartial rational spectator would adopt. My aim
here is not to settle the issue. It is merely to illustrate that Kant™s argument
here is controversial at best. Through his appeal to ordinary moral reason,
he falls far short of showing that the environmentalist is rationally compelled
to give up the notion that species preservation is unconditionally good.
Of course, there are many other candidates for unconditional goodness
besides a maximum number of species being preserved. Someone might,
for example, defend the view that knowledge, courage, friendship, beauty,
and so forth are good in themselves, independently of any agent™s desir-
ing them. It is open to Kant to challenge any item on such a list on the
grounds that, unlike a good will, it undergoes a value reversal in some con-
text. But I suspect that this tactic would be no more effective with regard to
these purportedly unconditionally good things than it was regarding species
preservation.52 I hope that my discussions of species preservation as well as
The Formula of Humanity 71

universal happiness have illustrated the vulnerability of Kant™s claim that
nothing other than a good will can be considered unconditionally good.
If I am correct that this claim does not really threaten the value realist po-
sition, then step iii of the regressive argument is without suf¬cient support.
We are free to hold that some of the objects of our desires are good not be-
cause we desire them, but are rather good in themselves. Given that iii lacks
suf¬cient support, the regressive argument does not prove that if we take our-
selves to have good ends, we must hold humanity to be unconditionally good.

3.8 Shortcomings in the Derivation of the Formula of Humanity
Let me now crystallize my main ¬ndings regarding Kant™s derivation of the
Formula of Humanity. There are two main reasons why I do not believe that
this derivation, even as reconstructed by Korsgaard, is successful.
First, Kant does not prove that if we take there to be a principle that
conforms to his basic concept of the supreme principle of morality, then we
must hold there to be something unconditionally good. Granted, if there is
a supreme principle of morality, then every agent must always have a motive
available to him for conforming to it. As our example of PW illustrated, how-
ever, this motive need not be the notion that conforming to the principle
is itself unconditionally good or enables the agent to secure something un-
conditionally good. The “ground” of a categorical imperative might be each
agent™s being rationally compelled to view his conforming to this principle
as something good for him (though not necessarily good from an impartial
The example I have offered of a principle “grounded” in this way “
“Maximize your power over rational beings” “ is, in my view, a repellent
candidate for the supreme principle of morality. I venture that most readers
would agree. Nevertheless, it is illegitimate to infer that if a principle con-
forms to Kant™s basic concept of the supreme principle of morality, then we
must take there to be something that all agents are rationally compelled to
hold to be unconditionally good. Kant does not establish that unless we hold
there to be something unconditionally good (in his agent-neutral sense), we
cannot hold there to be a universally and unconditionally binding practical
principle (a categorical imperative).
The second main dif¬culty with the derivation of the Formula of Hu-
manity is, I think, more important than the ¬rst. Even if holding there to
be a categorical imperative requires holding there to be something uncon-
ditionally good, Kant does not establish that this must be humanity. Even in
Korsgaard™s ingenious reconstruction, we ¬nd no good reason to rule out
the possibility that the unconditionally good “ground” of a categorical im-
perative is everyone™s happiness. The regressive argument fails to threaten
utilitarianism. Moreover, it contains no compelling arguments against var-
ious forms of value realism. Kant™s response to those who would hold that
Search for the Supreme Principle of Morality

what is unconditionally good is the objects of their inclinations “ that is, their
ends such as maintaining biodiversity or gaining systematic knowledge of the
universe “ is that these objects are really only conditionally valuable, for if
these persons did not value them, then the objects would be devoid of worth.
But why must we agree that the value of such objects derives solely from our
wanting them? It would hardly seem unreasonable for someone to maintain
that Kant has things backward; it is not that environmental preservation is
valuable ( just) because we desire it, but rather that we desire it because it is
valuable. Yet the argument Kant suggests against the notion that some ob-
jects of our inclinations are unconditionally good is the controversial and, in
my view, ineffective one that a good will alone is unconditionally good. Kant
does not show that humanity alone is capable of being the unconditionally
good “ground” of the supreme principle of morality.
In sum, as I have argued, the derivation of the Formula of Humanity
contains two highly questionable steps. First, Kant does not establish that if
there is a supreme principle of morality, then there is something uncondi-
tionally good. Second, even if we assume that his ¬rst step succeeds, he does
not show that this unconditionally good something must be humanity.

<< . .

. 9
( : 27)

. . >>