Allen Wood - Humanity as an end in itself

Text-only Preview

[Note to the copy-editor. This Commentary's endnotes should become footnotes, as with
the other three Commentaries. I don't know how to make this change. Nor do I know
how to eliminate the two black lines in the last few pages of Susan Wolf's Commentary
directly above.]

Part One: Rational Consent, Practical Reason, and Humanity as End-in-itself
There is a great deal in Parfit's chapters, especially in Chapters 8 to 10 (on which I
am going to concentrate these comments) with which I strongly agree. I think
Parfit provides a better account than O'Neill and Korsgaard do of what Kant meant
in saying that for me to treat another as an end in itself, the other must be able to
"contain in himself the end of my action" (G 4:429-430),515 and also a better account
of the relation of this idea to issues surrounding hypothetical rationally consent. I
also find very illuminating Parfit's remarks about the relation of possible rational
consent to actual consent and how each bears on the morality of actions.
At a deeper level, too, I think I favor a reading of Kant that puts him closer to what
Rawlsian style Kantians would regard as "dogmatic rationalist" views in ethics -
and I think this means closer to the position Parfit wants to defend. Thus I would
accept, as good Kantianism, what Parfit calls a `value-based' theory of reasons;
Parfit's rejection of `desire-based' theories therefore seems to me nothing but good
Kantianism. I therefore also accept his thesis that "no reasons are provided by our
desires and aims." But to this I would want to add two other things (which I don't
think Parfit means to deny): first, that our desires and aims are often merely the
rational expression of value-based reasons, and second, that our desires might
constitute a crucial aspect of some of our reasons, as long as they stand in the right
relation to values.
Where I think I part company with Parfit is on certain questions of method in ethical
theory. He seems to prefer a method descending (as I see it) from Sidgwick -- a
method that involves appeal to what Sidgwick called "the common moral opinions
of mankind" (or just "Common Sense') in the formulation and testing of moral
principles. By contrast, I favor a method, which I find not only in Kant but also in
utilitarians such as Bentham and Mill, that would draw the fundamental moral
principle from very general and fundamental considerations about the nature of
rational desire and action, and would then attempt to reconcile these principles with
common moral opinions only insofar as those opinions can be seen as applications
of the principles. Sidgwick seems to have thought that what he called "primary
intuitions of Reason" are to be used only systematize and correct Common Sense,516

which continues to exercise authority within moral theory independently of first
principles, and might even help to shape the formulation of moral principles.517
The Kantian and Millian method that I favor, by contrast, involves a fundamental
principle whose ground is independent of moral intuitions or Common Sense, and
then the derivation from the fundamental principle of various moral rules or duties.
Conclusions about particular cases are not inferred directly from the first principle
at all, but rest on it only mediately, through what Mill calls "secondary principles"
and Kant calls "duties" (of various kinds, of which he provides a taxonomy). The
derivation of moral rules or duties from the first principle, moreover, is also not
deductive. The first principle is instead fundamentally an articulation of a basic
value (that of rational nature for Kant, that of happiness for Mill). The rules or
duties represent an interpretation of the normative principles applying that basic
value under the conditions of human life. In their application, moreover, the rules
or duties themselves require interpretation, and admit of exceptions, by reference to
the first principle.518 More recent (Sidgwickian) theory sets itself the goal of
providing a precise principle or set of principles which, along with a set of facts,
enable one to deduce the "right" conclusion about what to do under any
conceivable situation. That's what it is for Sidgwick to make ethics "scientific".519
For Kantian or Millian theory, as I understand them, this is such a hopeless goal that
it would be wrongheaded to orient your theoretical method to it.
The system of moral philosophy, following the Kantian conception, consists of three
different things: first, a fundamental principle or value (which Kant thought was a
); second, a body of empirical information and theory about human beings and
their situation (which in the Groundwork Kant called `practical anthropology' (G
4:388) and later described as `empirical principles of application' for the moral
principles (MS 6:217)); and finally a set of rules, duties, or other moral conclusions
resulting from the interpretation of the former principle or value in light of the latter
information. This third part of Kantian ethical theory is the taxonomy or system of
duties expounded in the Metaphysics of Morals (the ethical part in the Doctrine of
Virtue). It corresponds roughly to the set of moral rules that Mill regards as
involved in every case of moral obligation. and relate only loosely to the principle
of utility, which he does not regard as imposing on us any obligations directly, and
from which Mill immediately derives (even together with facts about the
consequences of actions) no substantive conclusions about what to do in particular
I think this way of conceiving of moral theory, and the fact that Parfit favors a
different theoretical method, accounts for some of the ways Parfit disagrees with
my interpretation of Kant at the beginning of Chapter 10. He quotes me
interpreting Kant's Formula of Humanity as End in Itself (FH) as saying that "we
must always treat people in ways that express respect for them" and then objects
that "most wrong acts do not treat people in disrespectful ways." The remark he

quotes here occurs in the context of a more systematic exposition of Kant's theory,
which, as I read it, is what Parfit would call a `narrow' or `monistic' value-based
theory. For this theory, all reasons are grounded, directly or indirectly, on the
single value of rational nature, which Kant expresses in two ways: as the objective
worth of humanity as end in itself, and the dignity of personality as universally
Respect, as I understand it, is first of all a feeling or emotion. Contrary to the Stoics
(and to some grossly mistaken misinterpretations of Kantian ethics), Kant thought it
impossible for a finite rational being to act rationally at all without having certain
feelings and emotions and manifesting them in its actions. In the Metaphysics of
, Kant specifies four such feelings (moral feeling, conscience, love of human
beings, and respect). These feelings are rational rather than empirical in origin,
and susceptibility to them is a condition for being a moral agent at all (MS 6:400).
I would describe respect in general as the feeling appropriate to the rational
recognition of objective value.521
Respect is something we not only feel but also show in actions that express it. It is
the active expression of respect rather than the mere feeling that matters for moral
conduct. On Kant's monistic value-based theory of practical reasons, all reasons
for action are based directly or indirectly on the objective value of rational nature,
and this is especially true of moral reasons that take the form of categorical
imperatives. Obedience to every categorical imperative thus involves showing
respect for the objective value of rational nature. In that sense, what morality
demands most fundamentally is that we show respect for that value, and violations
of morality all involve treating that value - often, the value of rational nature in the
person of rational beings - with disrespect. Many morally wrong actions do not
"display disrespect for people" in any conventional sense of that phrase, but if
Kant's theory is correct, the moral wrongness of these actions always consists
fundamentally in the way they show disrespect for the objective value of rational
Parfit recognizes the Kantian distinction between values to be respected and values
to be promoted. But he is worried that the claim that dignity is a value above all
price may commit Kantians to the view that rational nature as a value to be
promoted must take absolute priority over other values to be promoted. This is,
for instance, the way. Parfit reads the following statement by Thomas Hill: "Kant's
view implies that pleasure and the alleviation of pain, even gross misery, have mere
price, never to be placed above the value of rationality in persons."522 That fear
seems to me based on a misunderstanding. Promoting rational nature (as one
value that can be promoted) is grounded in respect for rational nature (as the basic
value to be respected). It is the latter value that has a dignity that is beyond all
price, and it must be given priority over all competing values. But equally,
concern for the alleviation of human suffering (as a value to be promoted) is

grounded in this same fundamental value. But this implies no absolute priority of
the value of developing rational nature (as one of the values to be promoted) over
other values to be promoted that are also grounded in respect for rational nature.
If the above quotation from Hill is correctly read as asserting that priority, then his
position is not a correct interpretation of Kantian doctrines.
In Kant's view, the objective value of rational nature grounds two general kinds of
ends which are duties: our own perfection and the happiness of others. (The value
of our own happiness, except as an indirect duty, is for Kant an object of prudential
rather than moral reason; and the perfection of others is a duty for us only insofar as
we contribute to perfections they want to acquire, and therefore falls under the
heading of their happiness.) Perfection prominently includes our rational nature
(both moral and nonmoral) as a value to be promoted. Both kinds of duty are
wide or imperfect. Thus for Kant there is no systematic priority of perfection over
happiness as ends or values to be promoted.
Parfit is also in danger of misunderstanding Kant when he says that the `humanity'
which has dignity cannot refer to non-moral rationality. Kant says that humanity,
as the capacity to set ends according to reason, is an end in itself and that humanity
insofar as it is capable of morality has dignity. As I interpret him, Kant holds that
it is our humanity that is an end in itself - where `humanity' is has a technical sense,
referring to our capacity to set ends (which includes both instrumental rationality
and prudential rationality - the capacity to frame a concept of happiness and to give
our happiness priority over more limited aims of inclination). We should therefore
include the permissible ends of others, especially their happiness (as the general
and comprehensive conception of those ends), among our ends as well (though
there are no strict rules in general regarding the priority we must give all these ends
among one another). Dignity - by which Kant means that supreme worth which
must never be sacrificed or traded away -- belongs to rational nature not in its
capacity to set ends, but only in its capacity of giving (and obeying) moral laws (G
It is the capacity for morality, however, not its successful exercise, that has dignity.523
Thus I agree with Parfit when he interprets Kant as saying that even the morally
worst people have dignity, and in that sense they have exactly same worth as even
the morally best people. I also agree with Parfit when he says that this view of
Kant's expresses a "profound truth." Parfit is further correct to point out that none
of this implies that my having dignity as a human being makes me a good human
. Not everything having value is thereby something good, especially good of
its kind. For Kant, the good is that which is recognized as practically necessary
independently of inclination (G 4:412). Having a character like that of a bad
person is the direct reverse of what is practically necessary, though it is also
practically necessary to treat even the worst person with the respect due to the

dignity of rational nature, and so it is that treatment of the bad person, and not the
bad person, that is good.
Parfit denies that FH - the principle that we should always respect humanity as an
end in itself - is a practically useful principle. In response to my claims that it
provides us with the right value-basis for settling difficult issues and that on many
difficult issues, it is an advantage of FH that different sides can use it to articulate
their strongest arguments, Parfit asserts that on a wide range of disputed issues
appeals to FH do not in fact constitute the strongest arguments of each side. I
think we may be talking past each other here, because we are beginning from
different assumptions (which I have tried to clarify above) about the aims and
structure of moral theory and the relation of a theory's basic principle to
conclusions about what to do. Kantian theory is grounded on a supreme principle,
which is then applied interpretively to a body of empirical information and theory
about human nature and human life, yielding a set of moral rules or duties. These
in turn are applied to particular circumstances, through practical judgment, in
determining what to do.
FH is one of Kant's formulations of the supreme principle, the one he uses most
often in deriving his system of duties in the Metaphysics of Morals. That is the role
FH is playing when I make the claims about which Parfit is skeptical. I suspect
that Parfit, on the other hand, thinks of moral theory as the attempt to formulate
precise principles from which we can rigorously derive a set of conclusions about
what to do in all actual or imaginary cases. The acceptability of these principles,
for Parfit, depends on how the conclusions derivable from them match up with
Sidgwick's "Common Sense" or "common moral opinions of mankind." Principles
well-grounded might in difficult cases give us reasons for revising our conclusion
about particular cases, but flagrant and systematic conflict of a candidate principle
with our intuitions is regarded as invalidating that principle. Parfit is treating FH
as a principle to be evaluated by these criteria, and he is rejecting it as too
indeterminate to yield the specific conclusions such a principle is supposed to yield,
and hence also incapable of providing adequate arguments on different sides of a
moral controversy that would be required by this conception of moral theory.
When FH is regarded in this way, I think Parfit is right, but not when it is regarded
in the way I regard it - which is also the way I think Kant regarded it. (My way of
reading Kant obviously involves reading his four famous illustrations of the
Formula of Universal Law in quite a different way from that in which they are
customarily read - including, I think, the way Parfit chooses to read them in
Chapters 12 and beyond. But that difference will not be pursued further in these
Part Two: "Trolley Problems"

The rest of my comments here will contain some general reflections on some of the
examples Parfit uses, especially in Chapters 8 and 9. I think these comments are
relevant to the theoretical differences I have tried to sketch above, for they concern
one now fashionable way of executing the methodological strategy I have suggested
that Parfit draws broadly from Sidgwick. I don't think the following remarks do
anything at all discredit the Sidgwickian program broadly conceived. Like many
ambitious philosophical projects, is too formidable in its conception ever to be
refuted by a few clever arguments or examples. But I do intend to challenge some
fashionable ways of carrying out such a program. My comments also relate to FH,
in that they help to illustrate the way in which I think it can figure productively in
moral reasoning. I should also frankly admit that these comments give me the
opportunity to get off my chest some complaints about what many moral
philosophers do nowadays.
In May of 2001, the Tanner lecturer at Stanford University was Dorothy Allison,
author of the novel Bastard Out of Carolina. Allison didn't talk much about moral
philosophy as such, but she did discuss a `lifeboat problem' that she had heard
about from a philosopher. Her reaction was to reject the problem -- to refuse to
answer it at all, -- on the ground that we should refuse on principle to choose
between one life and five lives. Even to pose the question in those terms, she said,
is already immoral. The only real moral issue raised by such examples, she
thought, is why provision had not been made for more or larger lifeboats. To
many philosophers her remarks would no doubt seem naive or even unreasonable.
Yet I think Allison's reaction to the lifeboat problem is far more sensible and right-
minded than what we usually get from most of the philosophers who make use of
such examples.
I am going to refer to these kinds of examples not as `lifeboat problems' but as
"trolley problems". (None of Parfit's examples are actually about trolleys, though
two of them are about trains.) They are all examples where the main point is that
you must choose between saving more people from death and saving fewer. Since
we think a human death is in general something very bad, it is natural also to think
that the option involving fewer deaths must be preferable to the one involving more
deaths. The examples gain their poignancy from the fact that this apparently
obvious point suddenly begins to seem questionable or even counterintuitive when
the fewer deaths are caused in the wrong way. The intent of the examples is
usually to incite us to formulate principles that correspond to, or even justify, our
moral intuitions (or deliverances of Sidgwickian "Common Sense") about the
difficult or problematic cases presented in the examples. The hope is apparently
that principles arrived at in this way will help us decide difficult cases in real life
with Sidgwickian scientific precision.
Some might think that if FH regards every rational being as having dignity (or
worth that cannot be rationally traded away to get anything else), then it might very

well not only support Allison's judgments about the lifeboat problem, but also
entail that there could be no rational way of choosing between one life and five
lives, or if it comes to that, five billion lives. If so, then FH would appear to have
consequences that seem plainly unacceptable according to our intuitions. We
apparently could never permit even a single death, not even to save the whole
human race.
No doubt the fact that rational nature has dignity or incomparable worth does mean
that the lives of beings having rational nature are valuable and important. But
merely from the fact that the value of rational nature cannot be rationally sacrificed
or traded away, it clearly does not follow that the lives of rational beings can never be
rationally sacrificed. If a person heroically sacrifices her life to save others, or to
uphold some important moral principle, that is not a case of undervaluing her own
rational nature. Depending on the circumstances and the principle involved, it
might even be a case of preferring the value of her rational nature to the value of her
life, and Kantian ethics might even require it. Nor does FH lend unambiguous
support to the vague idea of the "sanctity of human life" - an idea that, in its
popular and political application, usually involves a lot of self-deceptive rhetorical
posturing, and is sometimes put in the service of some of the most pernicious moral
superstitions currently on sale in the marketplace of moral ideas (for instance,
dreadful superstitions about the unexceptionable wrongness of euthanasia, or the
right to life of human embryos and fetuses). I strongly caution against associating
FH with morally obscene popular prejudices such as these.
The bearing of FH on trolley problems is therefore also not entirely clear. One
thing I hope is clear by now is that for Kantian ethics, the point of a moral principle
such as FH is not directly to tell us what we should do. It is rather to ground a set
of rules or duties, and more generally to orient us as to how we should and should
not think about what we should do. We would be right to conclude from FH, for
instance, that we should be reluctant to treat human lives as having the sort of value
that can be measured and reckoned up. That is what I think Dorothy Allison was
getting right. It would follow that answers to problems like Parfit's Lifeboat, Tunnel
and Bridge, therefore, can never be as clear (or as trivial) as the arithmetical fact that
five is greater than one. The tendency of some moral philosophers to draw such
inferences is due to their bad habit of thinking that the canonical form of every
moral principle must consist in the scientifically precise way it preferentially ranks
states of affairs (as the outcomes of actions). But what FH tells us is that the
fundamental bearers of value are not states of affairs at all, but persons and the
humanity or rational nature in persons. This is not a kind of value that translates
easily into preferential rankings of states of affairs.
FH does not imply that it is always immoral to choose five lives instead of one, but I
think it does imply that we should be reluctant to think about such choices in those
terms, or indeed in terms of any preferential rankings of states of affairs. FH rather

implies that we ought to arrange things in the world so that agents are not faced
with choices of that kind. Of course this means arranging things, as far as possible,
so that one life need not be sacrificed to save five. But it also means arranging
things -- including our moral deliberations -- so that when numbers of lives are at
stake, the choices dictated by our moral principles are not based merely on the
numbers, as trolley problems -- in the very way they are posed, through the careful
selection of information included in and excluded from them -- often suggest they
have to be.
I have long thought that trolley problems provide misleading ways of thinking
about moral philosophy. Part of these misgivings is the doubt that the so-called
`intuitions' they evoke even constitute trustworthy data for moral philosophy. As
Sidgwick was fully aware, regarded as indicators of which moral principles are
acceptable or unacceptable, our intuitions are worth taking seriously only if they
represent reflective reactions to situations to which our moral education and
experience might provide us with some reliable guide.524 Poll-takers are well aware
that the way a question is framed often determines the answer most people will
give to it. What might seem to us genuine intuitions are unreliable or even
treacherous if they have been elicited in ways that lead us to ignore factors we
should not, or that smuggle in theoretical commitments that would seem doubtful
to us if we were to examine them explicitly.
Most of the situations described in trolley problems are highly unlikely to occur in
real life and the situations are described in ways that are so impoverished as to be
downright cartoonish. (In imagining Bridge, For instance, I can't help casting my
favorite cartoon superhero, Wile E. Coyote, in the role of the hapless single person
who may be toppled onto the track.) But this by itself is surely not a problem. It is
extremely rare for a man to lure teenage boys into his apartment, then kill,
dismember and eat them; and at this writing, at any rate, it remains an utterly
unique occurrence for a group of terrorists to hijack airliners and crash them into
skyscrapers filled with innocent people going about their daily lives. But the rarity
of such cases does not lead us to mistrust our moral intuitions about these cases.
Nor do we mistrust our moral reactions to the absurdly fantastic villainy sometimes
depicted in comic books and action movies.525
The deceptiveness in trolley problems is indirectly related to their cartoonishness,
however, in that it consists at least partly in the fact that we are usually deprived of
morally relevant facts that we would often have in real life, and often just as
significantly, that we are required to stipulate that we are certain about some
matters which in real life could never be certain. The result is that we are subtly
encouraged to ignore some moral principles (as irrelevant or inoperative, since their
applicability has been stipulated away). And in their place, we are incited to
invoke (or even invent) quite other principles, and even to regard these principles as
morally fundamental, when in real life such principles could seldom come into play,

or even if they did, they would never seem to us as compelling as they do in the
situation described in the trolley problem.
Trolley problems focus primary attention on the value or disvalue of certain
consequences or states of affairs (usually, more human deaths or fewer). But
trolley problem philosophers are by no means all consequentialists. Trolley
problems are quite frequently used, in fact, to support anti-consequentialist
conclusions in moral philosophy, and many of them appear to do so. But in these
problems, attention is directed exclusively to the consequences of certain actions for
the weal or woe of individuals and also the way those actions relate causally to
those consequences. Typically, the circumstantial rights, claims and entitlements
people would have in real life situations are put entirely out of action (ignored or
stipulated away). In the process, an important range of considerations that are,
should be, and in real life would be absolutely decisive in our moral thinking about
these cases in the real world is systematically abstracted out. The philosophical
consequences of doing this seem to me utterly disastrous, and to render trolley
problems far worse than useless for moral philosophy. I would like to illustrate
these general points by briefly discussing three problems used by Parfit in Chapters
8 and 9.
Lifeboat. It seems to me that when faced with a situation like Lifeboat, there is
only one morally defensible policy: You must seek to rescue all six people as quickly
and efficiently as possible. It might very well be true that, following this policy,
you should first set about rescuing the five and only then try to rescue the single
person, because in that way you will go farther, faster and with greater certainty
toward achieving your only legitimate goal (which is rescuing all six). But if you
thought you could go farther faster and with greater certainty toward the goal of
saving all six by rescuing the single person first (say, because this person's rock is
right on your way to the rock with the other five on it), then you obviously should
do that.
It is relevant here -- even decisive -- that in the real world, if both rocks are in
imminent danger of being swept under the water, then you would very likely not
know for certain that you must choose between saving the single person and saving
the five. (The stipulation that you are certain about this ruins the real moral issue
just as certainly as it would ruin some issue in rational choice theory to stipulate
that you are sure which box being offered you contains the larger amount of
money). Rather, in real life there would always be some chance that you would
save all six, and if both rocks were about to go under there would also probably be a
significant chance that no matter what you did, all six people would drown. When
a philosopher simply stipulates that we are certain you can save all and only the
inhabitants of exactly one rock, then we should be clear that he is posing a problem
so different from otherwise similar moral problems you might face in real life that

any "intuitions" we have in response to the philosopher's problem should be
There is one intuition about a situation such as Lifeboat that is perfectly clear and not
the least suspect. It is this: if any of the six drown, the result is tragic - it is
unacceptable. You will regard ourselves as having failed significantly in your
rescue efforts no matter what you did, even if you know your failure was inevitable
and not your fault. Another vivid and reliable intuition is that all concerned have
an urgent obligation to call to account whoever is to blame for the fact that there
were not enough life boats. They should to find out why this happened, and take
steps to minimize the chances of its happening ever again. We recently saw this
point illustrated dramatically in the universal reaction to the utter incompetence of
federal authorities to hurricane Katrina.
These intuitions are at least as strong and certain as any intuition we might have
about what you should actually do about the single person and the five. To many
trolley problems, as they are posed,526 I think the right reaction is to regard it as
simply indeterminate what the agent should do, and the only real moral issue
raised by the problem is (as Dorothy Allison rightly said), how the situation in
question was permitted to arise in the first place. The fact that lives are at stake is
intended to compel us to reject this correct reaction, and make us feel that we
simply must decide to do something - hence to decide that something is morally
right and something else is morally wrong.
Yet trolley problem philosophers would regard us as missing the whole point of the
problem if we even bothered to express any of the moral intuitions that don't
directly involve saying what the agent should do. These philosophers are focusing
our attention shortsightedly, even compulsively, solely on the question about what
you should do in the immediate situation, as if that were the only thing moral
philosophy has any reason to care about. In the context of the moral epistemology
that goes with Sidgwickian style moral theory, the reasons for this restriction of
attention are clear enough. But the fact that the clearer and more compelling
intuitions about such a case are irrelevant to what interests them ought all by itself
to make us distrust the philosophical value of the questions these philosophers are
Why trolley problems mislead. In real life, people go to a lot of trouble to arrange
things so that no one will ever be placed in the position that, for example, the
bystander in the train examples is placed. There are sound moral reasons why this
is so, reasons that could be derived from FH and that are closely connected to
Dorothy Allison's reaction that it is already immoral to ask anyone to decide
between one person's life and five people's lives. The way I would put the point is
to say that even if some choices do inevitably have the consequence that either one
will die or five will die, there is nearly always something wrong with looking at the