. 15
( 38 .)


strong altruistic reasons “ ones that can justify a great deal “ also require
a great deal, then it will be quite severely irrational to act on a desire to
harm or kill someone. But if one does harm or kill someone, just for fun,
or for some small pro¬t, then what seems to explain this is the lack of a
rationally required motive, or the possession of a rationally prohibited one.
That is, what explains one™s action is the very defect in virtue of which
it was irrational. It is hard to see how it would be fair to hold someone
morally responsible in such a case.
I am sure that the above argument will leave some readers unpersuaded.
But in fact I also think that the best argument for the claim that immoral
action is not always irrational is simply a direct appeal to intuitions about
rationality, and need not proceed indirectly, through intuitions about moral
responsibility. Immoral actions that do not also qualify as irrational based on
the harms they are likely to cause the agent simply do not seem irrational.
Clever low-risk embezzling schemes, if they really are obviously low-risk,
do not seem irrational. The raw exercise of power for personal gain, such
as that exhibited, for example, by Carlos Menem during his tenure as
president of Argentina, does not seem irrational, especially because by
stacking the courts with political allies he eliminated any real chance of a
criminal conviction in later years. In the face of the fact that so many people
who gain such power end up acting in immoral ways, the appropriate
response is to admit that what keeps us moral is, to some degree, our own
self-interested stake in behaving morally “ the possibility of punishment,
censure, loneliness, lack of friends, and so on. It is much less plausible,
as an explanation of the phenomena, to claim that the power to act with
impunity either tends to come to those who are already irrational, or that
such power tends to produce irrationality.
I do not think that those who hold that all immoral action is irrational
adopt that view because it seems correct on its face. Rather, I think they
are drawn to it on the basis of theoretical claims that are hard to avoid
if one does not acknowledge the distinction between the justifying and
requiring roles of reasons. For it is clear that altruistic reasons can indeed
be very powerful. If one doesn™t notice that the intuitively unproblem-
atic examples that demonstrate this power are exclusively examples that
demonstrate justifying strength, one is likely to think that altruistic reasons

as ˜suf¬ciently strong that in some cases one™s awareness of reasons that would make it
irrational to act on the desire would be psychologically incapable of dissuading one from
acting.™ For an account of disabilities of the will in these terms, see Duggan and B. Gert
(1967, 1979).

Brute Rationality

are very powerful simpliciter. And it is a short step from this view to the
idea that immoral behavior is irrational. Views that do not acknowledge
both functional roles of practical reasons tend to hold that there is gen-
erally a unique action that one has most reason to perform.33 But if this is
right, then moral theory (as opposed to a theory of practical rationality)
is in danger of losing all its practical importance. After all, once one has
determined which action one has most reason to perform, what more
could one wish to know, for purposes of deciding how to act?34 Perhaps
in cases in which a number of actions were tied for ¬rst place, it would be
interesting to know which of the actions was morally required. But even
in such a case, there would be no special reason to perform the morally
required action instead of one of the others. If, on the other hand, morality
is, as G. A. Cohen has put it, “a choice within rationality,” then when all
of the rationally permissible options are laid before us, those of us with a
concern to behave morally will have some genuine use for moral theory.35

33 As has been mentioned, there are ways of avoiding this by retreating to a satis¬cing view
of rationality, or by appeal to the notion of incommensurability. Chapter 5 argues that
satis¬cing views cannot capture some fairly uncontroversial intuitions about the rational
status of particular actions, and that incommensurability cannot capture others.
34 There is a related danger for views that identify rationality and morality.
35 See his reply to Korsgaard in Korsgaard (1996a), p. 173.

Accounting for our actual normative

In arguing for the theoretical need to distinguish the justifying and requir-
ing roles of practical reasons, some of the work in the previous three
chapters was devoted to arguing that we can take certain commonsense
normative claims at face value. Among those claims were, for example,
that being immoral is not necessarily irrational, but that making sacri¬ces
for other people is not irrational either; that it is irrational to refuse to take
medicine that will restore one to perfect health and a happy life, regardless
of one™s indifference to that prospect; that simply having a strong desire for
things like pain or disability does not, by itself, give one the slightest reason
to pursue them. This chapter will take these sorts of claims for granted.
The point here is to argue in a more formal way against views that attribute
a single strength value to practical reasons, and that go on to claim that
the rational status of an action is dependent only on the strengths of the
reasons that favor and oppose it.1 Once this is demonstrated, the second
half of the chapter will then go on to argue that the justifying/requiring
distinction explains the relevant phenomena better than do two other
proposals: incommensurability of reasons, and a technical device called
an ˜exclusionary permission.™ In this chapter, as in chapters 2 and 3, the
arguments will be presented in terms of subjective rationality. But they
can generally be taken to demonstrate formally similar points regarding
objective rationality, since the claims will remain true even if we stipu-
late that the agent is fully informed. And indeed, in the examples I make
use of in this chapter, I will make the simplifying assumption that all agents
are aware of the relevant reasons, and that they do not falsely believe that
there are any other relevant reasons. The possibilities of ignorance and mis-
take have important normative implications, as we will see in chapter 7,
but these possibilities can be bracketed for current purposes.

1 These strength values could be ordinal or cardinal, vague or determinate; it will make no
difference to the arguments presented here.

Brute Rationality

s i ng le - value v i ews
It is a common view that:
R1 If there is a reason in favor of an action, and no reasons
against it, then one is required, on pain of irrationality, to
perform the action.2
On the surface, of course, this is an extremely plausible view. Suppose,
for example, that there is only one action that will help one to avoid an
immediate painful injury: perhaps one is in the path of a thrown rock.
In such a case, and in the absence of any other reasons to stay put, one
is rationally required to step out of the way of the rock. Similarly, it is a
common view that:
R2 If there is a con¬‚ict between the only two reasons relevant
to a choice, and one reason is stronger than the other, then
one is required, on pain of irrationality, to perform the
action favored by the stronger reason.3
In fact, R2 is sometimes taken as explaining what it means for one reason
to be stronger than another.4 Again, this view has considerable intuitive
appeal. Consider, for example, the following modi¬cation of the above
example. It is still the case that one will be painfully injured by a thrown
rock, but one also knows that if one allows this to happen, then one will
receive a huge amount of money. Perhaps the rock-thrower is a drunken
tycoon who has invariably settled such cases in the past. In this case there
is a reason in favor of staying put, and also a reason in favor of getting
out of the way.5 To the degree that we regard the former reason as clearly
2 See, e.g., Nagel (1970), pp. 50“51; Smith (1994), pp. 148, 174“75, and 177; Korsgaard
(1996a), pp. 225“26; and Foot (1978b), p. 152. In fact, Nagel is committed to this view
only if he means by ˜suf¬cient,™ ˜suf¬cient to cause action in a rational agent,™ but this does
appear to be what he means. And Smith is more concerned with the rational requirement
to act when one believes there is a reason. But the position this book is defending tells
equally strongly against such a view.
3 See Raz (1999a), pp. 25“26. In fact, Raz is one of the few theorists who, although
committed to this view, explicitly recognizes that this cannot be the end of the story (1999a),
p. 35. Raz supplements his account by adding second-order reasons and other second-order
reason-affecting entities, and by committing himself to widespread incommensurability of
reasons and values. See Raz (1999b), p. 46. I argue for the superiority of my solution
4 What R2 actually de¬nes is the circumstance of one reason having suf¬cient requiring
strength that the justifying strength of the other reason is insuf¬cient to justify acting
against it.
5 Of course, this could also be called a reason against staying put.

Accounting for our normative judgments

stronger than the latter, it is plausible to regard jumping out of the way as
a case of irrationality “ of ˜weakness of will™ caused by immediate fear of
pain. Finally, it is a common view that:
R3 Rational action in general is action based on the best rea-
sons: ˜the rational alternative is the one supported by a pre-
ponderance of reasons.™6
This view, too, has a great deal of plausibility on its face. Suppose some
decision is to be made, and that it is suf¬ciently complex that a group
of people are assembled to research the various options. At the end of
the information-gathering and of the assessment of the reasons in favor
and against all the options, it would be strange indeed if the head of the
committee decided on an option that everyone (including the committee
head herself) agrees has less in favor of it than some other option.
The three views listed above increase in strength: R1 is implied by R2,
and R2 is implied by R3. To see that R3 implies R2, consider any case in
which there are only two reasons relevant to a choice, these two reasons
favor different options, and one of the reasons is stronger than the other.
This is the type of case about which R2 is making its claim. Now assume
that R3 is true “ that rational action is action based on the best reasons.
Then it follows that in the representative two-reason case, the rational
action is the action based on the stronger reason. That is, it follows that
R2 is true. To see that R1 is implied by R2, consider a case in which
there is only one reason in favor of an action, and no reason against it or
in favor of any other action. Such a case is plausibly regarded as a limiting
instance of two-reason cases. That is, such a case is plausibly regarded as a
case in which there is a reason of some strength in favor of an action, and a
reason of zero strength (no reason) against it. Assume that R2 is true “ that
when two reasons con¬‚ict, the rational action is the action on the stronger
of the two. Then it follows that in the one-reason case, the rational action
is the action based on the lone relevant reason. That is, it follows that R1
is true.
This chapter argues explicitly against R2: that one is always rationally
required to act on a stronger reason. Hence, by implication, R3 will also
6 See Gibbard (1990), p. 160 and Par¬t (1997), p. 99. See also Herman (1993), pp. 166“68.
There Herman indicates the common nature of this view by explaining an interesting way
in which it might be argued that Kant does not hold it. But her ˜defense™ of Kant suggests
that for all practical purposes he holds both this view and the view that the stronger reason
always generates a requirement.

Brute Rationality

be denied: the view that one is always rationally required to act on the
balance or preponderance of reasons. R1, which amounts to the claim that
all reasons create prima facie rational requirements, this chapter neither
af¬rms nor denies. The of¬cial position of this book is that R1 is false,
but as has been mentioned at the conclusions of chapters 2 and 3, a small
modi¬cation to the of¬cial view could easily accommodate the truth of
R1 without giving up any signi¬cant points. In particular, it is possible to
concede R1 while still holding that the requiring and justifying strengths
of reasons can come apart. That is, it is possible to concede R1 with-
out conceding R2. And the motivated denial of R2 is a view with very
signi¬cant philosophical implications.

a mot ivat i ng e xam p le
Suppose that I am thinking of donating two hundred dollars to a certain
charity. The money I donate will provide food for forty children for four
months, preventing them, at least for that period, from suffering from
serious malnutrition and the attendant risk of illness and death. It is clear
that I am rationally allowed to donate the money if it is reasonable to
believe that the money will be used for this purpose, and if there are
no other signi¬cant reasons bearing on the case. But I am not rationally
required to donate the money just because there is a reason in its favor that is
suf¬cient to justify doing so. Even though the reason in favor of donating
the money is suf¬cient to justify doing so, it is not suf¬cient to require it.
One point of the justifying/requiring distinction is that it allows us to say
that an altruistic reason™s insuf¬ciency to require action is not a result of
its being too weak to generate a requirement. For in one important and
intuitive way the fact that an action will prevent serious malnutrition in
forty children for forty days is a very strong reason: it can rationally justify
a great deal. In this sense, such a reason is in fact much stronger than the
reasons against the donation. For the mere prospect of saving two hundred
dollars cannot justify nearly as much. And yet it is not irrational to refrain
from donating just because one wants to keep the money for one™s own
inde¬nite future purposes.
The inability of the altruistic reason to make it rationally required to
donate the money is not the result of that reason™s weakness. Rather, the
insuf¬ciency of this altruistic reason to rationally require the donation is
the result of its being of a certain type: a type, the function of which is
to rationally justify, and not to rationally require. Even if I could help the

Accounting for our normative judgments

people a great deal more, for the same money, or if I could help them
an equal amount at less expense, it might still not be irrational to fail to
do so. Indeed, virtually every reader is in this circumstance, for virtually
every reader has at least two hundred dollars that she could donate to
charity, and that she is not going to donate. Even if a reader has recently
donated a large amount, she is still in this position, for unless her donation
has actually impacted her ¬nances in a signi¬cant way, she still has two
hundred additional dollars that could be disposed of relatively painlessly.7
And the reason in favor of donating those additional two hundred dollars
is the same powerful reason: the donation could prevent malnutrition for
forty children for four months. For most readers, the loss of two hundred
dollars would not make any signi¬cant impact on their ¬nances, happiness,
or opportunities. As a result, for most readers there is only a very weak
reason against giving the money away. And yet I, and they, may rationally
decide simply to spend the money on a new coat, or tickets to the opera,
even if we know that by donating it we could do a great deal of good for
It is the of¬cial position of this book that altruistic reasons can never,
in themselves, rationally require action, even though such reasons can
justify actions that stand in need of justi¬cation. But neither the motivating
examples nor the arguments in this chapter depend upon this view. Rather,
the examples simply depend upon the existence of a gap between the
justifying strength and the requiring strength of altruistic reasons. It is a
matter of indifference to the arguments presented below whether this gap
is the gap between some and none, as the of¬cial view holds, or between
more and less.
Again, even though the altruistic reasons in favor of a donation need
not rationally require me to donate two hundred dollars, these altruistic
reasons are not weak. And even though one might be rationally required to
spend the same amount of money for some other reason “ for example, to
avoid the loss of one™s index ¬nger “ this does not mean that the altruistic
reasons in the donation example are weaker than the reasons provided


. 15
( 38 .)