. 23
( 38 .)


requirement on practical reasons. If one uncritically accepts the uniqueness
assumption, then this form of reasons internalism may seem trivially true,
as indeed it has seemed to many theorists. For if reasons have a unique
strength, and if this strength corresponds to the motivation they would
provide to a rational agent, then any consideration that could provide a
rational agent with zero motivation would provide a rational agent with
zero motivation. It would then seem very odd to regard the consideration
both options in all three cases would be for her to acquire a new desire in a rationally
inexplicable way at some point. For all that is needed for the argument to go through is
the possibility that for one agent, all the normative claims in (1) through (3) could be true.
It may also be worth noting here that the way in which we acquire new desires is typically
not by reasoning, or by any intellectual processes at all, and that realistic internalists such
as Darwall and Williams allow an important role for imagination, or ˜becoming aware
of things in a vivid way,™ in the rational acquisition of new desires. See Darwall (1983),
pp. 39“40, and Williams (1981), pp. 104“5.
31 The best candidates for such reasons may be altruistic reasons that involve the promotion
of bene¬ts (as opposed to the prevention of harms) for others.

Brute Rationality

as a reason at all. But if one rejects the uniqueness assumption, then zero-
minimum reasons could still reasonably be called reasons. For as long as they
had nonzero maxima, they could account for the rational permissibility of
actions that were opposed by other reasons “ even by reasons with nonzero
minima. For example, although it may be rationally acceptable not to lift
a ¬nger merely to give a stranger some pleasure, it does seem that one
would be rationally justi¬ed in going to some pains to do so. That one™s
action would give pleasure to a stranger may therefore be a reason that
violates the internalism requirement.

conc lu s i on
This chapter has not argued in favor of ideal motive accounts of practical
reasons. Nor has it even tried to explain what precisely such accounts are
best regarded as trying to provide: meaning analyses or truth-conditions,
reductions of the normative to the nonnormative, or something else.
Rather, the point has been to suggest that whatever such accounts are
doing, they could do it better if they explicitly rejected the claim that
there is a unique degree of motivation that any given consideration would
provide to an ideal agent. But if they do reject this assumption, then they
must use two values to characterize the motivation a given consideration
would provide to the idealized agent: values that specify the minimum and
maximum of a range. Discussion of some examples revealed that these two
values function in a way that is isomorphic to the way in which requiring
and justifying strength function to determine the rational status of actions.
The fact that improved versions of ideal motive accounts give nor-
mative reasons features that are isomorphic to requiring and justifying
strength supports the view that reasons really do play the two normative
roles described in the preceding chapters. The appearance of these fea-
tures also expands the common ground between the view advocated in
this book, and ideal motive accounts of reasons and rationality, and allows
ideal motive theorists to understand justifying and requiring strength in
their own terms. It is my hope that this common ground will allow for
a real philosophical engagement between the two views. However, the
same sharing of features that allows for this philosophical engagement may
also lead some philosophers to claim that talking about the justifying and
requiring roles for practical reasons is just a needlessly complicated way of
talking about something that is more intuitively captured by ideal motive
accounts. So it may be worthwhile to explain why it is plausible to hold

The relation to other views

that the notions of justifying and requiring strength have an explanatory
priority over the notions of minimum and maximum rationally permissible
degree of motivation.
As was mentioned at the outset of this chapter, ideal motive accounts
come in two varieties: naturalistic and nonnaturalistic. The problem with
naturalistic versions is that for any naturalistically speci¬ed set of condi-
tions, it is possible that an agent might meet them, and yet still have a desire,
for example, to scratch the skin off of his legs for no reason. For an agent
might have a brain tumor, or a chemical imbalance, or some other physical
malady. It does no good here to protest that such accounts are intended
to be restricted to rational agents, for we are in search of an account of
what rational agents are. Moreover, the prospects are extremely dim for
producing a purely naturalistic set of criteria that will sort out chemical
and neurophysiological states into those that can be included as part of
˜ideal conditions™ and those that cannot. Purely statistical criteria will not
do. What, for example, would differentiate a statistically rare con¬guration
of the brain that made someone take extra delight in nature from one that
made a person self-destructive? One could, of course, use a naturalistic
description of the goals or ends towards which the state inclined the agent:
states that inclined agents towards self-desctruction would be disquali¬ed
as part of an agent in ˜ideal conditions.™ But this move gives the game away.
For what criteria would we use for selecting the relevant ends? The fact
is that we use a normative notion of rationality in sorting psychological
states into those that impugn the agent™s rationality, and those that do not.
Thus, the only plausible form of an ideal motive account will be one that
uses an antecedent notion of normative rationality: one that we have been
calling a nonnaturalistic ideal motive account. But once we see this, it
should become clear that what explains the fact that some considerations
yield reasons that have a signi¬cant range of rationally acceptable degrees
of motivation is the fact that we are using a fundamental principle of ratio-
nality with a form similar to those of principles P and Q of chapter 3. That
is, what explains the existence of ranges of rationally permissible degrees
of motivation is the prior existence of a gap between the justifying and
requiring strength of certain reasons.


Two concepts of rationality

This book has so far been primarily dedicated to arguing for and explaining
a distinction between two normative roles for practical reasons: justifying
and requiring. One reason for this is that this distinction is the most con-
troversial aspect of the theory of rationality advocated here. Acceptance
of the distinction entails the falsity of a number of extremely widespread
assumptions that philosophers make when talking about rationality. But
the distinction between these two normative roles cannot be the end of
the story. For, as was argued in chapter 4, we should take the notion of
wholesale rational status as prior to the notion of a reason for action, and
thus as prior to the justifying/requiring distinction as well. The functional
role analysis of reasons offered in that chapter took it for granted that we
had some way of determining which actions were rational, and which
not. So this book would be seriously incomplete without an account of
wholesale rational status. Moreover, chapter 4 also claimed that reasons are
directly relevant only to objective rationality, and not to subjective rational-
ity. Much more remains to be said about these two concepts of rationality.
It is the purpose of the current chapter to address these issues, yielding the
¬nal account of practical rationality. The two ¬nal chapters of the book
will then draw out some implications, and explain how the psychology of
a rational agent is related to the reasons available to her.

ob j e c t ive and sub j e c t ive rat i onal i ty
When we point out to someone that she has done something rude or
boring, we can expect that at least in some real-life cases she will respond
˜So what? I wanted to be rude to that person,™ or ˜Well, caf´ con leche is what
I like.™ But when we point out to someone that she has done something
wrong-headed or irrational, this type of response is not appropriate. If
someone acknowledges that the kind of action she is contemplating is
irrational, then she must also acknowledge that she should not do it “ that

Two concepts of rationality

the reasons in its favor are insuf¬cient to justify it, even by her own lights.
For if the person does offer us a reason for that type of action that she
takes to be suf¬cient to justify it (even: ˜it does one good to act like an
idiot once in a while™), then she does not regard the action as irrational.
Thus, to say that an action is irrational seems to be to say

(1) The action absolutely should not be performed.

On the other hand, there also seems to be a very tight connection between
the rational status of an action, and the mental functioning of the agent
who performs (or would perform) it. That is, to say that an action is
irrational also seems to be to say

(2) If someone performs the action, then something has gone
wrong in the practical mental functioning of that person.

Part of the purpose of this chapter is to argue that one notion of irrational
action cannot stand immediately behind both (1) and (2). Of course, this
claim has already been acknowledged by a signi¬cant number of philoso-
phers.1 But this chapter argues that the relation between actions one should
never do, which I have been calling ˜objectively irrational,™ and actions that
indicate a failure in practical mental functioning, which I have been call-
ing ˜subjectively irrational,™ is of a different nature than is assumed even
by philosophers who acknowledge two senses of rationality. In particu-
lar, subjective rationality cannot be viewed simply as objective rationality
relativized to the beliefs of the agent. Nor can it be viewed as objec-
tive rationality relativized to the beliefs that the agent should have, given
the evidence available to her. Interestingly, it is the distinction between
the justifying and requiring roles of practical reasons that is instrumental
in demonstrating and explaining the inadequacy of these two attempts
to explain subjective rationality in terms of objective rationality. Neither
the distinction between subjective and objective rationality, nor the dis-
tinction between justifying and requiring, are tremendously complex in
themselves. But taken together they yield a view of considerable subtlety
and power. It is the purpose of this chapter to explain and defend the view
that results from the combination of this pair of distinctions.

1 See Brandt (1979), pp. 72“73; Rawls (1971), p. 417; Gibbard (1990), pp. 18“19; Harman
(1982), p. 127; Raz (1999a), p. 22; Cullity and Gaut (1997), p. 2; Scanlon (1998), p. 30.

Brute Rationality

what ac t i on s sh oul d ab s olute ly not b e done ?
What is it we are trying to express when we call an action ˜irrational™ in
the sense expressed by (1)? If actions that are irrational in this sense should,
as a matter of conceptual necessity, never be done, then the question ˜Why
not be irrational?™ should be pointless. For this question to be pointless, it
should be analytic that there can never be a suf¬cient reason or a compelling
argument to perform an action that is understood to be irrational in this
sense, and that there is in fact always a reason not to do it. This gives some
clue as to what we may mean when we say that an action is irrational in
this sense. We may mean that no one could ever sincerely offer anything
as a suf¬cient reason for such an action. Of course, we can insincerely offer
reasons in favor of actions that we regard as irrational, and we can even
say to someone ˜I think you should do this action,™ when we regard the
action as irrational. For we may not care if the person does something
irrational. We may in fact want that person to behave in such a way.
This may be because we dislike the person, or because the action will,
in some way, bene¬t us. But if we are being sincere, we cannot say ˜You
should do the action, for such-and-such reasons™ if we regard the action as
irrational, in the sense given by (1). Now, we are most often sincere in our
recommendations when we are speaking with our friends. Therefore, a
good heuristic in thinking about what it is to regard something as irrational
in the sense given by (1) is provided by keeping in mind that it should in
general not be possible for anyone to recommend an action to a friend
if one regards it as irrational, in this sense.2 But one should not take this
heuristic too literally. The real question of relevance is whether anyone
could sincerely offer considerations in support of the claim that the agent
should perform a particular action. Whether the person is a friend or not
is really neither here nor there: it is simply a mental device to help us
see more clearly whether an argument in favor of performing the action
could actually be offered sincerely. Another way of putting the question
is therefore the following: are there features of the action that someone
could cite to the agent in support of the claim that the agent ought to do
2 This heuristic strongly suggests that formal accounts of objective rationality as the maxi-
mization of the satisfaction of one™s considered and fully-informed preferences are inad-
equate. For one™s friend may be in a state such that her considered and fully-informed
preferences would be for extremely self-destructive things. Nothing in the notion of ˜full
consideration™ or ˜full information™ rules out this possibility. At least this possibility remains
unless these notions illicitly import some substantive criteria: for example, that one cannot
count as ˜fully-informed™ if one™s preferences continue to be self-destructive. See Sen and
Williams (1982), pp. 9“12; Par¬t (1986), p. 500.

Two concepts of rationality

the action, such that the agent would not be puzzled to understand that
the action was being recommended on those grounds?
Note that the general impossibility of arguing sincerely that someone
should do something that one regards as irrational does not imply the
general possibility of offering an argument or reason in favor of any action
one takes to be rational. For not only is it analytic that one cannot offer a
reason that one takes to be suf¬cient to justify an action that one regards
as irrational in this sense, but it must also be true that one takes there to be
a reason against such an action: a reason that places the action in need of
justi¬cation in the ¬rst place. For if there were no such reason against an
action A, then it would be perfectly reasonable to ask ˜Why not do A?™ even
if there were no reason (and thus no argument) for doing A. This is why
twiddling one™s thumbs, doodling, or deciding to walk clockwise around
the block (as opposed to counter-clockwise) are all perfectly rational: it
makes sense to ask ˜Why not do it?™ Now, it may not be possible to
recommend such trivial actions, since sincere recommendation may always
involve having reasons. So if the heuristic offered above “ asking ˜Could
someone sincerely recommend the action to a friend?™ “ were regarded
as providing a de¬nition, such actions would be classi¬ed as irrational.
But someone could certainly allow a friend to do such actions. There
is no reason against them, and as a result, they are perfectly allowable.
Therefore we should widen the heuristic so that such allowable actions are
not considered irrational either.

th e of f i c i al account
The strategy for the of¬cial account of rationality is ¬rst to describe an atti-
tude that one might have towards a possible action. This attitude, which
featured prominently in the preceding section, will be called ˜regarding
an action as irrational,™ and it will correspond to the fundamental sense
of ˜irrational™ given by (1) above. The claim is then made that objective
irrationality is a property of those actions that are of a type that would
prompt this attitude in the overwhelming majority of people. That is, an
action type is objectively irrational if, and only if, it prompts this attitude
in the overwhelming majority of people, and an action token is objectively
irrational if, and only if, it is a token of an irrational action type. The ¬rst
and most important of these two biconditionals preserves an appropriate
vagueness in ˜objectively irrational,™ since it is a vague matter what ˜over-
whelming majority™ means. That biconditional, however, should not be


. 23
( 38 .)