. 2
( 38 .)


exactly the same question. In some cases there will be no good answer to
this question, and then the recommendations that ¬‚ow from the princi-
ple may lose their authority. There is, however, a signi¬cant philosophical
tradition according to which this sequence of principles and questions,
and more basic principles and further questions, cannot go on forever.
At some point, after the articulation of one of these principles, it will no
longer make sense, or be appropriate, to ask ˜But why should I follow

Brute Rationality

that principle?™ That is, there is a philosophical tradition that asserts the
existence of a fundamental normative principle applicable to action. Perhaps this
tradition goes back as far as Aristotle, who asserted that there was one gov-
erning end according to which all human action was to be judged. Hume,
when he argued that reason cannot by itself direct the will, was reacting
against the majority opinion of his contemporaries, according to whom
reason could do so. That is, the philosophers against whom Hume was
arguing held that if it could be shown that reason required or prohibited
an action, that was the end of the practical argument about that action: no
further appeal could possibly be made that could legitimately alter such
a judgment. Kant also is a prominent member of this tradition, advocat-
ing the existence of a categorical imperative that tells one how one must
act, and against which no further consideration can have any legitimate
Contemporary philosophers also defend the existence of a fundamen-
tal normative principle, or set of principles. Indeed, this is the sense of
˜rational™ that is central to contemporary ethical theorizing. For example,
Stephen Darwall writes that “It is part of the very idea of the [rationally
normative system] that its norms are ¬nally authoritative in settling questions
of what to do.” Thomas Nagel writes that it should not be possible to ask
why one should do what one has reason to do, and that for this reason
there cannot be a justi¬cation for acting rationally. And Allan Gibbard™s
notion of rationality “settles what to do . . . what to believe, and . . .
how to feel.”1 According to all of these philosophers it is a conceptual
truth that there cannot be a suf¬cient reason to act irrationally and that
there is a reason not to do so. Therefore, according to these philosophers,
the question ˜Could I have a suf¬cient reason to do an irrational act?™ is
as misguided (or trivial) as the question ˜Could there be an unmarried
When we are presented with any proposal regarding this fundamental
normative principle, there are two tests we can apply to see whether
it is adequate. The ¬rst is to see whether the question ˜Why should I
always follow that principle?™ makes clear sense. If it does make sense,
this casts the fundamental nature of the principle into doubt. For the

1 Darwall (1983), pp. 215“16; Nagel (1970), pp. 1“9; Gibbard (1990), esp. p. 49. See also
Korsgaard (1996a), p. 104 and Smith (1994), pp. 150ff. Smith claims that it is all and only
reasons which spring from the norms of rationality that make actions desirable. Of course
there are other conceptions of rationality. Robert Nozick (1993), pp. 40, 117 is concerned
with the human faculty of rationality, and is content to ask about its purpose or function.

Conditions on an adequate theory

principle is supposed to the most basic “ the principle that stands behind
all others. If the above question makes sense, then the putative fundamental
principle certainly is not wearing its fundamental nature on its face. That
is, it does not appear to be the end of the normative road. The second
test is to see whether one could ever sensibly offer reasons for acting
against the principle. If this is a real possibility then the principle cannot
be the fundamental principle that tells us how we ought always to act.
To illustrate these tests, it may be useful to use them to disqualify one
possible fundamental normative principle: always act so as to maximize
the satisfaction of your preferences.2 Does it make sense to ask ˜But why
should I always act so as to maximize the satisfaction of my preferences?™
Yes, it does. For one could elaborate the question in this way: ˜Why should
I always act so as to maximize the satisfaction of my preferences, if I know
my preferences are the result of a brain defect that tends to produce self-
destructive preferences?™3 This failure to pass the ¬rst test is related to the
way in which the proposed principle will also fail the second test. For one
way of sensibly offering a reason to act against the principle is to say ˜But
if you follow this principle you will cause yourself a lot of pain, without
any bene¬t.™
The second of the above tests is quite clearly one which a fundamental
normative notion must pass. If there can be an adequate reason to act
against a principle, that principle cannot be telling us how we ought
always to be acting. The ¬rst test, however, is more slippery, and it may
be useful to show how a principle may pass it without at ¬rst seeming to
do so. Consider then the following:
One should never perform an action that will harm oneself unless it will bring
some compensating bene¬t to someone (perhaps oneself). All other actions are
rationally permitted.

It seems obvious that one could sensibly ask ˜Why should I always follow
this principle?™ One reason it seems obvious is that there seems to be
an answer. For example, one might offer ˜Because then one will avoid

2 It is unclear if any contemporary philosophers hold such a simple version of this view. But
the criticisms offered here also tell against more sophisticated versions of such principles.
For a very clear presentation of these criticisms see Ripstein (2001).
3 This is not the place to descend into arguments about the various ways in which one might
patch up the suggested principle. But it is worth mentioning that the strategy of ruling
out desires that are the result of, say, a brain defect, is not a simple one. For we use the
notion of rational action in determining what counts as a brain defect, rather than (say) a
statistically rare con¬guration of neurons.

Brute Rationality

suffering harms.™ But in fact that is not an answer, since it is false that if
one successfully follows this principle, one will necessarily avoid suffering
harms. This is because the principle permits one to suffer harms in cases in
which one will thereby produce compensating bene¬ts for someone else.
One might then suggest the following amended answer: ˜Because one will
avoid suffering harms, except in cases in which one will thereby produce
compensating bene¬ts for someone.™ What is important to see is that with
this amended answer one has ceased to offer a further reason to obey the
principle. One has simply pointed out that by following the principle one
follows the principle. Of course, this brief discussion has not shown that
the above principle actually does pass the ¬rst test. It only shows one way
in which a principle may misleadingly appear to fail it. Moreover, though
the above principle may in fact pass this one particular test, it may be
inadequate for other reasons.
This book is part of the tradition that seeks to discover and defend a
fundamental normative principle applicable to action “ of course by some
means other than the production of a still more fundamental principle.
That is, it seeks to provide an account of a principle that passes both of
the tests mentioned above. It is devoted entirely to this principle, and not
to its employment in arguing for further normative claims. In particular,
no moral view is advocated, although it will be clear that the account has
signi¬cant implications for the development of moral views.
In the phrase ˜fundamental normative principle,™ the word ˜fundamen-
tal™ should not be taken to mean ˜most important.™ For there are many
other normative principles that, in different contexts, are likely to be
more important and more salient than the principle that is the central
topic of this book. Of course, we should never follow these more salient
principles if they can be shown to violate the fundamental one: that is
part of what it means for it to be fundamental. Another part of what it
means is that if it is clear that an action does not violate the fundamental
principle then there may be nothing we can say to dissuade even a ratio-
nal agent from performing it “ the agent may remain perfectly rational
in resisting all our arguments. As will become clear later in the book,
this means that the fundamental normative principle gives agents a very
wide scope in making decisions about how to act. Because of the lack
of guidance that the principle provides, some may be tempted to think
that it cannot really be fundamental. But that is to confuse being funda-
mental with being most generally useful, or with being salient. It will turn
out that, because we almost always act rationally without having to think

Conditions on an adequate theory

about it, the fundamental normative principle will very rarely be of much
use in particular decisions. It will not tell us, for example, which career to
choose, or whether to marry, or to have children, or whether to pursue
wealth over enlightenment. As I will argue in various ways in what fol-
lows, it will not even tell us whether to take the high moral road, or the
low. These questions we must answer for ourselves “ they are choices, and
it is futile to search for a basic principle that will authoritatively hand us
the correct answer. In a limited number of cases I have found that when
people are tempted to act against the fundamental normative principle,
it is sometimes effective simply to point this out. This tends to bring the
real source of the temptation into clearer focus, which helps in resisting it.
But the primary usefulness of a clear view of the fundamental normative
principle is not “ at least directly “ practical. Rather, it is theoretical: the
principle will ¬gure in an explanation of what it is for an action to be
rational, in a sense that is closely connected with mental functioning. This
notion, in turn, is often indispensable in restricting the scope of ˜every-
one™ as it is used in philosophical theories (such as contractualism). The
principle will also play a role in explaining why we should want to be
rational, in that sense. And of course the fundamental principle will have
many indirect practical implications, for very often such a principle plays
an obvious and central role in the development of moral theory. And a
moral theory, if it is clear, may have signi¬cant practical implications for
people who care about morality.

rat i onal i ty and m e ntal f unc t i on i ng
It seems fairly clear that whatever the fundamental normative notion might
be, it will use the facts about one™s situation in yielding its judgments. This
is why, when we are trying to decide how to act, we do not simply rest
content with our present beliefs or evidence about the consequences of
our actions, but seek out additional relevant information. Seeking this
information is part of the process of ¬guring out what to do. Sometimes,
through no fault of our own, we may fail to get the correct information,
or may form justi¬ed, but false, beliefs. Because of this we may often fail
to discover what we ought to do, and consequently we may fail to do what
we ought to do. In failing to act according to the fundamental normative
principle in such cases, we are not to be blamed. Nothing has gone wrong
in the mental processes that produced our action. We would not want to
call such actions ˜irrational,™ if we were taking irrational action to count

Brute Rationality

against the rationality of the agent in a way that was relevant to questions of
moral responsibility, competence to give consent, freedom of will, mental
health, and so on. Similarly, we may sometimes perform an action that
is permitted according to the fundamental normative principle, given the
facts; however, given our beliefs, it may be that our performance is obviously
the result of some mental malfunction. In such cases we may want to call
the action ˜irrational,™ if we are concerned with these same questions of
moral responsibility, competence, and so on.
Since there may often be adequate (but unknown) reasons to perform
actions that would be irrational in this ˜mental functioning™ sense, it should
be clear that the ˜mental functioning™ sense of rationality is not the fun-
damental normative sense. It fails the second test. Nevertheless, it should
be equally clear that the two senses of rationality are very closely related.
But there is an interesting puzzle that one encounters in trying to specify
exactly how they are related. It is very tempting to think that the ˜men-
tal functioning™ sense of rationality is nothing but the fundamental sense,
relativized to the beliefs of the agent, in place of the facts of the case.4 But
this cannot be correct. For it may be that an action would be rational, in
the fundamental sense, if the world were as my beliefs represent it, and yet
it may still be that my performance of the action would be irrational in
the ˜mental functioning™ sense. This may happen because I conspicuously
lack a belief that I should de¬nitely have: the belief that my action will
cause me a great deal of suffering, for example. I may refuse to believe
this, although I have more than enough evidence to believe it, because
it may be that the suffering will be caused by someone I love, and I may
deceive myself into thinking that the person would never hurt me. In such
a case my action would be rational, in the fundamental sense, if my beliefs
accurately represented the world. But it is nevertheless irrational in the
˜mental functioning™ sense. The next obvious strategy would be to de¬ne
rationality in the ˜mental functioning™ sense in the following way: it is
simply the same as the fundamental sense, but relativized to the beliefs that
the agent should have, given the available evidence.5 But this strategy also
fails, perhaps even more spectacularly. For it may be that an action would
be rational, in the fundamental sense, if the world were as I should believe
it to be, and yet it may still be that my performance of the action would
4 See Brandt (1979), pp. 72“73; Gibbard (1990), pp. 18“19; Harman (1982), p. 127; Raz
(1999a), p. 22.
5 This de¬nition follows a pattern used by Rawls in de¬ning what he calls ˜subjective
rationality™ in relation to what he calls ˜objective rationality™. See Rawls (1971), p. 417.

Conditions on an adequate theory

be irrational in the ˜mental functioning™ sense. How could this happen?
It may be that, though I should believe that a certain unpleasant action
will bene¬t me greatly in the long term, I do not actually believe it. In
such a case, the fact that the action will bene¬t me (and that I should
believe this) does nothing to mitigate the irrationality of performing it, if
it would be irrational to do so in the absence of the future bene¬ts. So
two initially plausible accounts of the relation between the two senses of
rationality are completely inadequate. And it is obvious that one cannot
simply relativize to the set of beliefs that one does or should have, for this
will typically be a set of inconsistent beliefs. Nor can one relativize to the
beliefs one does and should have, for if one believes that a certain action
will be quite painful, and will bene¬t no one, then it would be irrational
to perform the action, even if one should not have this belief.6 This book
provides an account of the relation between the ˜mental functioning™ and
˜fundamental™ senses of rationality in a way that not only avoids coun-
terexamples, but also explains why the above relativizing de¬nitions fail,
and why they fail in the particular ways they do.
The ˜mental functioning™ and ˜fundamental™ senses of rationality are
often distinguished by calling the former ˜subjective rationality™ and the
latter ˜objective rationality,™ and this is the terminology I will use in this
book.7 But quite often philosophers do not distinguish the two senses at
all. And sometimes the fundamental sense is the only sense of rational-
ity that is of¬cially recognized, so that the phenomena captured by the
˜mental functioning™ sense end up being described with phrases such as
˜rational, relative to the beliefs of the agent.™8 In earlier writing I some-
times borrowed a piece of terminology from Allan Gibbard, who uses the
term “advisable” as a label for “[w]hat it makes sense to do objectively,
in light of all the facts” “ that is, for what I am calling ˜objectively ratio-
nal™ action.9 Gibbard™s terminology has the advantage of minimizing the
risk of thinking that the objective notion has much to do with mental
functioning directly. However, I now prefer the terms ˜subjective ratio-
nality™ and ˜objective rationality™ because, despite the fact that a perfectly
(subjectively) rational person might often perform objectively irrational
actions, it is uncontroversial that there is a very close connection between
subjective and objective rationality. Using the two terms ˜rationality™ and
˜advisability™ wrongly lends an air of plausibility to objections that depend
6 See Cullity and Gaut (1997), p. 2.
7 See Rawls (1971), p. 417; compare Gibbard (1990), p. 89.
8 9 Gibbard (1990), p. 89.
Williams (1981), p. 103. See also Sobel (2001).


. 2
( 38 .)