. 14
( 38 .)


which the context is relevant. In fact, Philip™s ˜teleological considerations™ are just rule-
consequentialist calculations of wholesale moral status.

A functional role analysis of reasons

someone is a moral reason. For it is perfectly correct to say of a particular
action that the reason why it is immoral is that it involves deception. This is
an explanatory sense of ˜reason.™ Its use indicates that the deception will
¬gure in an explanation of why the action counts as immoral. On the
rule-utilitarian view outlined above, this means that the deception would
be part of an explanation of why it would be bad if that sort of behavior
were not liable to punishment. But whether or not this justi¬es counting
the fact that an action will deceive someone as a moral reason in some sense,
it should be clear that it is a reason in a sense that does not ¬t into the
framework of weighing.
The requirement that one must be able to use reasons to determine
the rational status of actions does not contradict the claim that wholesale
rational status is theoretically more basic than the concept of a reason. After
all, we typically determine the meanings of particular sentences based
on our prior understandings of the words they contain, and this does
nothing to undermine the theoretical position that it is better to take the
sentence as the basic unit of meaning. True, the relation between particular
reasons and the rational status of a particular action gives the reasons, rather
than the rational status, an explanatory priority. But the relation between
the concept of a reason and the concept of wholesale rational status gives
the concept of rational status an explanatory priority. And when we are
explaining what reasons are, it is this latter relation we are concerned

bas i c reas on s
˜But surely,™ it might be objected, ˜even when we are considering the
rational status of actions, rather than their moral status, there are consider-
ations that are obviously reasons for and against actions that do not have
constant weights, even within one particular normative role. For instance,
sometimes the fact that a particular person is waiting for me might be
a reason for going to the place where she is waiting, and sometimes the
fact that the same person is waiting there might be a reason against going
there. This is true even when the sense of rationality at issue is the rele-
vant objective one. Moreover, this reason provides intelligible motivation,
given appropriate circumstances, both for going to the place where the
person is waiting, and for avoiding that place. That is, someone can act for
this reason. Intuitively, it seems like a reason. We call it “a reason.” And it
meets all the criteria you offer, except for the obvious and drastic failure to

Brute Rationality

meet the “constant weight” criterion. Surely this shows that the “constant
weight” criterion should be abandoned.™
In a sense, this objection is right. But for an important class of reasons “
the class directly relevant to a fundamental normative notion “ it is wrong.
Let us ¬‚esh out the example in the objection. How might it be a reason for
going to a certain place, that a certain someone is there? Well, it may be
that I know that this person will be happy to see me. Of course, there may
be reasons against going to the place. It may be that it would prevent me
from getting important work done. But, if we hypothesize the prospect
of the person™s pleasure in a meeting, it is hard to deny that the fact that I
know the person will be in the place provides me with a reason for going
there. Now, how might it be a reason against going to a certain place, that
that very same person is there? Well, it may be that the circumstances of our
relationship have changed, and that I anticipate nothing but unpleasantness
from a meeting. Of course there may be other reasons for going to the place.
It may be that I need to go there in order to get some work done. But, if
we hypothesize the prospect of great unpleasantness in a meeting, it is hard
to deny that the fact that I know the person will be in the place provides
me with a reason against going there.
The answer to this objection lies in distinguishing basic from derivative
reasons. Using Mill™s methods of difference, we can see that what deter-
mines whether or not I have a reason to go to the place in these cases is
not whether the person is there, but whether it is likely that going to her
location will produce pleasure or pain for her.25 If an action is likely to give
someone pleasure, this is always a reason in its favor, and it is this reason
that stands behind the fact that the person™s presence at the place provides
me with a reason for going there. And if an action is likely to produce pain,
this is always a reason against it, and it is this reason that is behind the fact
that the person™s presence at the place provides me with a reason against
going there. This is not to deny that, given the proper circumstances, it
is a reason to go to a certain place, that a certain person is there. But in
order for this kind of fact (˜that such-and-such person is waiting there™)
to be a reason, there must be another sort of reason standing behind it.
And this supporting reason (˜that I will cause such-and-such person to
have some pleasure if I go there™) does not need anything to stand behind

25 This is not meant to suggest that pleasure and pain are the only reason-giving considera-
tions. Rather, they are the basic relevant considerations in the example.

A functional role analysis of reasons

it. It does not even need the support of my desire to cause that person to
have pleasure, for it a reason whether or not I desire this.26 We can call
these latter reasons “ reasons that do not need any other considerations to
stand behind them “ ˜basic reasons.™ And we can call the former ˜derivative
reasons.™27 When we are using reasons to calculate the rational status of an
action, it is basic reasons that we are concerned with. And these reasons
do make systematic contributions to the rational status of actions.
Although philosophers often use examples of reasons that are nonbasic,
it should be clear that anyone who takes seriously the idea of balancing
or weighing reasons is committed to there being a preferred or basic level
of description for reasons. Otherwise one will end up adding the same
reason into the calculation under a variety of descriptions. That is, in the
example above, one would end up having the following two reasons to
go to a certain spot, where one should have only one: (1) ˜that P will get
pleasure if I go there,™ and (2) ˜that P will be there.™ It is basic reasons that
should lend themselves to weighing, and it is therefore basic reasons that
should make systematic contributions to the rational status of actions. The
objection relies on the failure of derivative reasons to make such systematic

th e f i nal account
Here then is the ¬nal formal account of normative reasons.
FA3 In the sense of ˜rational™ that has to do with objective
rationality, a consideration is a basic reason if and only if:
(1) it corresponds to an intelligible object of human moti-

26 This is easiest to see when one keeps ¬rmly in mind that the type of rationality at issue is
not the rationality of proper mental functioning. Rather, it is the sense of rationality that
is related to claims about whether or not anyone could sincerely recommend the action
to the agent, based on the likelihoods of its various consequences.
27 Another way of trying to achieve the same effect as the distinction between basic and
derivative reasons is with a distinction between complete and incomplete reasons. See
Raz (1999a), pp. 18“35. This strategy does avoid the problem of there ever failing to be
a reason whenever some particular complete reason obtains, even if other circumstances
change. But it is extremely plausible that any one of Raz™s complete reasons becomes
complete precisely in virtue of implying that there is some basic reason to do the action:
that the action will, for example, avoid pain for someone, or save their lives, or give them

Brute Rationality

(2) it plays at least one of the functional roles (i) or (ii), and
has constant strengths, and is comparable to all other
reasons, within and across these roles28
(i) making it rationally permissible to do actions that
would, without it, be irrational, or
(ii) making it rationally required to do actions that
would, without it, be rationally permissible to omit.
If a reason can ful¬ll role (i), then it is said to have
justifying strength. If a reason can ful¬ll role (ii), then
it is said to have requiring strength.
Condition (1), admittedly, makes FA3 less than purely formal. But it is
also quite plausible that it is eliminable. For it may well be that only intel-
ligible objects of human motivation will meet condition (2); it is certainly
hard to think of a counterexample to this hypothesis. (1) is included in FA3
primarily as a reminder of the considerations offered at p. 72. One reason
to eliminate (1), in addition to making the account more formal, would
be to make it more plausible that a similar account might also apply to
reasons for anger, hope, and so on, and especially to reasons for belief. On
the other hand, it may be that on a functional role account of reasons for
belief, there would be some claim parallel to (1), expressing a claim about
reasons for belief corresponding to possible objects of human perception
or credence.
Of course, since FA3 is so formal, it does not tell us which substan-
tive considerations actually are reasons for action. But this is as it should
be for a functional role analysis. In order for FA3 to help us here, we
need some independent way to determine the wholesale rational status of
actions.29 But FA3 will take a substantive account of wholesale rational
28 C1 and C2 above explain how reasons can be compared with respect to strength within
each role. But the relation between justifying and requiring is such that it is also possible
to compare the strengths of reasons across roles. For example, we can say the following:
C3 Given two reasons, R1 and R2 , the requiring strength of R1 is greater than
the justifying strength of R2 iff it would be irrational to perform any action
against which there was a reason with the requiring strength of R1 , and in
favor of which there was a reason with the justifying strength of R2 , and
to which no other reasons were relevant.
29 Internalist full-information accounts of normative reasons might also easily be modi¬ed
to become accounts of wholesale status instead. Modi¬ed in such a way, these accounts
become much more plausible. For it is not plausible that a fully rational agent would have
a desire corresponding to each and every reason applying to his choice, as these accounts
must assume. See Smith (1996). For adherents of such views who continue to want to

A functional role analysis of reasons

status, whatever shape such an account might take, and yield a substantive
account of basic reasons. And, prima facie, the prospects seem better for
the production of an independent account of wholesale rational status
than for the production of an independent account of reasons. For reasons
are pro tanto in nature, and may sometimes elicit no noticeable behav-
ioral or phenomenological response at all. This may happen, for example,
when other relevant considerations provide reasons that are much more
important, whether those reasons support or oppose the weaker reason.
But with regard to the wholesale rational status of actions, there are char-
acteristic motivational and behavioral responses. At least there are such
responses to the status of an action as irrational, and this is all that is needed
to de¬ne ˜rationally required™ and ˜rationally optional.™30 The ˜observa-
tional™ advantage of starting with wholesale rational status is parallel to an
advantage, in accounts of linguistic meaning, of starting with sentences
instead of words. For it is sentences that do things. Single words can do
similar things only in the degenerate cases (˜Run!™) in which they form
a sentence on their own. Similarly, single reasons can sometimes provide
rational requirements or prohibitions. But just as it would be a mistake to
base a linguistic theory exclusively on one-word sentences, it is a mistake
to examine actions to which only one reason is relevant. Such a restricted
view makes it impossible to get a clear view of the justifying role of reasons.
For this role is interestingly manifested only when reasons justify acting
in the face of other reasons.31 Unfortunately, a casual survey of discussions
of normativity reveals an overwhelming preponderance of oversimpli¬ed

i m p l i cat i on s
It would be hard to overestimate the signi¬cance, for contemporary ethical
theory, of a general appreciation of the distinction between the justifying
regard them as accounts of reasons, and not wholesale rational status, chapter 6 offers one
suggestion that will still allow such theorists to distinguish the requiring and justifying
strengths of reasons.
30 The perceptive reader will notice that with this sentence the sense of ˜rationality™ seems
to have changed from objective to subjective. In fact, what there is a reliable motivational
and behavioral response to is neither objective nor subjective irrationality, but something
that might be called apparent objective irrationality. This is enough, however, to give content
to a notion of actual objective irrationality, as chapter 7 explains.
31 In fact, one requires three distinct reasons, yielding three actions corresponding to each
of the three possible opposing pairs, in order to produce examples that demonstrate how
justifying and requiring strength need not co-vary.

Brute Rationality

and requiring roles of practical reasons. For it is this distinction which
allows us to formulate a view of practical rationality that is consistent with
the following two claims: (1) morally required action is always rationally
permissible, and (2) not all immoral action is irrational. It is obviously
desirable to be able to hold the ¬rst of these claims. For if one concludes
that, based on all the relevant reasons, an action is not rationally permissi-
ble, then nothing remains that one could adduce in favor of performing it.
Indeed, one would have to admit that it ought not be performed. These
would be unpleasant things to have to say about a morally required action.
The reason to hold (2) is that otherwise it seems that we should regard
people who perform immoral actions as less than fully rational “ and the
more egregiously immoral, the more severely irrational. Then we will
either have to absolve such people, at least partially, from responsibility
for their immoral actions, or we will have to sever or attenuate the con-
nection between rationality and moral responsibility. Neither of these is
an attractive option. But if we acknowledge the distinction between the
justifying and requiring roles of practical reasons, then it is open to us to
construct a moral theory according to which the reasons that favor any
morally required action are always suf¬cient to rationally justify it, even
though they may not be the sort of reasons that could make it rationally
It is true that the attraction of (2) depends on the view that it would
be a bad idea to sever or attenuate the connection between rationality
and moral responsibility. It is possible to challenge this view, or to hold
that the connection is already rather more attenuated than I represent it as
being. Certainly, the bare fact that an action is irrational is not suf¬cient
to absolve the agent who performs it from moral responsibility. But if
someone performs an action because of a phobia or a compulsion, we do
tend either to excuse her or lessen the degree to which we hold her morally
responsible, should that action be one that would have drawn signi¬cant
moral condemnation if it had been performed by someone without that
mental illness. This is because if a desire or aversion is suf¬ciently strong
to cause one to act irrationally “ and this is a plausible account of when
desires or aversions qualify as compulsions or phobias “ it is reasonable
to regard them as, in a certain sense, ˜irresistible.™32 Now, if we hold that
32 Of course not all of the desires that stand behind such illnesses are literally irresistible “
perhaps none are. A compulsive hand-washer could probably be persuaded to refrain from
washing his hands if his life was threatened, and an addict could probably be persuaded to
defer an injection by similar means. ˜Irresistible™ should be understood, in such contexts,

A functional role analysis of reasons


. 14
( 38 .)