. 32
( 38 .)


and discuss. In such situations, we say that although she appears trustwor-
thy, she is not. But this does not mean that anything is going on here
that is like an object™s appearing blue. Rather, it is simply a way of noting
that the untrustworthy woman elicits behavior or beliefs or feelings in a
way that is appropriate only when dealing with a genuinely trustworthy
On an analysis of the above sort, what would it be for an end achievable
by action to seem good? It would simply be for it to elicit responses that
are appropriate when dealing with an end that genuinely is good. What
are such responses? They are simply desires, or states of motivatedness, or
affective states of that sort. Thus desire is not to be explained by normative
appearances in the strong sense that the judgment thesis maintains. Rather,
the notion of ˜appearing good™ gives us a way of talking about desires
(and other affective states) for things that we judge not to be desirable.
This is consistent with the claim that the notion of ˜appearing good™ also
gives us a way of talking about things that we do (or would) judge to
be desirable, based on their appearance: ˜That peach appears particularly
good.™ But there is a strong case to be made that seems-talk is parasitic on
is-talk, even in cases where ˜seems™ can legitimately be taken to suggest
˜is.™ That is, it is unlikely that we would have seems-talk in a domain
if there were no possibility of misleading perception, wrong judgment,
or inappropriate reaction in that domain. And the claim here is that for
blueness, the relevant possibility is misperception, while for goodness and
other normative terms, the relevant possibility is inappropriate affective
None of this requires us to deny that desires are quite often to be
explained by reference to one™s awareness of something that is in fact good:
for example, by reference to one™s awareness that one™s action will give
pleasure to a friend.11 It is only to deny that desires are to be explained by
reference to the fact that something appears to be good: not even in those
10 I do not mean to suggest that there is a uniquely appropriate pattern of response to
trustworthy people. Rather, there are a wide variety of responses to particular people
that would only have been appropriate had the person been trustworthy, and when we
note that we tend to have a signi¬cant number of these responses, despite our knowl-
edge of the untrustworthiness of the person, we can usefully say that the person appears
trustworthy. This same point goes for other normative properties and the responses they
characteristically elicit.
11 Awareness of something that is in fact good is different from awareness of the fact that
something is good. One of the targets of the current argument is a tendency to con¬‚ate
these two things.

Brute rationality

cases in which the thing that motivates is in fact not good. In both cases,
the desire can be explained by awareness of the fact: whether it is the fact
that the action will help a friend, which does give a reason, or the fact that
the action will hurt someone of whom one is envious, which (arguably)
does not. In the latter case, if we say that the fact that the action would
hurt the person ˜appeared to the agent to provide a reason™ we need only
be taken as indicating that the agent desired to do the action because of
it.12 We need not be committing ourselves to the idea that the prospect
of hurting the other person appeared as having any normative property. One
may wish to object here that desires can clearly be explained by reference to
something™s having appeared to be the case to the agent; perhaps it appeared
to the agent that a certain object was a real peach, when in fact it was only
a convincing fake. But there is no need to deny this, for even in such cases
the relevant judgment would not be a normative one.
One should not be fooled by the use of the same word in ˜appears blue,™
˜appears trustworthy,™ and ˜appears good.™ This mere linguistic similarity
is very weak evidence that anything formally the same is going on in
each case. There are, admittedly, similarities between appearing blue and
appearing good: the linguistic similarity does suggest this much. But it is
very plausible that perceptual appearance is the paradigmatic case, and that
the other cases are derivative, or that some other relation holds between
them. In cases in which an object appears blue, an agent will tend to behave
as if the object really were blue, until the agent comes to believe that it is not
actually blue. And this behavior is to be explained by a phenomenological

12 It seems to me that this is the strongest claim that Dennis Stampe can defend in Stampe
(1987), although it may be suf¬cient for his purpose in that paper, which is to show that
the fact that one desires something gives one a (possibly quite bad) reason to pursue it. He
writes that “desires are reliable indicators of what would be good, and that their authority
involves nothing more, and nothing less” (p. 374). But this ˜nothing more™ means that
their authority does not depend on their involving a presentation of their object as good.
And indeed, I think that Stampe™s “seemsd ” can simply be taken as a misleading technical
replacement for ˜is desired.™ I don™t deny that desire shares many points of similarity with
perception “ including being a fairly reliable indicator of something “ but taking the point
as literally as Stampe does forces him to look for an existing object of perception, which
he ¬nds in one™s own physiological state. This does not sit well with his claim that the
desire that p is a state ideally caused by the fact that it would be good that p. For suppose I
desire my daughter to marry well. This desire, ideally, would be caused by the fact that it
would be good for her to marry well. But the desire is caused by some physiological state
of mine. In order to avoid the obviously false claim that the goodness of my daughter™s
marrying well actually is a state of my own body, Stampe instead makes the claim that the
relevant state of my own body is such that it would be good if my daughter married well.
See Stampe (1987), pp. 355, 372“75.

Brute Rationality

experience that the agent is having, by way of his faculty of sight. Similarly,
in cases in which a person appears trustworthy, or an action appears good,
the agent tends to behave as if the person were really trustworthy, or the
action really good, until the agent comes to believe that the person is
not really trustworthy, or the action is not really good. This similarity is
suf¬cient to explain the use of the word ˜appears™ in both types of case. But
in the latter case the antecedent behavior “ believing what the seemingly
trustworthy person says, taking her advice, etc. “ is not to be explained
by any phenomenological experience the agent is having, distinct from
the experiences of sight, sound, and so on. Rather, it is to be explained
by the following fact: that we humans are set up to respond more or less
automatically to certain perceptible features of the world, including other
people™s tone of voice and facial expression.
A defender of Scanlon may wish to point out here that objects appear to
us as cups or trees all the time, and that this fact does nothing to imply that
we have a special faculty that perceives cups or trees. Nor does it suggest
that there is any distinctive phenomenology of cup or tree perception.
When we say that something appears as a cup, this can simply mean, for
example, that the object looks like a cup would be expected to look under
normal conditions, or that the way it looks would lead one to believe that
it is a cup.13 Why cannot ˜appearing to be a reason™ be modeled on this sort
of appearance? Why need it be compared to appearing to have some basic
phenomenal quality like blueness? The answer is: let ˜appearing to be a
reason™ be modeled on as complex a sensory appearance as one likes, it will
not help Scanlon. For, as William Alston points out, when we understand
˜appears as™ in these ways, “a phenomenal concept is, so to say, always in the
background, even when not explicitly employed.”14 Suppose then that we
interpret ˜appears as a reason™ as ˜appears as a reason would appear under
normal conditions.™ What is the phenomenal concept in the background?
It does not seem that there always is one. Sometimes, perhaps, a reason
elicits a felt desire, or some other felt affective response. Perhaps sometimes
there is even a distinct phenomenology of something appearing to be a
13 These interpretations of ˜appears as™ are what William Alston calls, respectively, the com-
parative and the doxastic. He also describes an epistemic concept, and presumably there are
a good number of other ways of interpreting our ˜appears as™ claims. See Alston (1991),
p. 45. It may be worth noting that the doxastic interpretation would not be an attractive
way for Scanlon to understand the normative appearances that he claims stand behind
desires, since children have desires before they have suf¬cient conceptual apparatus to
form a belief that something is a reason.
14 Alston (1991), p. 45.

Brute rationality

reason: a felt experience that one would wish to describe in exactly this
way. But it is just implausible to assert that this sort of experience precedes
or underlies all desire, or even all rational desire.15
Now, in attempting to understand what the normative judgments might
be that Dancy, Quinn, Raz, and Scanlon have in mind, it is not extremely
helpful merely to be told that they are not to be understood by strict
analogy with perceptual appearances. But surely the burden here is on
these philosophers to explain what they mean when they speak of an
agent™s taking a fact to show an action to be good, or an agent™s conceiving
a consideration as favoring an action.16 Perhaps Scanlon has done so, in
offering his analogy with perception. But that analogy cannot be taken
strictly. So it remains for Scanlon to explain how the analogy is actually
to be taken. In what follows it will be assumed, for the sake of argument,
that there is some plausible understanding of ˜normative judgment™ that
makes such judgments less explicit than occurrent thoughts, but also gives
them a distinctively normative content. And it will be argued that no such
normative judgments are required for rational action, or for intentional
action more generally.

th e bas i c p i c ture
The view offered in this book does not merely dispute the claim that ratio-
nal action requires agents to make correct normative judgments. It holds that
rational action, and intentional action more generally, does not require the
making of any normative judgments. While agreeing that virtually all of
our desires, even at the most basic level, are held for reasons, and that
virtually all of our actions are undertaken for reasons, I also hold that it
is quite common for human action and desire to involve no normative
judgment, even of an unconscious, inexplicit, inchoate sort. Rather, it is
suf¬cient for one to act for a reason that there be a reason to act, and that
one act because of it “ at least if the ˜because™ here is interpreted as indicat-
ing the presence of the right sort of causal or psychological mechanism.17

15 Stampe (1987), p. 359 seems to appreciate the truth of this point, but not its signi¬cance.
16 These phrases come from Raz (1999b), p. 24 and Dancy (2000), p. 129.
17 This quali¬cation is necessary to deal with the problem of wayward causal chains. To take
a well-known example, I may become so unnerved by my sudden awareness of a reason
to kill my mountain-climbing partner that my hands begin to shake and I end up letting
go of the rope, killing him. This should not count as an action done for a reason. The
problem of characterizing the ˜right way™ in which the reason must be a cause of my

Brute Rationality

Similarly, it is suf¬cient for one to desire something for a reason that there
be a reason to desire it, and that one desire it because of that reason. In
neither case need one judge that the reason is a reason. One need not even
have the concept of ˜a reason.™ And as long as the reasons for one™s actions
or desires are suf¬cient to justify them, that is all it takes for those actions
or desires to be rational.
Here is the picture of reasons and rational action that will be set up
against the judgment thesis. Although the ¬rst two claims, which concern
normative reasons, are part of the developed theory worked out in previous
chapters, the ¬ve points taken together are by no means intended as a full-
blown theory of rational agency. Rather, they are offered as a framework
within which one might develop such a theory.18
(1) There are facts of the matter about whether or not there are reasons
that speak in favor of particular actions, and facts about what those reasons
are. For example, it is a fact that the prospect of having one™s ¬ngers burnt
provides a reason to avoid touching a certain very hot object, though
it can be rationally permissible to act against such a reason, if there are
countervailing reasons of suf¬cient justifying strength.
(2) Reasons for actions are, at least generally, completely independent of
the desires or other noncognitive attitudes of the agents who have them.
This claim and the previous one express views that are shared by Dancy,
Quinn, Raz, Scanlon, and myself. Again, because this chapter is directed
primarily at these philosophers and at those who hold similar views, no
additional defense of these ¬rst two claims will be offered here. This chapter
does not attempt to persuade Humeans to abandon their desire-based
views, but only to convince those who already reject such views that they
should take a further step. Nevertheless, in order to clarify the view being
offered, it may be worth noting that claim (2) goes signi¬cantly beyond
claim (1). Claim (1) is consistent with the view that the reasons that speak
in favor of a particular action are somehow determined by the desires of
the agent “ either as she is, or in some ideal state. Claim (2) denies such a

action is beyond the scope of this book. See Davidson (1980), p. 79 and Mele (2000).
It may be worth noting that simply adding an appropriate normative judgment to the
causal or psychological mechanism does nothing to eliminate the possibility of these sorts
of wayward chains.
18 For readers worried that no plausible theory will ¬t into the basic picture, it may be worth
mentioning that one of the most in¬‚uential causal accounts of intentional action, Alfred
Mele™s, is fully consistent with the view advocated here.

Brute rationality

(3) A desire may be understood in an intuitive and very broad way as a
disposition to act so as to bring about certain states of affairs. We can call this
state of affairs the end of the desire. ˜End™ should be as broadly understood
as ˜desire™ here, so that it can include, for example, the state of affairs of my
avoiding some pain, or the state of affairs of my acting as God commands.
This particular picture of desire, however, is not at all crucial to anything
that follows. Those who favor some alternate account of desire should be
able to substitute their own favored picture, modi¬ed in relevant ways,
into the basic view I am here presenting. In particular, those who wish to
regard desire as a state of motivatedness, where the motivation comes from
the antecedent apprehension of some fact, should feel free to do so. Such
states of motivatedness will still have ends, as they are understood here.
What is important is that ends are states of affairs that would be realized or
promoted by acting successfully on a desire. Whether one has the desire
because of one™s logically antecedent perception that such ends could be
achieved by one™s action, or whether desires are somehow brute, is neither
here nor there for current purposes. The only accounts of desire that are
incompatible with these purposes are accounts that tie desires conceptually
to normative judgments.
(4) If an agent acts on a certain desire, then the end of that desire “
something quite distinct from the desire itself “ forms part of the reason why
the agent acted as he did.19 One and the same thing can be both a reason for
action, and a reason why an agent does an action. In fact, it is typically true
that the reasons why an agent acts are also reasons for acting in that way. But
this does not mean that there is only one notion of reason in play. For one
thing, it is not typically true that all the reasons for an action are reasons why
an agent acts. One important reason for this is that there can be reasons of
which the agent is completely unaware, or which the agent has insuf¬cient
conceptual apparatus to appreciate. And even when we restrict attention
to the reasons of which the agent is aware, generally there will be too many
for an agent to take them all into account. Moreover, it seems possible that
a language could have the means to express the notion of a reason why,
without having the means to express the notion of a reason for. It is only the
notion of a reason for that is normative.20 In giving the reason why an agent
19 ˜Part of,™ only because there may be more than one desire involved. In cases in which
an agent clearly acts because of one and only one dominant desire, then the end of that
desire simply is the reason why the agent acted as he did. The reasons why an agent acts
as he does are always to be given in terms of the ends of the agent™s relevant desires.
20 That the notion of ˜reason for™ is a normative notion is consistent with the claim that
particular reasons for action are themselves nonnormative matters of fact. For example, it

Brute Rationality

acted as he did, one is neither making a normative claim, nor imputing
one to the agent. One is only making a claim about what motivated the


. 32
( 38 .)