matters of fact, as that an action is the only means of avoiding getting
oneā™s feet wet.21 In the following discussion the bare term ā˜reasonā™ should
always be understood in the normative sense, as meaning ā˜reason forā™:
what Dancy calls ā˜a good reason.ā™22 In order to avoid confusion, the terms
ā˜motive,ā™ ā˜end,ā™ ā˜motivating reason,ā™ or ā˜explanatory reasonā™ will generally
be used instead of ā˜reason why.ā™ Also, when the phrases ā˜reason whyā™ or
ā˜explanatory reasonā™ are used, they should be interpreted in a stronger
sense than merely ā˜causally explanatory.ā™ Rather, they indicate reasons that
play the right sort of causal or psychological role in the production of the
(5) By the time a normal human being has grown to adulthood, she will
have acquired the concept of a reason. Normal adults can, unsurprisingly,
make wrong judgments about reasons, just as they can make wrong judg-
ments about the colors of objects, but typically their judgments are correct.
Moreover, although there are disagreements about whether something is
a reason, or about how much it can justify or require, these disagreements
are as inevitable, marginal, and conceptually unimportant as disagreements
as to whether something is green or blue. This is shown by the fact that
almost all of us can see the reasons for almost all of the actions that almost
all people actually perform, whether we approve of them or not.23
is a reason for taking some medicine that it will stop the pain. That this fact has normative
signiļ¬cance does not mean that it itself is a normative fact. Normative facts are of the
following sort: that pain is bad, that providing pleasure to others provides a reason, etc.
21 Or again, if one likes, what typically motivates an agent is her awareness of, or beliefs about,
such nonnormative matters of fact. For current purposes there is no need to enter this
22 Dancy (2000), p. 4. Dancy would perhaps deny that there are two senses of ā˜reason,ā™
holding instead that there are two sorts of questions that reasons are intended to answer.
Since I do not dispute Dancyā™s claim that normative and explanatory reasons are the same
kind of entity ā“ states of affairs ā“ this disagreement is relatively unimportant.
23 In fact, the general correctness of our reasons-beliefs is not essential to the argument, since
it only bears on how many of our actions are rational, and not whether our desires are
produced by these beliefs. Nevertheless, see Scanlon (1998), p. 71 for similar skepticism
about widespread disagreement as to what counts as a reason. In order to appreciate
this point it is important to realize that agreement about what counts as a reason, and
agreement about how much such reasons can justify or require, is perfectly consistent
with widely divergent tendencies to act on those reasons, even among rational people.
This is because normative reasons are plausibly regarded not as uniquely ļ¬xing a rational
choice (or small range of optimal choices), but rather as setting limits on a relatively wide
range of rationally permissible options. See Raz (1999b), p. 100.
Here is how the basic picture describes some typical rational actions.
One reason for desiring to eat ice cream, and for eating it, is that one will
get some pleasant sensations by eating it. If a three-year-old desires to eat ice
cream, and eats it, because he knows he will get those pleasant sensations,
then that child acts for a (motivating) reason, and also has a (motivating)
reason for its desire. Since that motivating reason is also a normative reason,
we can say that the child acts for a good reason. If this reason is unopposed
by any other reasons of which the child is or should be aware, and which
would make it irrational to eat the ice cream, then the action is rational. If
a ļ¬fty-three-year-old desires to eat ice cream, the same things are also true.
In this latter case the ice cream eater, being older and wiser, does indeed
have normative concepts, and would probably immediately assent to the
claim that pleasure was good, and that the pleasure of eating ice cream
provides a reason to eat it. But these facts about what the older agent would
immediately assent to are the result of his having, unsurprisingly, certain
concepts and beliefs. Such an agent would also probably immediately assent
to the claim that ice cream is a dairy product, or, for that matter, that
chickens are animals. But none of these beliefs has anything to do with
the genesis of his desire to eat ice cream, or with his eating it, or with the
fact that his eating it is a perfectly rational action. For philosophers who
are disposed to believe that desires are somehow the result of normative
judgments, the immediate assent that most adult agents would give to the
relevant normative judgments will seem to conļ¬rm their opinion.24 But
this immediate assent is also explicable as the result of motivationally inert
basic knowledge of what is good and what is bad.
e x p la i n i ng th e j udg m e nt th e s i s
Why are Dancy, Quinn, Raz, and Scanlon all drawn to the judgment thesis?
There is no uniļ¬ed answer to this question. Each philosopher says slightly
different things in its defense. But very often one of them can be seen as
ļ¬lling a gap in anotherā™s argument, or as answering a further objection. The
following four sections attempt to do three things: to present the relevant
arguments, with some indication of how they reinforce each other, to
explain how one might come to hold them, and to criticize them, lending
support to the basic picture outlined above.
24 See Raz (1999b), p. 233.
Quinn endorses the judgment thesis at the end of an argument against
the view that desires, understood as bare functional states, could possibly
justify action.25 His argument makes use of the example of a person who
is disposed to turn radios on whenever he sees them off. This person sees
nothing good in doing this, but is quite strongly disposed to do it anyway.
Quinn reasonably expects his readers to agree that such a person has no
reason for turning on radios simply in virtue of being in this functional
state. Perhaps the person might feel uncomfortable until he turned on a
visible radio. And in such a case he might have some reason to do so. But
then it would be the prospect of relieving the discomfort that provided
the reason, and not the stipulated end of the desire, which is simply to
have the radio switched on. Quinn then goes on to point out that if the
bare disposition to act in a certain way does not provide a reason in the
bizarre cases, then it does not provide a reason in the normal cases either.
Rather, in the normal case, in which an agent is disposed, for example, to
promote his own health, it is not the disposition that provides the reason
to do so, but something else: something to do with the goodness of the
prospect of increased health. It is at this point in the argument, up to which
I wholeheartedly agree, that Quinn makes an additional unnecessary move
and commits himself to the judgment thesis. Here is the relevant passage.
That I am psychologically set up to head in a certain way cannot by itself rationalize
my willā™s going along with the set-up. For that I need the thought that the direction
in which I am psychologically pointed leads to something good (either in act or
result), or takes me away from something bad.26
The ļ¬rst sentence here is correct. Simply being set up to switch radios
on does not give one a reason to do so. Nor does being set up to pursue
pleasure and health give one a reason to pursue these things. Something
else is needed. But Quinn adds the wrong thing. It is not the thought that
the end of my desire is good that makes action rational; it is the fact that
25 Quinn (1995), pp. 189ā“95.
26 Quinn (1995), p. 195, italics Quinnā™s. It is possible to give a de re reading to the content of
the thought referred to in the second sentence. Then the thought would be the following:
ā˜the direction in which the agent is psychologically pointed leads to something which is,
in fact (but perhaps not in the agentā™s thought) good.ā™ But this reading is inconsistent with
Quinnā™s repeated assertions elsewhere that āsome kind of evaluationā is ātypically present
in basic desireā (p. 200) and that ādesires and preferences rationalize only because of the
value judgments they involveā (p. 201).
the end of my desire is good.27 That fact alone is enough to explain the
difference in rational status between actions based on a desire to switch on
radios, and actions based on a desire to maintain oneā™s health. The claim
here is not that merely having a genuinely good end is sufļ¬cient to make
oneā™s action rational. Surely it is not, since one may well be irrational
in pursuing even a good end. Rather, the claim here is only that what
accounts for the difference in rational status between switching radios on
and pursuing health need not be the presence of a normative judgment.
Rather, one can give a sufļ¬cient explanation for the difference in status in
this case simply by reference to the pointlessness of the one end, and the
goodness of the other.
Quinnā™s own strategy for arguing against the normative signiļ¬cance of
desire works just as well against his own view that normative judgments
have normative signiļ¬cance. Consider a person who not only desires to
switch radios on, but who also makes the normative judgment that it is
good to do so. Suppose that he is totally sincere in making this judgment,
and that all the evidence supports attributing it to him. Suppose also
that this judgment is basic: i.e., that it is not based on false beliefs about
the usefulness, for some further ends, of switching on these radios, but
that it is a judgment that switching radios on is simply, in itself, good.
This agent and his actions do not seem any more rational than the agent
who merely has a bare desire to switch radios on. So the basic normative
judgment does nothing to rationalize action in this case. Thus, following
Quinnā™s reasoning, even when basic normative judgments are correct (as, for
example, that getting pleasure is good), it is not the normative judgment
that rationalizes action in accord with those judgments. To paraphrase
No normative judgment can, by itself, make the contribution to rationalizing
action that adherents to the judgment thesis suppose it to have. This is true even
if the judgment is correct: i.e., that pleasure or health are good. For pleasure or
health provide a point to their pursuit that does not consist in the fact that they
are judged to have a point.28
27 One should not be misled by the language of this sentence into thinking that the current
issue is whether one should regard beliefs or facts as the kind of thing that motivates us.
That is emphatically not the issue. For even if one holds that it is the belief that jogging will
promote my health that motivates me, it is still the fact about the goodness of health that
makes the motivation rational. And I need make no corresponding normative judgment
for it to do so.
28 This is a modiļ¬cation of a passage from Quinn (1995), p. 195.
In line with this paraphrase, Quinn does in fact seem to deny that a bare
normative judgment is itself sufļ¬cient to rationalize desire or action. That
is, he considers cases in which one makes a false normative judgment ā“
assessing some end as choice-worthy when it is not ā“ and acts on the
corresponding desire. And what he says of the resultant choice is that it is
intelligible, but not rational.29 But why should we postulate the existence
of a normative judgment that mediates between the perception of an
end, whether it be choice-worthy or not, and an action, if the difference
between merely intelligible and actually rational action depends not on
the judgment, but on the choice-worthiness of the goal? One answer to
this question might be that the normative judgment is required to make
the action intelligible, and that this is a necessary condition on its being
rational. We will assess this suggestion when we discuss Joseph Raz, but
since Raz also picks up a thread of argument started by Jonathan Dancy,
let us turn to Dancy ļ¬rst.
Dancy holds that ā[w]e can normally explain an agentā™s doing what he did
by specifying the reasons in the light of which he acted.ā30 Understood in
a certain way, there is nothing objectionable in such a claim. But Dancy
further holds that ā[i]t is required for this sort of explanation that those
features be present to the agentā™s consciousness ā“ indeed, that they some-
how be conceived as favoring the action.ā31 Similarly, Dancy writes that
āthe explanation of action, at least that of intentional action, can always be
achieved by laying out the considerations in the light of which the agent
saw the action as desirable, sensible, required.ā32 It is these further claims
about what is involved in being motivated by a reason that this chapter is
Why does Dancy endorse the judgment thesis? It may initially be smug-
gled in with the technical phrase ā˜in the light of ā™. Dancy deļ¬nes this notion
early in Practical Reality as the relation between an agent and the reasons for
29 30 Dancy (2000), p. 5.
Quinn (1995), p. 201.
31 Dancy (2000), p. 129. It is only the second part of this claim that will be disputed.
32 Dancy (2000), p. 136. There is a way of reading the āalwaysā in this claim so that Dancy
is specifying only a sufļ¬cient condition for motivational explanation. In light of other
claims, however, it seems most consistent to read the claim as meaning ā˜in principle, it is
always possible to explain intentional action by laying out the considerations in the light
of which the agent saw the action as desirable.ā™
which the agent acted.33 Since, for Dancy, reasons are not psychological
states, this relation is between the agent and (typically) an external state of
affairs, as that it is very likely to rain. The relation holds whenever the state
of affairs motivated the agent. So to say that an agent acted in the light of
x is merely to say that x motivated the agent to act. Deļ¬ned in this way,
the phrase ā˜in the light ofā™ is unproblematic. Moreover, it serves a useful
purpose, since acting in the light of some consideration is different from
acting (partly) because one believes that some state of affairs obtains. For
example, and as Dancy mentions, it turns out that people are much more
likely to give to charity if they believe that others are doing the same. Yet it
does not generally seem appropriate to say that the fact that other people
are giving to charity is an agentā™s reason for donating, or that an agent
is motivated by it. That other people are giving does have an inļ¬‚uence,
and this inļ¬‚uence involves something psychological, but it is insufļ¬cient
to ground the claim that the agent acted for the reason that other people
were donating ā“ that the agent acted in the light of that consideration.
What is required to ground a claim that an agent acted in the light of
some consideration? Dancy seems to assume that part of what is required
is the normative judgment, by the agent, that the consideration favors
the action. At one of the points where Dancy makes the claim that the
purpose of psychological explanation is to reveal the light in which the
agent came to do what he did, he adds that this light cannot but be regarded
as an evaluative light.34 But there is no argument in support of this claim.
There is, however, an explanation for why it should have misleadingly
appeared to Dancy that psychological explanations of actions must take
for granted that the agent saw the things that motivated her in a positive
evaluative light.35 For it is indeed bizarre for an agent to be motivated by
some consideration, and yet not to regard that consideration in a positive
evaluative light. Since it is bizarre, we do not feel that we have received a
satisfying explanation when this happens. But the reason why it is bizarre
is that we humans are typically motivated by considerations that are in
33 Dancy (2000), p. 6.
34 Dancy (2000), p. 97. It is unclear whether Dancy is speaking for Nagel or for himself at
this point. But even if he is only representing Nagel with this claim, in many other places
he says essentially the same thing. See, e.g., the passages above, taken from Dancy (2000),
pp. 129, 136.
35 I write ā˜take for grantedā™ here so that these evaluative judgments can function as what
Dancy calls ā˜enabling conditions,ā™ rather than as a proper part of the explanation of action.
The current argument will speak equally strongly against this sort of role for normative
fact good reasons, and, equally typically, we (adults) have correct beliefs
about those considerations being good reasons. When we are motivated