. 30
( 38 .)


Par¬t™s argument here is good, insofar as we are considering reasons
in their requiring role. But Par¬t™s short ¬nal sentence here exposes his
assumption that all reasons are prima facie rational requirements.
It might seem that a parallel argument would also support the claim
that Humean internalism must be false of justifying reasons, and therefore
that Kantian internalism must be a correct account of reasons in both
roles. One might try to argue in the following way: suppose that one
has a reason that would justify acting in a certain way, but not require
it. Then this reason would also justify being motivated to act that way,
although it would not require it. This shows that we would have the
reason for being motivated in that way, even if we were not so motivated.
So far so good. But although this conclusion does argue against Humean
internalism for justifying reasons, it does not support Kantian internalism
for such reasons. For the conclusion does not establish that a fully rational
agent would be motivated by the justifying reasons. It only establishes that
those reasons would exist independently of the motivations of the agent.
If Kantian internalism were true of such a reason, the agent would be
rationally required to be motivated by it. But by stipulation, the reason is
not a requiring one. So it seems in fact that the strong form of externalism
presented above will be true of justifying reasons. But before we endorse
this conclusion we should re-examine the case that seemed to provide
such strong support for Humean internalism: the Air Florida case.

The Air Florida case revisited
In the Air Florida case the reason at issue “ that the man™s action saved the
lives of other people “ is relevant as a justi¬cation of action, not as some-
thing that requires action. This can be seen by noting that in discussion of
the case, the question is emphatically not whether the heroic man™s actions
were rationally required. Rather, the discussion centers on the question of
whether the action of the ¬ctional passenger is made rationally permissible “
i.e., is rationally justi¬ed “ by the reason. The action, involving such grave
16 Par¬t (1997), p. 130.

Brute Rationality

personal risk, is obviously one that stands in need of justi¬cation. That
is, without a powerful justi¬cation, it would have been irrational to have
done what the heroic man did, refusing the life-line time after time, until
he ¬nally succumbed to the cold. But it was not irrational; the altruistic
reason provided the necessary justi¬cation. In the discussion of this case,
the contingent motivations of the agent seemed to make a great deal of
difference to the question of whether or not the agent™s action was indeed
rationally justi¬ed (the question of moral justi¬cation is neither here nor
there). When he was motivated by the fact that he could save lives, then his
action was rationally permissible, though not required. But when he was
motivated to act in such a risky way only by the prospect of a second date,
then his action did not seem rationally permissible. This seemed to sup-
port the Humean internalist. But now we can see that, at best, it supports
Humean internalism as it applies to reasons in their justifying role.
In fact, I write ˜at best™ above because there are differing intuitions about
what one should say about the second, ¬ctional case: the case in which the
man saved the lives of the other passengers only because of the prospect of
a second date. On the one hand, it is very tempting to say that this man™s
actions were irrational, since it is simply not worth it to risk one™s life
to get a second date (especially with someone who refuses a second date
because one is not heroic enough). On the other hand, it is also tempting
to say that whenever one can save the lives of a number of other people,
it is rationally permissible to risk one™s life to do so. The difference in
intuitions here may seem to stem from a difference in the kind of thing
one is evaluating: action tokens, or action types.17 If one takes this route in
explaining the different intuitions, then one can say the following: the ¬rst
intuition (˜irrational™) is strong when one is evaluating the token action
of the pathological date-seeker, while the second intuition (˜rational™) is
strong when one is evaluating the action type of this ¬ctional man “ an

17 This is the line I took in the paper from which this chapter was derived, although I
now regard it as an inferior solution. Robert Audi also addresses this issue. See Audi
(1985). Joseph Heath makes a similar distinction, in terms of ex ante and ex post questions
about actions. See Heath (1997). Heath points out that Humeans like Williams con¬‚ate
these two forms of justi¬cation. But Heath, like Audi, does not distinguish requiring
from justifying reasons. This explains, in part, why Heath (p. 471) ends up accepting the
Kantian intuition that “acting morally is, in one sense, just acting rationally.” Here the
quali¬er “in one sense” does not indicate, benignly, the reasonable claim that moral action
is rationally permissible. Rather, it expresses the much stronger view that ideal rational
justi¬cation tells us to abide by moral requirements.

Internalism and different kinds of reasons

action type of which the action of the actual hero of Air Florida™s Flight
90 was also a token. The question then becomes: are justifying reasons
relevant to action types, or to action tokens, or perhaps to both?
It seems to me that it is misguided to attempt to assess the rational status
of action tokens by appeal to reasons, unless this is simply understood as
assessing the rational status of the relevant action type that the action token
instantiates. Here is one reason for such a doubt. If a consideration can
fail to be a justifying reason for an action token because that token was
not based on the consideration (as one is tempted to say in the ¬ctional
Air Florida case), then it seems we could have action tokens that were
required but not justi¬ed. For example, suppose that an agent is rationally
required to take a signi¬cantly painful two-day treatment that will cure
him of a life-threatening disease and restore him to perfect health and the
prospect of a long and happy life. We can stipulate that the painfulness of
the treatment, though signi¬cant, is not so great that it would be rational
to refuse the treatment and die, just in order to avoid the pain. Suppose
now that the agent does undergo the treatment, but only because he has
been promised a bowl of cherry ice cream; he had refused the treatment
and all other inducements until the prospect of cherry ice cream made its
appearance. If the potential justifying reason here “ that the agent will be
restored to health “ fails to be an actual justifying reason because it does
not ¬nd a corresponding motivation in the agent, then we seem forced
to say that taking the treatment was rationally required, but that there was
nevertheless no reason that would justify the patient in undergoing it. This
simply sounds too strange. We should favor an alternative description, if
a plausible one is available. And one is: the action was both required and
justi¬ed in terms of objective rationality, which is the sense of rationality
to which “ as we have already seen in chapter 4 “ reasons are directly rele-
vant. That is, there is a reason for the treatment that justi¬es undergoing it.
But despite its favorable objective rational status, and despite the presence
of the justifying reason, it remains the case that the agent performed the
action for the wrong reasons. Because of this, and because (in this case)
such a performance was the result of his being insuf¬ciently averse to the
painfulness of the treatment (or of his being much too concerned with
cherry ice cream) his action was subjectively irrational. That is, we can
say the following three things in this case: that undergoing the treatment
was objectively rationally required and justi¬ed; that there was a justifying
reason to do so; that the actual agent™s undergoing of the treatment was
subjectively irrational. One important consequence of this discussion is

Brute Rationality

therefore that it highlights one context in which a failure to distinguish
subjective from objective rationality is likely to result in a wrong con-
ception of practical reasons. In the case of ˜acting for the wrong reasons,™
the simple stipulation that the agent is fully informed is not guaranteed
to bring the subjective and objective rationality of an action into agree-
ment. If, in such a context, one tries to gain insight into practical reasons
by stipulating full information and then using the intuition that a cer-
tain action is subjectively irrational, one will be led into error. For it is to
objective rationality that practical reasons are directly relevant, and in this
context one will have come to a wrong conclusion about the objective
rational status of the action.18 No argument in this book has made use
of an intuition that a particular action was subjectively irrational in sup-
port of any claim about that action™s objective rational status, unless, in
addition to stipulating full information, it was also made clear that the
relevant action would have been subjectively irrational no matter what its
etiology. For example, if a fully informed agent performs an action, the only
relevant consequences of which will be a moment™s pleasure and a lifetime™s
misery, then the subjective irrationality of this action implies its objective
irrationality because there is no way such an action could be subjectively
A second reason for regarding the justifying role of reasons as relative to
relevant action types, and not to action tokens, is the following. There is
reason to wonder what exactly the justi¬cation of an action token could
possibly be, if it is not simply the justi¬cation of the relevant action type.
For it is very plausible that arguments seeking to show that an action is
required or justi¬ed must proceed in general terms if they are to make the
status of those actions intelligible.19 That is, if the available reasons show
a certain action token to be, say, rationally justi¬ed, then it seems fair to
say that any other action that is similar in the relevant respects must also
be rationally justi¬ed by the same reasons. If this is right, then what seems
to be the justi¬cation of an action token is really only the justi¬cation of
18 See pp. 8“9 for a discussion of Gibbard™s equation of subjective and objective rationality
under conditions of full information. See also pp. 69“72 for other cases in which a focus
on subjective rationality misleadingly suggests that, for example, normative judgments
provide reasons because acting against one™s normative judgment is a species of subjective
rationality. At pp. 163“64 many distinct forms of subjective rationality are described, and
each of these can also form the basis for mistaken conclusions regarding what sorts of
considerations provide normative reasons.
19 See Raz (1999b), p. 220.

Internalism and different kinds of reasons

the relevant action type. Thus, I conclude that the justi¬cation of actions
is always the justi¬cation of action types. Both the heroic rescue, and the
rescue motivated by a pathological desire for a second date, may be equally
rational. For despite differences in the motives that produced them, they
may still be tokens of the same relevant action type. This is especially
plausible when one recalls that the relevant type, for purposes of assessing
objective rationality, does not include the motives from which the action
springs. Moreover, any difference between the two rescues does not seem
likely to be of much consequence to their objective rational status, for
they are alike in their most important reason-giving features: both involve
the risk of drowning, and the chance to save several people from a similar
Because the two rescues are instances of virtually the same relevant
type, either both rescues are objectively rational or both are objectively
irrational. It is objectively rational to save several people at the risk of one™s
own life, though it would also be objectively rational to decline to do so,
in favor of one™s own safety. This means that the reason ˜that the action will
save the lives of several strangers™ is quite a strong justifying reason, for it
can justify actions that involve a signi¬cant risk of death for the agent. So
this altruistic reason is a justifying reason independent of the motivations of
the agent. Humean internalism is therefore false of such altruistic reasons.
Is Kantian internalism therefore true of these reasons? Not necessarily.
For if it is subjectively rationally permissible, but not required, to act on
this reason at the risk of one™s own life, then it seems possible that even
a fully rational agent might be unmotivated to act on the reason. This
would mean that externalism was true of justifying reasons: that even a
fully rational agent might have such a reason and be unmotivated by it.
And in fact, even if all rational agents must be motivated to some minimal
degree by such a reason, in which case Kantian internalism would still
be technically true of such justifying reasons, it would also have lost much
of its interest. For the strength of such a reason, as a justi¬cation, would
not correspond to this minimal rationally required degree of motivation.
Rather, this minimum would correspond to the strength of the reason
as a requirement. When one is determining the justifying strength of the
altruistic reason, it doesn™t matter whether all rational agents would be
motivated by it to some minimal degree. The technical truth or falsity of
Kantian internalism is irrelevant when one is considering reasons in their
justifying role.

Brute Rationality

My conclusion therefore is that justifying reasons do not depend in
any important way on the antecedent motivations of the agent, so that
Humean internalism is false of them. And Kantian internalism is also false
of such reasons, since it seems that even a fully rational agent might not be
moved by them. To the claim that a fully rational agent would be motivated
to some minimal degree by primarily justifying reasons, the response is
that this minimal motivation is completely unconnected to the justifying
role of those reasons. Instead, it gives a measure of the minimal requiring
strength that those reasons possess. Therefore, when considering reasons in
their justi¬catory role, externalism is the most illuminating view. Humean
internalism only seems plausible when one wrongly looks to subjective
rationality for direct insight into the existence of justifying reasons. That is,
there are cases in which a justifying reason fails to motivate an agent, and in
which that very reason provides the justi¬cation without which the action
(which is objectively rational) would have been objectively irrational. In
such cases the reason does not make the agent™s action subjectively rational.
And this can lead to the wrong conclusion that the reason is in fact not a

ev i de nc e
Is a failure to appreciate the justifying/requiring distinction really hin-
dering progress in the current internalism/externalism debate? It should
seem very plausible that it is. Requiring and justifying simply are two
conceptually distinct but equally important roles for normative reasons. If
an argument for some particular form of internalism were to rely at some
point on the assumption that reasons provide pro tanto rational require-
ments, this would render the argument invalid for reasons that do not
provide such requirements. And if an argument relied on some feature
of justi¬cation, this might render the argument irrelevant to the capacity
of reasons to require. Despite these a priori claims about the destructive
effect of a failure to recognize the justifying/requiring distinction, it may
be worthwhile to point to at least a few places in which this failure has a
noticeable effect.
In his seminal “Internal and External Reasons,” Bernard Williams
argues against externalism and against Kantian internalism by arguing
against the existence of reasons that are completely independent of the
contingent motivational setup of the agent. One now famous example is

Internalism and different kinds of reasons

that of Owen Wingrave, whose father (on Williams™s rephrasing) insists
that Owen has a reason to join the army. Williams explains that Owen
hates the thought of the army, and has no other existing motives from
which rational processes could generate a desire to join the army. Now,
it is admitted by internalists and externalists alike that Owen might well
not have any reason to join the army, so the example may perhaps not
have been the fairest that Williams could have chosen. But, in fairness to
Williams, he is here concerned not so much with the truth of Owen™s
father™s claim, as with what such a reason claim could possibly mean.
Williams thinks that the only thing the externalist could possibly mean
by a reason claim is that the agent would be irrational if he failed to act


. 30
( 38 .)