<<

. 18
( 38 .)



>>

have explicitly considered a justifying/requiring distinction, he ultimately
rejects such a view because he does not see the possibility that the very same
reason might have different justifying and requiring strengths.18 Rather, he
only considers the possibility that some reasons are exclusively justifying,
while others are exclusively requiring.
Raz™s strategy for explaining the basic belief is to claim that the reasons
that favor incompatible actions are often incommensurable. His inter-
pretation of incommensurability is clearest in the two-option case, and
what it means there is that neither of the opposed reasons defeats the
other, but that they are also not of equal strength. Because of this, Raz
holds that reason cannot adjudicate between them, or between the actions
they favor. Raz also holds that as long as one acts on an undefeated rea-
son, one is acting rationally. This claim, together with Raz™s interpreta-
tion of incommensurability, explains why action on either reason, in the
17 Raz (1999b), pp. 100“1.
18 Raz (1999b), pp. 101“2. Raz puts the distinction in terms of enticing and requiring reasons,
rather than in terms of enticing and requiring roles or strengths, and this contributes to his
rejection of it.

102
Accounting for our normative judgments

two-option case, would be rationally permissible.19 Such an explanation
allows Raz to make all the normative judgments expressed in the two-
gap and equal justi¬cation arguments, without falling into the inconsis-
tencies that undermined single-value views. For example, in the two-gap
case he could assert the incommensurability of the relevant altruistic and
self-regarding reasons. This would explain the rational permissibility of
choosing either A or D when presented with those two options, as well as
the rational permissibility of choosing either C or D when presented with
those two options. And if he continued to hold that the reasons involved
in actions of type A, B, and C were all commensurable, he could continue
to hold that it is rationally required to choose actions of type B over actions
of type A, and actions of type C over actions of type B. Similar reasoning
would apply to the examples used in the equal justi¬cation argument.
One might think that the preceding brief presentation of Raz™s view of
incommensurability must be unfair. Surely he must provide some expla-
nation for the paradoxical claim that there can be pairs of reasons, neither
of which is at least as weighty as the other.20 But Raz gives no positive
explanation. Rather, he shifts the burden of proof, so that the challenge
becomes ˜How could reasons involving apples be compared in weightiness
with reasons involving oranges (or aches)?™ Dif¬culties in explaining how
such comparisons could be made or defended, combined with a need to
explain the ˜basic belief,™ push Raz to endorse the thesis of widespread
incommensurability of reasons without much explanation of how it could
be true. So in arguing against Raz here, it will be useful to do two things.
The ¬rst is to mention a problem with the notion of incommensurability
that undermines its usefulness in accounting for our intuitions in cases
like those used in the two-gap and equal justi¬cation arguments. And the
second is to try to dispel some of the air of mystery surrounding the ability
to compare reasons that have very different substantive content.21
One of the most powerful objections to incommensurability has been
called ˜the nominal-notable objection.™22 Before presenting it, it will be
useful to give a quick argument that seems to support the idea that there can
be incommensurable reasons. Suppose, then, that I am trying to choose
19 Raz (1999b), pp. 102“3.
20 For this alternate expression of the mutual nondefeating nature of incommensurable
reasons, in terms of weight, see Raz (1999b), p. 182.
21 In fact, Raz himself provides one way in which mixed values, such as romance novels,
can be compared, even though they involve a number of distinct goods. See Raz (1999a),
pp. 182“201.
22 See Ruth Chang™s introduction to Chang (1997).

103
Brute Rationality

between two options, A and B, and that the only relevant differences
between them is that choosing A will spare me a month of annoying but
not debilitating knee pain but cost me four hundred dollars, while choosing
B will save me the four hundred dollars, but will result in the month of
knee pain. In this case, let us grant that it would be rationally permissible
to choose either A or B. This could be true because the reasons favoring
each option are exactly equal in strength. But this explanation loses appeal
once one grants the plausible claim that even if one increased the cost of A
to ¬ve hundred dollars, both choices would remain rationally permissible.
This suggests, in favor of Raz™s position, that the permissibility of the two
original options was not the result of their being favored by reasons of
precisely the same strength, but was the result of their being favored by
incommensurable reasons. However, this same strategy of changing the
cost can be used against an advocate of incommensurability. For suppose
that we now reduce the cost of A to some nominal value “ say, three
dollars “ or that we increase the duration of the knee pain to some quite
notable degree “ say, ¬ve years. These changes do not alter the nature of
the values underlying the reasons, and should therefore not alter either
the fact of incommensurability or the consequent rational permissibility
of either choice.23 But they do, since it is irrational to refuse to spend three
dollars to avoid a month of annoying but not debilitating knee pain. That
is the nominal-notable objection.
An advocate of the justifying/requiring distinction can agree with Raz
that the rational permissibility of either option in the ¬rst choice was
not the result of a precise (or even a rough) balance of reasons. But it
was not the result of incommensurability either. Rather, it was a result of
the fact that each reason, in the original case, possessed greater justifying
strength than the requiring strength of the other. This response is not
open to the nominal-notable objection. For when the cost of the cure is
reduced to three dollars, the justifying strength of the economic reason is
greatly reduced and can no longer justify suffering any signi¬cant amount
of annoyance or pain. Thus it becomes irrational to refuse to spend the
money.
It now remains to suggest how it is possible to compare reasons that
involve diverse values. A full solution to this problem is beyond the
23 If the reader believes that changes in these quantities of pain and money bring with
them other normatively relevant changes, other examples can be chosen that involve less
complex goods. Despite any quarrels with the details of the example, the point should
remain clear.

104
Accounting for our normative judgments

scope of this book. But it is worth noting that one consideration that
has led philosophers to doubt that such reasons can be compared is that
in con¬‚icts of such reasons there is no uniquely rational choice. The
requiring/justifying distinction accommodates this fact. This said, here
is a suggestion as to how to solve the comparison problem. It was argued
in chapter 4 that the basic unit of normative assessment should not be ˜a
reason.™ Rather, the notion of wholesale rational status should be taken
as more basic than that of a reason. In arguing for this conclusion, appeal
was made to an analogy with language: just as it is more pro¬table to
take the sentence as the basic unit of meaning, it is more pro¬table to take
wholesale rational status as the basic unit of normative assessment relevant
to actions. And just as we can identify the various meanings of individual
words with the systematic contributions that they make to the meanings
of sentences, we can understand the various normative roles of reasons,
and their strengths in those roles, by understanding how they contribute
systematically to the wholesale normative statuses of actions. In this way,
the ability to compare apples and aches is not a prerequisite for the ability
to judge it rationally permissible to give up an apple to spare oneself an
ache. Rather, the ability to make wholesale normative judgments about
the rationality of action “ an unsurprising ability to ¬nd in normal human
beings “ is a prerequisite to learning about the relative normative capacities
of reasons involving apples and aches.
Why have philosophers been led to embrace a notion as paradoxical as
the sort of incommensurability described above? The justifying/requiring
distinction provides an explanation for this fact. Philosophers have typi-
cally assumed that if reasons have comparable strength values, then those
values will be characterizable with single values, whether cardinal or ordi-
nal, rough or precise. But if the justifying/requiring distinction is valid,
and if reasons with equal justifying strengths often have very different
requiring strengths, then attempts to characterize the relative strengths of
reasons with a single value will often fail: two values will be required. The
thesis of incommensurability may be a response to the perception of this
systematic failure for particular pairs of reasons. In essence, the thesis of
incommensurability is an inference from the true premise ˜the strengths
of these reasons cannot be compared as if they were single values™ to the
false conclusion ˜the strengths of these reasons cannot be compared at all.™
Put in this way, the inference is obviously formally invalid. But it is easy
to understand how one might make this inference if one did not see that
reasons might have more than one dimension of strength.

105
Brute Rationality

Exclusionary permissions
A second proposal by Raz for accommodating the relevant phenomena
agrees with single-value views in holding that ¬rst-order reasons possess a
single strength value. Moreover, this proposal also agrees that the strength
of a ¬rst-order reason is a measure of its power to override other ¬rst-
order reasons. But Raz goes on to claim that it is sometimes rationally
permissible, but not required, to exclude some ¬rst-order reasons from
one™s deliberations. By making this additional claim, Raz can account for
our intuitions in the two-gap and equal justi¬cation arguments. The name
that Raz gives to the normative entity that makes it rationally permissible
to exclude some ¬rst-order reasons is ˜an exclusionary permission.™
Raz uses exclusionary permissions to explain the phenomenon of
supererogation.24 Here is how the explanation works. Suppose that an
agent has, as one of the options open to him, a morally supererogatory
action. Raz assumes that since such action is praiseworthy there must be
reasons in its favor outweighing all con¬‚icting reasons. Indeed he thinks
of such actions as in a sense required by reason.25 These are themselves
problematic assumptions, but let them stand. That is, let us assume that
¬rst-order reasons unambiguously favor the supererogatory action. Now,
since the action is supererogatory, we are also assuming that it is neither
immoral nor irrational to refrain from it. How can all of these claims be
true? How can all the relevant reasons unambiguously favor the action,
and yet not make it rationally required? Raz™s answer is that in cases
of supererogation there is an exclusionary permission that allows us to
24 See Raz (1999a), pp. 89“95 and Raz (1975). It is worth noting that cases of moral
supererogation only form a subset (albeit an important one) of the set of actions whose
rationally optional status is explained by the justifying/requiring distinction. This is partly
because altruistic reasons, which, on my account, have much more justifying strength
than requiring strength, can just as easily be reasons for immoral action as for morally
good action. An altruistic reason can rationally justify a risky immoral action, undertaken
for the sake of one™s family, friends, or colleagues. Accounts that attribute rationally
optional status only to morally supererogatory action are inadequate accounts of practical
rationality.
25 Raz (1999a), pp. 91, 94, and Raz (1975), p. 165. From the claim that it is always wrong
to say one ought not perform a supererogatory act, Raz (1975), pp. 165“6 infers that
such acts are always favored by conclusive reasons. This suggests that he only allows for
two sorts of actions: those one ought to perform, and those one ought not perform. An
alternate view claims that there are actions of which it is false that one ought to perform
them, and also false that one ought not. I suspect that most actions fall into this class, and
that Raz™s notion of incommensurability also commits him, against this 1975 argument,
to such a view.



106
Accounting for our normative judgments

exclude some of the ¬rst-order reasons that would otherwise make the
action rationally required. Presumably these excludable reasons include
all or some of the altruistic ones. In effect, an exclusionary permission
generates rational options by allowing two different calculations based on
the relevant ¬rst-order reasons: one that uses all the ¬rst-order reasons,
and one that uses only a subset of them. Since an exclusionary permis-
sion only says that one may exclude a certain subset of ¬rst-order reasons,
but does not require one to do so, either of the calculations is permit-
ted. Since the calculations favor different actions, both actions are also
permitted.
Here is how an exclusionary permission would help Raz avoid the force
of the two-gap argument. Suppose that, in cases in which only ¬rst-order
reasons are relevant, the ¬rst-order reasons favoring an altruistic action of
type D would be suf¬ciently strong to rationally require action in cases
in which the only other option was an action of type A or an action of
type C. But suppose further that in the actual case there is an exclusion-
ary permission, allowing us to exclude the altruistic reasons favoring D.
Given these assumptions, all of our intuitions about the examples in the
two-gap argument can be preserved. For although one calculation would
favor an action of type D over an action of type A, there would be an
alternate calculation available to the agent who had to choose between
the two types of action: the calculation based on RA alone, excluding RD
completely. Clearly this calculation would favor choosing A. Since either
calculation is permitted, so is either action. Parallel reasoning explains why,
in a choice between actions of type C and actions of type D, either action
would again be rationally permissible. And yet, since there is no exclu-
sionary permission that allows the agent to exclude RA , RB , or RC , it can
remain true that the agent is rationally required to choose actions of type
B over actions of type A, and actions of type C over actions of type B. This
accounts for all of the intuitions that were used in the two-gap argument.
Similar reasoning could be used to show that a theory that includes exclu-
sionary permissions might also avoid the force of the equal justi¬cation
argument.
But despite the adequacy of exclusionary permissions in accounting for
our intuitions in many cases, they are inferior to the justifying/requiring
distinction as a means of saving the phenomena. Their greatest liability is
their peculiar ontological status. Where do exclusionary permissions come
from? Raz is clear that they cannot be taken for granted, and require some


107
Brute Rationality

sort of justi¬cation.26 But it seems very implausible that there will always
be a justi¬cation for an exclusionary permission whenever we need such
a permission to account for our intuitions. This objection, however, is
unlikely to impress anyone who is antecedently committed to exclusion-
ary permissions. Such theorists will simply say that either our intuitions
are incorrect in those cases, or there is indeed some justi¬cation for the
exclusionary permission that we have not yet discovered.
A more formal dif¬culty with exclusionary permissions can be seen by
imagining a case in which such a permission allows the exclusion of just
one reason from one™s deliberation. This allows two possible calculations:
one in which the reason is included with all of its strength, and one in
which it is excluded completely. The problem here is that we sometimes
think that a reason must play a certain minimal role in the decision of
an agent, even if we agree that it could permissibly play a much greater
role. Exclusionary permissions cannot accommodate this. They can only
accommodate such a reason™s playing no role, or its playing a full role. For
example, suppose that an agent is rushing to catch a bus into the city.
She is quite late, and if she misses the bus it will cause her quite a bit of
annoyance, although it will not affect anyone else. As she is jogging along
the sidewalk, a man with a map and a puzzled face tries to stop her to ask
directions. Here the agent has the choice of stopping to help, which will
increase her chances of missing the bus, or not stopping, which will leave
the man unaided. Let us grant that in this case either action would be per-
mitted, and that this is the result of an exclusionary permission that allows
the agent to exclude the reason provided by the puzzled man™s interests.
Considered in isolation, this choice situation seems to be explained ade-
quately by appeal to this exclusionary permission. But suppose we wish
to make the additional conditional claim that, were the inconvenience of

<<

. 18
( 38 .)



>>