. 3
( 38 .)


Brute Rationality

on the false premise that a fully informed agent, performing a rational
action, might nevertheless be performing an inadvisable one. Also, the
˜subjective/objective™ terminology allows me to use phrases such as ˜an
account of rationality™ in order to indicate an account both of subjective
and objective rationality, and of the relation between the two. In what
follows, when I use the words ˜rational™ or ˜irrational™ without any quali¬-
cation, they should be understood in the subjective sense, which ¬ts more
with the everyday understanding of these words.
Although I will not provide a full account of the relation between sub-
jective and objective rationality until chapter 7, some limited claims about
their relation are independently plausible, and will be very useful in a
number of earlier arguments. First, if an agent knows all the facts relevant
to his action, then if that action is objectively irrational “ that is, if it is
prohibited by the fundamental normative principle “ it is also subjectively
irrational. This connection will allow us to move from the objective irra-
tionality of an action to its subjective irrationality (and therefore from its
subjective rationality to its objective rationality) in all cases in which it is
permissible to stipulate that the agent has all relevant beliefs. This claim is
very similar to one of Gibbard™s: “in the special case in which I know all
that bears on my choice, what is rational for me to do is what is advisable
for me to do.”10 My claim is slightly weaker, however, for it does not
entail that we can always move from objective rationality “ what Gibbard
calls ˜advisability™ “ to subjective rationality (or, therefore, from subjec-
tive irrationality to objective irrationality) even in the case in which the
agent is fully informed. As chapters 7 and 8 will explain, this move can
be illegitimate when the agent does not care about the considerations that
make his action objectively rational, or only performs that action because
of failures of instrumental rationality: cases in which an agent does ˜the
right thing for the wrong reasons.™ These are cases in which the etiology
of the action is what makes it subjectively irrational, and this gives rise
to the possibility that if the same action had been done for other reasons
it would have been subjectively rational “ and therefore it also gives rise
to the possibility that an action can be objectively rational despite being
subjectively irrational, even in a fully informed agent. Acknowledging this
possibility, we can make the following claim: if the agent is fully informed,
then if his action would have been subjectively irrational no matter what its
etiology, it is also objectively irrational. Since, as has already been noted, if

10 Gibbard (1990), p. 19.

Conditions on an adequate theory

the action of a fully informed agent is objectively irrational then it is also
subjectively irrational, Gibbard™s claim is very close to correct.11
Another interesting feature of subjective rationality is the following. It
seems that if one is simply unmoved by awareness of the prospect of some
signi¬cant harm for oneself “ say, that one™s action will cause one a great
deal of pain, or will risk some nontrivial injury “ this does nothing to
the normative force of the reason that one is aware of. This is not to
deny that one can be perfectly rational in willingly suffering such harms,
if there are suf¬cient countervailing reasons. But the fact that one needs
signi¬cant countervailing reasons shows that a rational person cannot be
very indifferent to such harms for himself. On the other hand, relative
indifference to the harms that one™s actions may cause other people is not
nearly as universally regarded as irrational in the ˜mental functioning™ sense
that is relevant to questions of moral responsibility and so on. Rather, when
we speak of such indifference, we use words such as ˜callous,™ ˜sel¬sh,™ or
˜mean.™ No one denies that it is rationally permissible to be motivated by
other-regarding reasons. But it does not seem to be rationally required
to the same extent that it is rationally required that one avoid harms for
oneself. Of course there are views of rationality according to which one
is in fact required to be as strongly motivated by altruistic as by self-
interested reasons. This introductory chapter is not the place to combat
such views. Rather, it is the place to mention that such accounts need to
make us comfortable with some apparently counterintuitive judgments as
to whether certain actions are subjectively rational or not “ rational in the
sense that is relevant to questions of moral responsibility and so on. That
11 It may be worth mentioning at the outset that subjective rationality, as understood here,
is not to be confused with Thomas Scanlon™s technical and restricted sense of rationality,
according to which actions are rational or irrational depending solely on whether or not
they are in line with the agent™s normative judgment that he or she ought to perform the
action. Scanlon™s sense is inadequate if we want a general notion that captures the wide
range of failures of practical mental functioning that are relevant to questions of mental
illness, competence to give consent, moral responsibility, and so on. One reason for its
inadequacy as such a general notion is that, as I argue in chapter 9, our actions very rarely
involve the normative judgments that are presupposed when one calls an action rational
or irrational in Scanlon™s sense. This is not to criticize Scanlon™s choice of terminology.
One is free to use whatever terminology one wants, as long as one is clear about what
one means. It is only a reminder that Scanlon himself recognizes many other failures in
practical mental functioning: insensitivity to certain reasons, compulsions, and phobias,
even when they are accompanied by rationalizing normative judgments (or no normative
judgments), etc. It is to this more general class that I am applying the term ˜subjectively
irrational.™ Of course, acting against one™s considered judgment as to how one ought to
act is one species of the sort of irrationality with which this book is concerned, and it will
be captured by the account offered in chapter 7.

Brute Rationality

is, in order to succeed in convincing us that it is irrational, in this sense,
to be indifferent to the harms one™s actions will cause other people, they
will have to account for the fact that we generally wish to hold extremely
immoral people fully responsible for their sadistic actions.
But even if it is a mistake to defend the normative equivalence of self-
interested and altruistic reasons, one cannot simply deny that there are
such things as altruistic reasons. That is, even if it is only callous, and not
irrational, to be indifferent to the harms one causes others, one should not
therefore deny that it can be perfectly rational to make great sacri¬ces “
even the ultimate sacri¬ce “ for others. It would be a poor theory of ratio-
nality that insisted that it was irrational to sacri¬ce one™s life to save a group
of strangers, or that held that one™s ˜real™ reason in such a case was essentially
self-interested.12 If an agent is strongly motivated to save a group of other
people, and acts accordingly at the cost of his own life, this may be perfectly
sel¬‚ess, and perfectly rational. These two facts “ that it is rationally permis-
sible to be sel¬sh, but also rationally permissible to make sel¬‚ess sacri¬ces
for others “ seem to suggest that whether or not one has an altruistic
reason depends upon whether or not one has a corresponding altruistic
desire. And indeed there have been philosophers who explicitly claim that
while one™s objective interests or needs provide desire-independent reasons,
there is another class of reasons, which includes altruistic reasons, that
stem from one™s desires or values.13 The plausibility of such views derives
entirely from their ability to capture some otherwise elusive phenomena.
Moreover, such accounts will need some way of limiting the content of
one™s reason-giving desires or values, so that they do not end up claiming
that one has a reason to drink paint simply in virtue of a desire to do so,
or that one has a reason to exterminate some offending race of human
beings because of one™s racist values. It is one goal of the present book to
provide an explanation for the differential impact of desire on the relevance
of reasons to the subjective rationality of action. Moreover, this explana-
tion will limit the importance of desires in such a way that one™s desire to
drink paint or to hurt someone else never rationally justi¬es one™s action,
while one™s desire to help someone else can provide such justi¬cation.
Because the following arguments will be so much at odds with the
Kantian view that moral requirements are also rational requirements, it may
seem as though they must be concerned with a more stringent notion of
12 It would be as poor as a theory that insisted that it would be irrational not to make such
a sacri¬ce, because one would, by failing to act, cost more lives than one saved.
13 See Copp (1995), pp. 172“85 and Foot (1978b), pp. 148“56.

Conditions on an adequate theory

subjective rationality: perhaps something closer to the colloquial notion
of insanity. But this would be a misperception. On the view that will
be put forward here, smoking generally counts as mildly irrational, as
does postponing a trip to the dentist. The difference with a Kantian view
is not a conceptual one, or a question of the severity of the charge of
irrationality. Rather, it is a substantive disagreement about what really
does count as a defect (large or small) in practical reasoning. Of course
the notion of irrationality in play here is related to the notion of insanity.
But it is unlikely that there is any plausible notion of irrationality, either
practical or theoretical, such that an agent might do countless extremely
irrational actions, or hold countless extremely irrational beliefs, and still
avoid the charge of insanity.

rat i onal i ty and moral th e ori e s
Very roughly speaking, contractualist moral theories hold that morality is
the system of rules that people would agree to, under certain conditions.
However, if ˜people™ is understood here as ˜actual people™ then it is unlikely
that contractualism will yield very extensive or determinate results. After
all, some people hate to agree with other people, other people are too
stupid to agree to anything, and still others are simply self-destructive
lunatics. So contractualist theories will need to restrict the scope of ˜people™
in some way. Intuitively, the kind of people we would like to exclude are
irrational people. That is, contractualism is plausible as a moral theory if
it claims that morality is the system of rules that rational people would
agree to, under certain conditions. It should be clear, therefore, that the
notion of rationality is likely to play a crucial role in such a moral theory.
It should also be clear that being rational cannot simply be equated with
being moral, for then contractualism would be trivial.
Now, one interesting question for contractualist moral theories is
whether a person, if rational, would or could actually act on the set of
rules that they would have agreed to under the relevant counterfactual con-
ditions. We can grant the contractualist the plausibility of the claim that
the way a rational person will act is quite similar to the way that such a
person would advocate that others act, and that it is also quite similar to the
way that such a person would agree to act, given that others also agreed
to act in that way. That is, we can agree that there is liable to be very
signi¬cant overlap in these differently speci¬ed classes of action. But sup-
pose that there is any mismatch between the way a rational person would

Brute Rationality

actually act, and the rules that a rational person would agree to act on, on
the hypothesis that others similarly situated would also agree. If there is any
mismatch then there is the danger that it will sometimes be irrational to
be moral, according to a contractualist moral theory. That would be a very
bad consequence for such a theory, if rationality is taken in the objective
sense. For it would mean that there could not, even in principle, be an
adequate reason to do what was, in those situations, morally required. And
it would not be much better if rationality were taken in the subjective,
˜mental functioning™ sense. For one thing, this would mean that someone
would have to be a little ˜wrong in the head™ to act morally in those mis-
matched situations. And in any case, the best explanation for the subjective
irrationality of such action is likely to be its objective irrationality.
Is this possibility of mismatch a real one, and if real, does it mean
that contractualist theories are doomed? On many popular conceptions
of rationality the answer seems to be ˜yes.™ If we understand rationality in
maximizing terms then it is extremely plausible that contractualist moral
theories will always yield at least a small class of actions that are morally
required but rationally prohibited.14 Maximizing views of rationality say
that, in a given choice situation, there is one class of rationally permitted
actions: those that maximize some measure. On some views this is a mea-
sure of preference satisfaction, on others a measure of pleasure, and on still
others it is a weighted sum of a number of distinct goods. It is extremely
unlikely that the class of actions that actually maximize the relevant mea-
sure in actual circumstances will always include the class of actions that are
required by a set of rules that have the following feature: in the relevant
counterfactual circumstances it would maximize the measure to advocate
or to agree to those rules.
A similar sort of trouble will af¬‚ict consequentialist moral theories. Such
theories, very roughly, claim that morality is a matter of acting in such a way
as to bring about the best consequences.15 This may be a matter of trying
14 I focus on subjective rather than objective rationality in the following discussion because
both morality and subjective rationality are plausibly regarded as somehow relative to
the agent™s epistemic situation. Thus no difference in this epistemic relativization is avail-
able to remove the sting from the possibility of mismatch between morality and subjective
rationality, as there is in the case of a mismatch between morality and objective rational-
ity. Moreover, maximizing views of subjective rationality typically go hand-in-hand with
maximizing views of objective rationality.
15 These may be the best actual consequences, or the best foreseeable consequences. If the
former, then the theory will have to separate moral wrongness from blameworthiness.
But the argument offered here will still apply, with ˜objective rationality™ substituted for
˜subjective rationality.™

Conditions on an adequate theory

to bring about the best consequences each time one acts, or of acting on
motives that tend to produce the best consequences, or of acting according
to a system of rules that is best with regard to consequences. Since ˜best
consequences™ means ˜best for everyone™ in this context, consequentialist
theories inevitably run into a problem if they also advocate a maximizing
theory of rationality. For unless the theory of rationality simply says that
rational action is action that brings about the best consequences in exactly
the same way as the moral theory says that moral action does, then there is
always the possibility that the action that maximizes the measure relevant
to morality will not be the action that maximizes the measure relevant to
rationality: some morally required action will be irrational. In fact, Mill
runs into a very similar problem in Utilitarianism. According to Mill, one
ought always to act so as to bring about the greatest amount of overall
happiness for those whom one™s action will affect.16 But Mill also holds
that the way one actually will act is determined by how much happiness
one believes one will get for oneself. Certainly these two sorts of actions will
not always coincide. Because Mill sees this, he is explicit in his advocacy of
an education that will bring people™s ideas of personal happiness more in
line with their ideas of overall happiness. But until someone is successfully
educated in this way, it may be impossible “ if Mill is right “ for that
person to act as they ought. This problem is not exactly that it is irrational
to act as one ought. This is because Mill is a psychological egoist, and not
a rational egoist. That is, he says we are actually psychologically set up to
maximize our own perceived happiness, not that it is rationally required
that we do so. But the problem is identical in form. In a nutshell it is the
following: if one has a maximizing view of rationality, and a maximizing
consequentialist view of morality, then unless the two views are really the
same view, one will sometimes have to choose whether to be moral or
One avenue of escape from the above problems is to embrace the idea
that rationality and morality really do amount to the same thing. This
claim is characteristic of Kantian moral theories. These theories of morality
begin by offering some morally neutral characterization of what it is to

16 I use the phrase ˜ought to act™ advisedly here, instead of ˜is morally required to act,™
for Mill™s maximizing view applies directly to ˜ought,™ and only indirectly yields moral
requirements as actions that ought to be punished. See Mill (1979), ch. 5. This actually
results in a much more plausible moral theory than Mill is generally credited with, and
one that is not maximizing at all. For this reason the important con¬‚ict discussed above
must be cast in terms of Mill™s maximizing nonmoral ˜ought.™

Brute Rationality

be a rational being. For example, rational beings might be beings who
act on universal laws, or who have the capacity to evaluate their desires


. 3
( 38 .)