Meeting of the Aristotelian Society held at Senate House, University
of London, on 9 March 2009 at 4:15pm.
X—A DEFENCE OF CATEGORICAL REASONS
RUSS SHAFER-LANDAU
In this paper I offer two arguments designed to defend the existence of categorical reasons, which I define as those justifying considerations that obtain independently of their relation to an agent’s commitments. The first
argument is based on certain paradigm cases meant to reveal difficulties
for practical instrumentalism—the view, as I define it here, that categorical reasons do not exist, because all reasons must serve the commitments
of the agents to whom they apply. The second argument relies on considerations of responsibility and blame to establish the existence of categorical reasons.
I
Categorical reasons, as I will define them here, are reasons that obtain independently of their relation to an agent’s commitments. Such
reasons do not depend for their existence on their being instrumental to the achievement of any of an agent’s desires, goals or cares. I
believe that there are categorical reasons for action, and will offer
two arguments on their behalf.
If there are categorical reasons for action, then practical instrumentalism is false. Practical instrumentalism (henceforth, just instrumentalism) is the view that the only reasons there can be are socalled hypothetical reasons, i.e. reasons to do things that are in
some way ancillary to the achievement of one’s commitments (cares,
desires, wants, goals, etc.).
Apart from the intrinsic interest of the matter, showing that there
are categorical practical reasons, and that instrumentalism is false,
is important for at least two reasons. First, it would enable us to resist relativistic arguments that assume that moral requirements entail excellent reasons for action, but make reasons contingent on our
commitments, thereby making the content of moral requirements
contingent on our commitments. Second, it would provide us with
©2009 The Aristotelian Society
Proceedings of the Aristotelian Society, Vol. cix, Part 2
doi: 10.1111/j.1467-9264.2009.00264.x
190
RUSS SHAFER-LANDAU
an adequate reply to arguments that assume a commitment-independent source of moral requirements, and then proceed, with the
help of instrumentalism, to the conclusion that there may be no
good reason to abide by morality’s demands.
II
Both of my arguments for categorical reasons, and against instrumentalism, begin by directing our attention to a familiar sort of example: that of the dedicated, successful immoralist. Imagine a
person who is very sharp, very cunning, but also deeply malicious.
His happiness is directly proportioned to the misery he wreaks. His
top priority in life is to cause pain and suffering, even if, as he
knows, such conduct will likely bring an early death, or a long incarceration. We intuitively regard such a person as (at the least)
morally obligated to desist from the cruel treatment he longs to impose. Don’t we also believe that there are excellent reasons for him
to so refrain—namely, all of those considerations that constitute the
wrongness of his actions? The reasons to refrain from cruelty are (at
a minimum) the very considerations that make his actions wrong in
the first place.
Consider an experienced torturer working on behalf of an authoritarian government. Such a person not only endorses the legitimacy of the regime, but takes active pleasure in breaking his
victims. His greatest joy is stripping the last vestiges of dignity from
those who initially resist his demands. At a given session, as he is
about to apply the electrodes, he pauses to consider the merits of his
action. He sees that doing so will get him what he most wants, and
will frustrate none of his desires. He proceeds accordingly.
Consider a different case, one in which a person can very easily
rescue another. A child has strayed from her parents on a busy city
street, and is about to toddle into the path of an oncoming car. The
bystander sees what is happening. He need only reach an arm down
to the child to save her from an awful death. Rather than doing so,
he watches in delight as the child is run over and killed.
If there were nothing to be said against these actions and
omissions—no considerations that opposed, extinguished, or overturned the case these agents might make for their cruel conduct—
then it is hard to see how their actions could be wrong. But they ob©2009 The Aristotelian Society
Proceedings of the Aristotelian Society, Vol. cix, Part 2
doi: 10.1111/j.1467-9264.2009.00264.x
A DEFENCE OF CATEGORICAL REASONS
191
viously are wrong. And the sorts of considerations just mentioned—
those directly relevant to matters of justification (and, in this particular context, those that indict the agent’s cruelty)—are precisely
what reasons are. Reasons are, by definition, considerations that favour or oppose, that make something appropriate, legitimate, or
justified (or the reverse). So, if we think that there is a plausible story to tell about why the dedicated evildoer is wrong for indulging
his inclinations, then we are committed to there being reasons for
him to refrain. And this despite the fact that, by hypothesis, he has
no commitments that would be furthered by his doing so.
But surely, some will say, the moral monster has some commitments that will be furthered were he to refrain from cruelty. And
that shows that he will, after all, have some reason to refrain. But
we’ve no grounds for thinking this a categorical reason. Whatever
reasons he has to desist will stem straightforwardly from his aversion to jail, or his desire to avoid the potential harms inflicted by his
vengeful victims, etc.
There are two things we can say here. First, and fairly obviously,
even if all real people in the real world do have at least some ends
that would be served by avoiding cruelty, we can imagine a possible
world in which our misanthrope does not. The instrumentalist’s rejection of categorical reasons is meant to express a necessary
truth—reasons must further what an agent cares about—and so is
vulnerable if there are possible contexts in which this truth fails to
obtain. In the scenarios I am imagining, the ruthless immoralist has
no commitments that would be served were he to refrain from his
cruelty. But there are, nevertheless, excellent reasons for him to so
refrain—namely, and at the very least, all of the considerations that
make his proposed actions immoral.
More importantly, we don’t want to make the case against cruelty
dependent on an instrumental link with this man’s goals. Suppose,
for instance, that our torturer wanted to avoid the censure that he
would earn were his actions publicized. The best way to minimize
his risk is to stop doing what he does. Though this, let us grant,
does provide him a reason to stop, it isn’t the only, or nearly the
strongest, reason to do so. The cruelties he perpetrates are opposed
by a host of considerations that make no mention of his aims. These
considerations are reasons—reasons to refrain from deliberately inflicting misery. And these reasons will, first and foremost, mention
the suffering of his victims, and the absence of their consent to his
©2009 The Aristotelian Society
Proceedings of the Aristotelian Society, Vol. cix, Part 2
doi: 10.1111/j.1467-9264.2009.00264.x
192
RUSS SHAFER-LANDAU
treatment. If the immoralist’s aversion to being found out enters
into it at all, it is only in a subordinate role, as a consideration that
may supply an additional reason to refrain from his actions, one
that is likelier than the others to motivate him to do the right thing.
Here is the argument in a nutshell:
(1) If there are reasons for these dedicated immoralists to refrain from their evil deeds, then practical instrumentalism
is false.
(2) There are such reasons.
(3) Therefore practical instrumentalism is false.
The first premiss is meant to be acceptable both to fans and to critics of categorical reasons. The dedicated immoralists that I am imagining are precisely those who lack any commitments that would
be furthered were they to hew to the moral path. So there should be
no controversy on this score. Premiss (2) is another matter. I have
tried to reveal its attractions with the examples of the dedicated evildoers. So long as we think—as all of us do—that there are genuine
considerations to oppose their cruelty, and also think that such considerations obtain independently of their commitments, then premiss (2) is secure. That would be enough to establish the existence of
categorical reasons.
Instrumentalists will likely charge that the argument begs the
question, because (as they will see it) the examples and considerations that support premiss (2) are insufficiently independent of the
conclusion being argued for. An ideal argument is one whose premisses can find support from those who are as yet uncommitted on
the matter at hand. If the only reason to endorse a premiss is that
one already accepts the conclusion it is meant to support, then the
argument that incorporates that premiss is question-begging. The
instrumentalist will likely insist that the only ones willing to ratify
premiss (2) are those who are already committed to rejecting instrumentalism.
I don’t think that instrumentalists are right about that. But before
saying why, we might undertake a brief excursus on the matter of
begging questions. Begging a question is sometimes unavoidable. In
ethics, the likeliest scenarios are ones in which one is advancing fundamental normative or evaluative commitments. It is hard to avoid
begging a question if one encounters someone who denies that pain
is ever bad, or denies that there is anything immoral about humiliat©2009 The Aristotelian Society
Proceedings of the Aristotelian Society, Vol. cix, Part 2
doi: 10.1111/j.1467-9264.2009.00264.x
A DEFENCE OF CATEGORICAL REASONS
193
ing vulnerable innocents. Perhaps the only way to avoid a petitio in
these circumstances is to show that one’s interlocutor is contradicting himself. This is the fond hope of Kantians and others—to show
that those with patently immoral sensibilities are in some way undercutting their own commitments and displaying some internal incoherence.
Perhaps the Kantians are right. It would be lovely were it so. But
let us pursue other possibilities, ones that do not vindicate the existence of categorical reasons by attributing a contradiction to those
who refuse to acknowledge them. On this alternative line, those
who oppose our basic normative and evaluative commitments can
coherently reject the claims we hold so dear. Any defence of our
deepest commitments will have to come from the sorts of bolstering
considerations that are involved in revealing a belief to be situated
within a network of mutually supporting beliefs. But such a defence
will not be able to avoid the charge of begging the question. The
other beliefs we enlist on behalf of our original claim may be no
more persuasive to opponents than the position originally in need of
support.
Unless we can reveal a contradiction in our opponent’s position,
we may have to beg a question somewhere. The most likely point is,
as I have said, with regard to our fundamental normative and evaluative commitments. And we are certainly in the neighbourhood,
when considering an endorsement (or rejection) of categorical reasons.
This is not yet to concede that this first argument is questionbegging. But what if it were? There is independent reason for thinking that question-begging claims and arguments are ones that agents
may sometimes be justified in believing. I am justified in believing
myself to be conscious, even if others regard me as an unthinking
automaton whose protestations are merely programmed behaviours. If I am imprisoned on false charges, I am justified in believing
myself innocent, even if all publicly available evidence convinces
everyone else of the justice of the sentence. If, having cried wolf one
too many times, my next cry is unheeded and disbelieved, I am
nonetheless justified in believing that there is such an animal before
me if I see it approaching and ready to attack.
In each of these cases, we can easily imagine that any evidence
that I offer to substantiate my claim will be taken as confirming the
hypotheses of the doubters who surround me. The credibility of my
©2009 The Aristotelian Society
Proceedings of the Aristotelian Society, Vol. cix, Part 2
doi: 10.1111/j.1467-9264.2009.00264.x
194
RUSS SHAFER-LANDAU
testimony will invariably be rejected, as it is expressive of a conclusion that the sceptics will not accept. In this context, anything I say
on my behalf is bound to be question-begging. But I am nevertheless
justified in regarding my supporting beliefs, and the claims they seek
to vindicate, as eminently plausible.
I don’t say that our belief in the existence of considerations that
oppose the actions of the immoralist is as epistemically secure as the
contested beliefs in the examples just given. That is not the point of
introducing them. Rather, the examples are designed to show that
some question-begging claims are credible and justifiedly held. So
even if the various beliefs that condemn the actions of the immoralist beg the question against the instrumentalist, such beliefs might be
epistemically justified.
To pursue this path, we would need to distinguish between those
question-begging beliefs that are, and those that are not, justifiedly
held. I don’t intend to embark on such a discussion, because I do not
believe that the considerations that support premiss (2) are, in fact,
question-begging.
I don’t believe that only those already convinced of instrumentalism’s failure will find these considerations compelling. What is true
is that dedicated instrumentalists will find something to resist. I submit that those who have yet to develop a considered view about the
existence of categorical reasons will find the considerations that
support the second premiss natural and highly plausible. They won’t
need convincing that there is something to be said against the torturer’s actions, and something to be said in favour of easily preventing a child from being needlessly killed. They will then discover that
such considerations, when conjoined with an uncontroversial definition of reasons, and the absence of any relevant commitments on
the immoralist’s part, entail the existence of categorical reasons. The
only ones who will deny the appeal of such considerations are those
whose theoretical commitments already require them to do so. That
is not enough to show that the argument is unsound.
III
A second argument on behalf of categorical reasons also relies on
our views about the dedicated immoralist, but shifts the focus to
matters of responsibility and blameworthiness. Consider those who
©2009 The Aristotelian Society
Proceedings of the Aristotelian Society, Vol. cix, Part 2
doi: 10.1111/j.1467-9264.2009.00264.x
A DEFENCE OF CATEGORICAL REASONS
195
freely commit themselves to blowing up civilians in crowded areas.
Such people are (with perhaps rare exceptions) highly capable of assessing options, gathering information to discover how best to pursue their chosen goals, and taking the needed steps to ensure that
their goals are met. They are not insane. They are genuine agents,
responsible for their deeds. They are as blameable as agents can be.
We would rescind our condemnation if such people were literally
compelled to do what they did. We would mitigate the blame were
we to discover that they had been coerced or manipulated into doing what they had done. But on the assumption that the killers have
autonomously elected to proceed in their undertaking, then they
are, at the very least, rightly subject to blame.
One is blameworthy for an action only if there is some reason to
refrain from committing it. Because the killers are blameworthy for
their deeds, there is a reason that opposes their actions. Since this
reason does not depend on the ends that the killers happen to have,
the reason is a categorical one. That they have violated or ignored it
is the basis of their blameworthiness.
We have here the makings of a second argument for categorical
reasons:
(1) If someone is blameworthy for doing something, then there
is a reason for that person not to do it.
(2) Autonomous fanatics are blameworthy for their killings.
(3) Therefore there is a reason for these fanatics not to perpetrate such killings.
(4) Such a reason, by hypothesis, is neither the content of one
of the fanatic’s commitments, nor instrumental in securing
or protecting one of his commitments.
(5) Therefore such a reason applies to the fanatic independently of his commitments.
(6) Therefore there is at least one categorical reason.
There are only three premisses to the argument. I think that each
one is highly plausible.
Premiss (4). Premiss (4) is a stipulation that comports with the relevant possibilities, and should be granted by all parties to the debate.
It is easy to conjure situations in which perfect instrumentally rational deliberation, begun from a fanatic’s existing commitments,
would generate no consideration that opposed his deadly undertak©2009 The Aristotelian Society
Proceedings of the Aristotelian Society, Vol. cix, Part 2
doi: 10.1111/j.1467-9264.2009.00264.x
196
RUSS SHAFER-LANDAU
ings. It seems to me that any weak points contained in the argument
would have to be found in association with premiss (1) or premiss
(2). So let’s turn our attention to these.
Premiss (1). Premiss (1) seems to me a conceptual truth. If there are
no reasons for an agent to refrain from a given action, then committing it cannot merit blame. Being deserving of blame entails that one
has (at a minimum) ignored a relevant consideration that opposes
the action that one has performed. If one has complied with all relevant and applicable reasons, then there is no room for criticism.
And if there is nothing criticizable about an agent’s actions, then the
person is not properly blameable for his or her behaviour. So a person is blameworthy for an action only if there is some reason that
stands against it. That is what the first premiss says.
There are two important routes to criticizing this first premiss,
but I don’t think that either of them is successful in the end. The
first argues by counterexample. The second claims that blameworthiness requires only that there be a reason that opposes the culpable action, but not that there be a reason for the agent to refrain.
Let’s take these criticisms in order.
(a) Here is a purported counterexample to premiss (1): a person intentionally sets out to do something that strikes us as just terrible.
He has no justification and no excuse. So he is blameworthy. But
through his inadvertence, or some fortunate accident, what he ends
up doing isn’t at all bad, and may even be quite good. Imagine a disgruntled employee who believes that the coffee he is about to hand
his supervisor is laced with poison. He is delighted—he put the poison in there, and hopes for a lethal effect. But unbeknownst to him,
the ‘poison’ is just saccharine, and the coffee is served up exactly to
his employer’s tastes.
Surely such a person is blameworthy for what he’s done. But
there is no reason to avoid handing someone a cup of coffee prepared just the way he likes it. So we have warranted blame for an
action without there being a reason to refrain from it. Thus premiss
(1) is false.
We can concoct many such similar examples, but I think that they
will all fail of their purpose, and for the same reason. The key to understanding why is to focus on the precise basis for the assignment
of blame. In this case, it is the possession of malevolent intentions.
©2009 The Aristotelian Society
Proceedings of the Aristotelian Society, Vol. cix, Part 2
doi: 10.1111/j.1467-9264.2009.00264.x
A DEFENCE OF CATEGORICAL REASONS
197
There is a reason not to act with such intentions. If there weren’t, it
is difficult to see why we would criticize a person for having them.
As we know, actions can be described in many different ways.
There is no reason to refrain from handing a perfectly tasty cup of
coffee to someone who will enjoy it. But that is not what we are
blaming the employee for. We are blaming him for trying to poison
his boss. And there is a reason not to do that. Further, it’s precisely
because the employee ignores this reason, or takes this disfavouring
reason as a favouring reason, that we believe him to be culpable for
the way he’s acted in this case.
For all such examples—ones in which a person is clearly culpable, though his action under at least one description is innocuous or
even attractive—we must attend carefully to the specific aspect of
his behaviour that is earning criticism. My contention is that once
we do this, we will identify a reason that is being ignored or intentionally disrespected, and that this indifference to the force of applicable reasons is going to explain the blameworthiness of the agent
in question. If that is right, then we don’t yet have grounds for rejecting premiss (1).
(b) Here is another criticism of the first premiss. This criticism begins with a subtle distinction between a reason that opposes an action, and a reason for an agent which opposes that action. Critics
who advance this second objection concede that blameworthiness
requires a reason that opposes the blameable action. But they deny
that this reason must be one that applies to each blameable agent. In
effect, they reject my argument’s first premiss,
(1) If someone is blameworthy for doing something, then there
is a reason for that person not to do it,
while endorsing a close cousin:
(1⬘) If one is blameworthy for doing something, then there is a
reason not to do it.
Critics will argue, however, that (1⬘), when conjoined with premiss
(2), does not yield (3). And they are right about that. So if we must
replace (1) with (1⬘), my argument collapses.
Here is the core thought that underlies this criticism. Warranted
blame occurs within a practice that is defined by a legitimate set of
rules that are effective in achieving a valuable purpose. Blame is de©2009 The Aristotelian Society
Proceedings of the Aristotelian Society, Vol. cix, Part 2
doi: 10.1111/j.1467-9264.2009.00264.x
198
RUSS SHAFER-LANDAU
served whenever these justified rules call for it, and they may do so
even when there is no reason for a given individual to abide by these
rules.
Consider a familiar illustration. Suppose that hard determinism is
true, but rather than abandoning our practices of blame, we choose
to retain them. We implement them through a set of rules that assign blame when it is socially efficacious to do so. There may be no
reason for an agent who is unconcerned with the rules (and the consequences of violating them) to refrain from disobedience. But this
will not immunize her from blame, on the assumption that the rules
which specify its assignment are justified.
Once we grant that assumption, it is true that there is a reason to
adhere to the relevant rules. This reason is provided by the justification of the social practices that the blaming rules are designed to enforce and protect. But the reason is a general one; it need not apply
to each person whose actions fall within the scope of these rules. So
(1⬘) is true, but (1) is not. And that’s bad news for me, since it is (1)
that is needed to make my argument go through.
I agree that (1) is what is really needed. But I also think that the
challenge to it is unsuccessful. Any system of warranted blame must
allow for the existence of legitimate excuses. And it seems to me
that an agent has an excellent excuse if there was no reason for her
to refrain from doing what she did. If a justified system of rules and
criticism fails to provide an agent with a reason for obedience, then
it is illegitimate to criticize her for failing to meet its standards.
Blame points to a personal failing. But there is nothing necessarily
amiss with the character of an agent who strays from norms that fail
to provide her with reasons. Indeed, it would be grossly unfair to
criticize a person for flouting rules that supply her with no reasons
for compliance. (1⬘) allows for this kind of unfairness. (1) does not.
(1⬘) severs the tie between blameworthiness and personal shortcoming. (1) does not. These points give us reason to prefer (1) to (1⬘).
Premiss (2). Premiss (2) tells us that autonomous fanatics are blameworthy for their killings. It can be supported thus: if any agents are
blameworthy for their actions, surely those who are bent on evil are
among them. This is so whether the immoralists are doing evil for
its own sake, or doing what is in fact evil, all the while characterizing their actions to themselves as ones that are aimed at a good. An
informed, uncoerced, rational fanatic is the perfect exemplar of the
©2009 The Aristotelian Society
Proceedings of the Aristotelian Society, Vol. cix, Part 2
doi: 10.1111/j.1467-9264.2009.00264.x
A DEFENCE OF CATEGORICAL REASONS
199
blameable agent. His consistency is no proof against criticism. His
intelligence and cunning, his ability to select appropriate means to
his chosen ends, render him more, rather than less, liable to blame.
The standard exculpation conditions do not apply here. The dedicated evildoers are not compelled to act as they do, but have chosen
their path and have ruthlessly pursued it in the absence of duress,
coercion, necessity or factual ignorance.
As far as I can tell, there are only three ways to try to falsify the
second premiss, and so deny that autonomous fanatics are blameworthy for their murderous deeds. The first emphasizes the importance of a reason’s accessibility as a precondition of blame. On this
line, fanatics are immune from blame, because their existing commitments will prevent them from seeing the reasons that oppose
their misdeeds. If they cannot see these reasons, then (it is said) they
are not blameable for ignoring them. The second way to reject
premiss (2) is to deny the existence of autonomous fanatics. The last
is to deny that anyone is blameworthy for anything.
(a) The first way in which we might deny that blame accrues to autonomous fanatics is to insist that being blameworthy for something
requires that an agent has a reason that she has failed to adequately
respect. Having a reason implies the reason’s accessibility—one has
a reason to f only if one either appreciates that reason, or can, in
some suitable way, be brought to appreciate it. And since the relevant evildoers are those whose commitments bear no instrumental
relation to the relevant reasons to refrain, such people do not have
any reason to avoid undertaking their atrocities. This means that
they are not to blame for the harm they do.
The following counter-argument sets out the relevant line of
thought:
(1ca) If an agent is blameworthy for doing something, then that
agent must have an accessible reason not to do it.
(2ca) Dedicated immoralists with no commitments that would
be served were they to refrain from evildoing do not have
any accessible reason to so refrain.
(3ca) Therefore such immoralists are not blameworthy for their
evil deeds.
Although I cannot give a full assessment of this argument, I think
that there are problems for both premisses, and that the accommo©2009 The Aristotelian Society
Proceedings of the Aristotelian Society, Vol. cix, Part 2
doi: 10.1111/j.1467-9264.2009.00264.x
200
RUSS SHAFER-LANDAU
dations needed to make its conclusion palatable lead directly to a
recognition of categorical reasons.
Though initially plausible, premiss (2ca) may yet be false. Depending on how we are to understand a reason’s accessibility, it may
be that a reason is accessible to an agent even if acting upon it does
not serve one of her pre-existing commitments. For one might arrive
at true practical conclusions through a variety of routes that extend
well beyond flawless instrumental reasoning. If that is so, then
premiss (2ca) could be false for many or even all dedicated immoralists.
There is also reason to worry about (1ca). The premiss gains support from this thought: if we are rightly subject to blame for something that we did (or failed to do), then we must have had an
opportunity to avoid the blameworthy behaviour. And there is such
an opportunity only if the bases of the blame—the considerations
whose flouting explains our culpability—are accessible to us. If we
cannot recognize the error of our ways, then we are blameless for
our deeds, since we will have had no effective opportunity to avoid
culpability. So dedicated immoralists (like the rest of us) are truly
blameless if the reasons that oppose their actions are inaccessible to
them.
I am unsure whether these reasons really are inaccessible (see the
brief discussion of (2ca) above), but let’s proceed as if they were.
Still, an inability to appreciate the existence or force of such reasons
does not immunize us from blame, if we are blameable for having
endorsed our ends in the first place.
To see this, imagine a person who has promised another to meet
him at a certain place and time, but then, through her culpable negligence, finds that it is impossible to fulfil the promise. This current
inability does not cancel her liability to blame. So, too, if the fanatic’s prior culpable choices are rendering him unable to see the merits
of refraining from his actions, then his present inability to appreciate these considerations does not immunize him from blame.
The question thus devolves to one about whether the fanatic’s initial choices to ally himself with evil ends are choices that he is
blameable for. And why wouldn’t they be? I am not imagining a person who has been brainwashed or neurologically manipulated, but
someone who makes choices that are as uncoerced and as informed
as those of anyone with more ordinary moral preferences. As far as
I can see, the only reason to suspend blame here is owing to an as©2009 The Aristotelian Society
Proceedings of the Aristotelian Society, Vol. cix, Part 2
doi: 10.1111/j.1467-9264.2009.00264.x
A DEFENCE OF CATEGORICAL REASONS
201
sumption that no one’s choices are blameable. Such a view may be
true. But so long as we are willing to blame anyone for the choices
that he makes, then we should be prepared to blame the fanatic for
his. And that means that his subsequent immoral choices and actions, even if they are endorsed by his instrumentally rational deliberations, are ones for which he is blameworthy. That is just what
premiss (2) states.
Defenders of the counter-argument know that freeing the worst
among us from blame is not a pleasant prospect. They may try to remove the sting of its conclusion by insisting that they are still able to
seriously criticize the various malefactors we have been discussing.
We can’t say of them that they have culpably done what they ought
not to have done, for (on this line) this would imply that they had a
reason to do otherwise. But we can say of them that they are evil
people, that they are abhorrent, and that their actions are despicable, atrocious, etc. We can register certain negative evaluations, and
strong ones at that.1 All we are blocked from doing is ascribing culpable wrongness to such agents, since proper blame entails an accessible reason that an agent is failing to adequately respect.
I think that there is a deep problem with this line of thought. The
problem is this: all of these negative evaluations themselves presuppose the existence of reasons that have been ignored. We are meant
to be barred from depicting consistent immoralists as blameworthy.
Yet we are allowed to say that they are evil, awful, horrific, etc. But
what is the basis for such criticisms? Isn’t is the very fact that these
agents have not paid sufficient attention to relevant, highly important (indeed peremptory) reasons to refrain from their misdeeds? To
merit such epithets means that these agents have failed to adequately respect the reasons that oppose their actions.
There must be something that explains why an agent’s actions
qualify as atrocious, depraved, evil, etc. The explanation will require citation of considerations that oppose such actions. And such
considerations are reasons. By hypothesis, these reasons are not
ones that the moral monsters have. But they are not thereby immune from all evaluative criticism. And this forces us to ask: why
should the assignment of blame require that an agent have an accessible reason that she is ignoring, although other very strong (quasi-)
1
See Harman (1975) and Williams (1995) for this criticism of premiss (1), as well as for this
specific line of defence regarding negative evaluations of the consistent evildoer.
©2009 The Aristotelian Society
Proceedings of the Aristotelian Society, Vol. cix, Part 2
doi: 10.1111/j.1467-9264.2009.00264.x
202
RUSS SHAFER-LANDAU
moral criticism does not? I don’t think there is a good answer to this
question.
Instrumentalists see how jarring it is to be unable to denounce the
moral monster. So, while denying that he is blameworthy, they nevertheless insist that he is evil, awful, demonic, etc. Such criticisms
implicate the existence of reasons that have been ignored. These reasons exist even if the moral monsters do not have them. These reasons exist even if they bear no instrumental link to the commitments
of such agents. These are categorical reasons.
Admittedly, I have not offered anything like a full evaluation of
the counter-argument (1ca)–(3ca). But for present purposes we can
do without such an assessment. For even if we were to ignore my
reservations about its premisses and concede the soundness of the
counter-argument, this would be at best a Pyrrhic victory for instrumentalists.
To see this, assume for the moment that the conclusion of the
counter-argument is true, and that dedicated immoralists are not to
blame for their actions. We must then ask whether they are properly
subject to other, strong forms of evaluative criticism. Summing up
my line of replies from the previous paragraphs, we can see that
there is trouble either way.
Suppose they are beyond all criticism. Those who blow up buses
of schoolchildren or detonate bombs at civilian funerals are not evil,
malign, or depraved; their actions are not horrific, atrocious, odious,
or despicable. There is surely something deeply wrong with an instrumentalist position that licenses such immunity. Yet if dedicated
immoralists are properly subject to these negative evaluations, this
susceptibility to criticism implies the existence of categorical reasons.
On the safe assumption that dedicated immoralists really do merit
some of the criticisms just mentioned, then categorical reasons
exist—even if such evildoers are not blameworthy for their deeds.
(b) A second criticism of premiss (2) claims that autonomous fanatics are not blameworthy for their actions, because there can be no
such thing as an autonomous fanatic. This criticism, sometimes
heard in Kantian corners, strikes me as highly implausible. If the
claim is more than an instance of a wholesale denial of personal autonomy, then there must be some special reason that agents are unable to autonomously elect evil, though they are able to freely attach
themselves to the good. But what could this special reason be? The
©2009 The Aristotelian Society
Proceedings of the Aristotelian Society, Vol. cix, Part 2
doi: 10.1111/j.1467-9264.2009.00264.x
A DEFENCE OF CATEGORICAL REASONS
203
evil, recall, need not be conceptualized as such—the autonomous
fanatic may tell himself that what he is doing is good, and be a dedicated evildoer nonetheless. And he may surely pursue what he really cares about while free of coercion, and in possession of relevantly
full information. Certainly, absent clear and compelling argument,
we are warranted in abiding by the general maxim that anything apparently conceivable is possible. It appears that we can conceive of
the autonomous fanatic. Thus, absent a very strong argument to the
contrary, we are right to suppose that such fanatics can exist.
(c) A last basis for rejecting premiss (2) comes from the assertion
that no one is blameworthy for anything. This might be true. If so,
my second argument is unsound. I can’t say anything here to falsify
this potential criticism. All I can do is to express the conviction,
shared by almost everyone, that at least some people are rightly
blameable for their poor choices and actions. The examples used to
substantiate this conviction seem to me more compelling than any
of the premisses employed in arguments to defeat them.
Because my second argument rests in part on this undefended
conviction, it is best to conceive of its conclusion conditionally: if
anyone is blameworthy for any of her choices, then there are categorical reasons. We get to this conclusion by means of a conceptual
truth (premiss 1), an uncontroversial statement of possibilities
(premiss 4), and a highly plausible premiss (2) that expresses a deeply commonsensical assessment of evildoing.
IV
The notion of a categorical reason has a storied past, and this lineage may contribute to some misunderstandings that it is best to
forestall. There are certain elements that are commonly associated
with the idea of a categorical reason, but whose defence is no part
of my current brief. I am thinking specifically of the notions that
categorical reasons always override any possibly competing reasons,
that they apply of necessity to all rational agents, and that they exist
either by virtue of being the outcome of successful rational deliberation or by virtue of being entailed by presuppositions common to all
exercises of rational agency.
The categorical reasons whose existence I am defending are not
©2009 The Aristotelian Society
Proceedings of the Aristotelian Society, Vol. cix, Part 2
doi: 10.1111/j.1467-9264.2009.00264.x
204
RUSS SHAFER-LANDAU
defined by reference to any of these features, but rather are to be understood simply as those reasons that exist independently of any instrumental link to an agent’s existing commitments. A fuller account
of the source of such reasons may lead naturally to one or more of
these traditional ideas, but they are not essential elements in the story that I wish to tell here.
I have offered the beginnings of a case for categorical reasons.
The complete account would have to reconstruct and criticize the
arguments for instrumentalism that have been offered by numerous
philosophers. That is certainly a task for another day. But I think it
reasonable to venture a thought about where we stand prior to such
an extensive examination of instrumentalist arguments. And that, I
think, is squarely on the side of categorical reasons.
The explanation for this is one that Philippa Foot has recently offered, in the recantation of her earlier instrumentalist views (Foot
2001, ch. 2).2 She has become puzzled by the predominance of instrumentalist views, and thinks that a more natural starting point for investigating practical reasons is one that assigns intrinsic importance
to considerations of self-interest, morality, and what we care about.
Indeed, when we reflect on deeply held assumptions about the ultimate sources of our reasons, it strikes her, as it does me, that we do
and ought to start with a recognition that each of these three sources is fundamental, and that none is to receive preferential status over
the other two. This does not foreclose the possibility that excellent
arguments may be introduced to reduce the sources from three to
two, or perhaps even to one. But absent such arguments, we are
right to reject the demand that we begin our thinking with the claim
that just one of these sources is to be privileged above the others. Instrumentalism makes this very claim, and is therefore suspect.
Though it may eventually overcome our suspicions, the burden is
on instrumentalism’s defenders to justify such a move. It should not,
in any event, be regarded as the default view about the nature and
source of our practical reasons. If we are ever to accept instrumentalism, we must not only find fault with the arguments that I (and
others) have offered on behalf of categorical reasons. We must also
be impressed enough to move away from the default position of pluralism about the ultimate sources of practical reasons.
2
She there expresses reservations about the views defended in Foot (1972).
©2009 The Aristotelian Society
Proceedings of the Aristotelian Society, Vol. cix, Part 2
doi: 10.1111/j.1467-9264.2009.00264.x
A DEFENCE OF CATEGORICAL REASONS
205
V
The most powerful kind of philosophical criticism is one that reveals a contradiction in its target. I have not presented such a criticism of practical instrumentalism, because I do not believe that
instrumentalism entails a contradiction. Nor must instrumentalists
exemplify any kind of practical inconsistency in behaviour or commitment. Most defenders of categorical reasons, following Kant,
have tried to sustain such charges. Their vindication would be welcome news for friends of categorical reasons. But I am not optimistic about this most direct route to instrumentalism’s refutation.
If a view is not internally contradictory, then any successful criticism of it must proceed by adducing non-conclusive but highly plausible reasons, cogently put together to make a strong, albeit
defeasible, case. That is what I have tried to do here. Of course, what
counts as a plausible reason is relative to antecedent beliefs and commitments. If one is already devoted to instrumentalism, then one will
find the considerations I have offered less plausible than anyone else.
But that does not distinguish the instrumentalist from (say) the sceptic about other minds. Such people can have internally consistent
views, and will regard with great suspicion the supporting evidence
introduced by their critics. Still, for those not antecedently wedded
to this scepticism, the falsifying evidence can be compelling.
I think that the very same thing is true of practical instrumentalism. We cannot prove that there are categorical reasons. But when
we vividly contemplate a world without them, one in which there is
literally no consideration that stands against the actions of a torturer, and none in favour of easily rescuing a child from imminent
death, most of us will find that instrumentalism has as much appeal
as the various sorts of scepticism that we take seriously only in the
study.3
3
This paper has gone through a number of different incarnations over the past couple of
years. I am grateful to Harry Adamson, Simon Blackburn, Jerry Cohen, Steve Darwall,
David Enoch, David Killoren, Hallvard Lillehammer, Mike Martin, Ellie Mason, Anthony
Price, Peter Railton, Geoff Sayre-McCord, Tom Senor, Barry Ward and David Wiggins for
forcing me to think much harder about some of the claims and arguments that appeared in
earlier drafts. Many thanks to audiences at the University of Michigan, Edinburgh University, Northern Illinois University, the University of Arkansas-Fayetteville, Union College,
the Institute for Advanced Study at Hebrew University, the Colloquium in Legal and Social
Philosophy at University College London, the Aristotelian Society, and the Moral Sciences
Club at Cambridge University for their acute comments on earlier versions of this essay.
©2009 The Aristotelian Society
Proceedings of the Aristotelian Society, Vol. cix, Part 2
doi: 10.1111/j.1467-9264.2009.00264.x
206
RUSS SHAFER-LANDAU
Department of Philosophy
University of Wisconsin, Madison
600 N. Park Street
Madison, wi 53726
usa
[email protected]
REFERENCES
Foot, Philippa 1972: ‘Morality as a System of Hypothetical Imperatives’.
Philosophical Review, 81, pp. 305–16.
——2001: Natural Goodness. Oxford: Oxford University Press.
Harman, Gilbert 1975: ‘Moral Relativism Defended’. Philosophical Review, 83, pp. 3–22.
Williams, Bernard 1995: ‘Internal Reasons and the Obscurity of Blame’. In
his Making Sense of Humanity. Cambridge: Cambridge University Press.
©2009 The Aristotelian Society
Proceedings of the Aristotelian Society, Vol. cix, Part 2
doi: 10.1111/j.1467-9264.2009.00264.x