zyxwvutsr
zyxwvut
zyxwv
@ Basil Blackwell Ltd. 1993, 108 Cowley Road, Oxford OX4 IJF, UK
and 238 Main Street, Cambridge, MA 02142, USA.
Ratio (New Series) VI 2 Deccmbcr 1993 00344006
DISCUSSION
zyxw
MOTIVATION, UNIVERSALITY AND T H E G O O D
A critical notice of Jonathan Dancy, Moral Reasons, (Oxford:
Blackwell, 1993).
Roger Crisp
Professor Dancy’s book is a powerful, imaginative and stimulating
defence of several related views on some of the oldest questions in
moral philosophy. Dancy approaches these questions in a refreshingly nonpartisan way, and embeds his discussions soundly within
the parameters of current debate. I shall comment on what he says
about the springs of moral action (section I ) , the universality of
moral reasons (section 2) and problems with consequentialism
(section 3 ) . I n addition, Dancy has much of interest to say on the
following: silencing and virtue, supervenience and universalizability, conflict and regret (where he introduces his idea of a
situation’s having a certain ‘shape’), and supererogation. His
discussion of objectivity and the analogy between values and
secondary qualities is one of the clearest and most useful I have
read. I remain unpersuaded by several of Dancy’s main arguments.
But there is no doubt that this is a significant book. Anyone
working seriously on any of the above topics must read at least the
relevant sections.
zyxw
I . Moral Motivation
The debate about how to explain moral motivation is usually
couched in Humean terms, the various positions taken differing in
the r6le they ascribe to desire. Dancy aims to move beyond this
debate by arguing that a theory of motivation requires nothing
resembling desire as conceived of by Hume. This attempt is highly
original and interesting. But I believe it ultimately fails because
Dancy has no answer to a powerful argument for the Humaen view.
Internalism about moral motivation is the view that a person
who makes some moral judgement will necessarily be motivated by
it. Dancy defines a non-cognitivist in ethics as someone who ‘sees
moral judgements as expressions of desires’ (p. 3). So, if these
182
zyxwvut
zyxwv
ROGER CRISP
desires are the ones we expect, the non-cognitivist will be an
internalist. A cognitivist, however, can be either an internalist, who
claims that moral judgements express necessarily motivating
beliefs, or an externalist, who requires for motivation the addition
of a desire.
Externalist cognitivists, such as W. 1). Ross, have been attacked
for their externalism. Dancy suggests that this is because there is a
strong intuitive argument in favour of internalism. This is based on
‘the sense that morality is essentially practical, so that i t would be
odd for someone to say “This action is wrong but I don’t see that as
at all relevant to my choice” ’ (p. 4). Dancy spells out the
argument with the claim that a moral imperative is, at least in a
weak sense, categorical. It has ‘practical relevance’ for someone
who accepts it which does not depend on the presence of a desire.
This argument appears to depend on an equivocation on the
notion of ‘practical relevance’. I n one sense, an imperative is
practically relevant for me if it applies to me. Salman Rushdie
could have accepted that in this sense certain Islamic imperatives
had practical relevance for him. I n another sense, an imperative is
practically relevant for me if I am concerned about it. Rushdie was
not concerned about the Islamic imperatives, and might have
repudiated them ‘by saying “I don’t care about that sort of thing” ’
(p. 4). It is at least possible that a Thrasymachean character might
make the same sort of claim about morality. Of course, we should
find such a character ‘odd’. But, the externalist will suggest, this is
not because we assume that moral judgements necessarily motivate,
but because we find that most people are in fact motivated by them.
Dancy in fact goes on to accept that, in the debate as it stands,
the externalist position is stronger. This is because of the person
with ‘accidie’, that is, the kind of moral indifference often caused by
depression. The strongest form of internalism, according to Dancy,
is a motivated desire theory, according to which a moral belief gives
rise to a desire, and an action results. But, Dancy suggests, in the
case of a person with accidie desire may be absent.
Dancy attempts to transcend the internalisdexternalist debate
by rejecting the Humean thesis that the r6le of belief in motivation
is inert. According to his ‘pure theory’, belief alone can motivate.
Imagine that I believe that my close friend is in distress, and that I
believe also that were I to comfort her, her distress would be
alleviated. According to Dancy, these two representations may be
sufficient to motivate me. Dancy does speak of desire, but it turns
out that desire consists just in my being motivated by the gap
zyxw
@ Basil Rlackwell 1,td. 1993
zyxwv
M O T I V A T I O N , IJNIVERSA1,ITY AND T H E GOOD
183
between the two representations (p. 19). He might, then, have
avoided talking of desire at all, speaking only of motivation.
There might be thought to be a problem here with weakness of
will. I n the case just mentioned, according to the pure theory, my
beliefs alone were sufficient to motivate me. But if I were weakwilled I would not be motivated. So the pure theory is lacking the
notion of something like a desire which will explain why I do
act.
Dancy avoids this problem with an interesting differentiation of
two distinctions: that between essentially and contingently motivating states, and that between states which do and states which do
not motivate in their own right. According to him, motivating
representations motivate in their own right, but not essentially.
Many are likely to feel, however, that the pure theory still fails to
explain successful motivation. This point can be put in terms of the
directions of fit between states of the agent and the world. Dancy
expresses this distinction in terms of beliefs’ being such as to aim to
be caused by the truth of their content, while desire are such as to
aim to cause their content to become true (p. 28). O n the pure
theory, both representations have the direction of fit of belief. But
there is a simple argument for the Humean claim that an element
with the opposite direction of fit is required to explain action.’
Actions are attempts to change the world. Therefore, some state
with mind-to-world direction of fit is required to explain action.
Dancy’s attempt to do the job of desire in explanation of action
with the second, conditional representation fails (pp. 28-9). I n the
case of my distressed friend, I may also believe that were I to insult
her, she would become even more distressed, that were I to ask her
what is the capital of France, she would reply ‘Paris’, and so on. W e
require an explanation of why I act on the second representation on
which I do act, and this will be in terms of a state functionally
equivalent to a dcsire. If this argument is correct, and if Dancy is
right about the accidie problem, the most plausible form of
cognitivism will be externalist.
Dancy denies the force of the argument based on directions of fit,
and attempts to explain the attraction of the belief/desire thesis as
resting on the assumption of a Cartesian split between the agent
and a n inert world. States of such a world could not motivate, since
they could not constitute reasons (pp. 30-1). H e goes on to suggest
’
z
zyxw
See Michael Smith, ‘The Humean throry of motivation’, Mind 96 (1987), pp. 54-5.
@ Basil Rlackwrll I.td. 1993
184
zyxwvut
zyxwvu
zyxwv
zyxw
ROGER CRISP
that the notion of directions of fit requires such a Cartesian
metaphysics, and then criticizes this metaphysics. Dancy claims that
it is not really even beliefs which motivate, but matters of fact
(what motivates me is her being distressed and its being the case
that I could help, not my belief that this is so). Since Cartesianism
cannot make room for ‘facts which intrinsically make a difference to
how we should act’ (p. 3 2 ) , both it and the notion of directions of
fit which rests upon it are mistaken.
O n the very first page of his book, Dancy rightly points out that
in his discussion of internalism Nagel fails to distinguish between
the truth of a proposition and its being believed. But here Dancy
does not separate the notion of the content of a belief from that of a
fact, nor that of a motivating reason from what we might call an
agcnt’s reason. I may believc that she is distressed; in fact she has
hay fever. I am motivated by among other things my belief that she
is distressed; if you ask me why I a m trying to help, I shall probably
refer you to the content of my belief and say, ‘She is distressed’. So
the only matter of fact that could motivate me is one involving the
content of my belief.
Things are different with justzfication. By talking of facts as
reasons here, Dancy has moved away from questions of motivation
and into those of justification. True, Cartesians as described by
Dancy will find it difiicult to make sense of the notion of a fact’s
being a justifying reason for action. But someone who denies
Cartesianism and is therefore able to speak of facts’ being justifying
reasons can still appeal to the Humean argument for the belief/
desire thesis based on the notion of directions of fit.
The failure to distinguish between motivating reasons and the
reasons a n agent would offer in explanation or justification of her
action perhaps explains why Dancy is prepared to accept that his
second representation - which appears inertly to the agent as a
ground for her action - can motivate.
zyxw
2. ParticulariJm
Years ago, Professor Hare distinguished between universality and
generality in moral principles. Universal moral principles apply
universally, but they may be highly complex (e.g., ‘You should not
kill, except in circumstances c, d, e . . .’) and so not general but
specific. Dancy attacks universality, though he calls it generality. I
suspect that this use of terminology may reflect a mistaken
assumption on Dancy’s part t h a t universal principles must be
@ Basil Blackwell I.td. 1993
zyxw
MOTIVATION, UN1VERSAI.ITY A N D T H E GOOD
185
general. For many of his criticisms work more effectively against
generality than universality.
According to Dancy, we should accept that a belief-state can be
sufficient to motivate in one case and yet not in another without
feeling bound to introduce any additional item into the explanation
of motivation. Moral properties, and the moral reasons which they
ground, behave in a similar way. This is what Dancy calls
particularism: a property F of one action may be a reason to perform
it, while the F-ness of another action may be morally neutral or
even a reason not to perform it.
Arguments against particularism might be a priori, resting on
some account, perhaps, of what counts as a property or a
satisfactory causal explanation or moral justification. T h e argument for particularism, however, must proceed by example. T h e
first example Dancy gives is that of the property ‘giving pleasure’
(p. 56). In many cases, the presence of this property appears to
provide a reason for doing an action or approving of it. But it can
also provide a reason against, as when I tread on a worm for
pleasure.
What should be the generalist (in Dancy’s sense, not Hare’s)
response to such an example? One response - allegedly made by
Hare (p. 56) - is to suggest that the pleasurableness of the wormcrushing is in fact a rzaJon for doing it, though the balance of reasons
may well be against doing it. This response does not seem entirely
unacceptable. Explanation and justification are in practice fairly
closely connected. Imagine a case where a person treads on a worm
and can offer no decent explanation for what she has done. She
appears to have no justification available either. But someone who
explains his worm-crushing by saying that it gives him pleasure
puts his act in a context where we can not only make sense of it, but
also see that he did in fact have Jome justification for it. Of course,
this justification may not be a moral one.
An alternative generalist response is to hone further the original
suggested reason. Perhaps it is not pleasurableness which justifies,
but pleasure-in-what-is-not-evil. Yet another option would be to
combine elements from both of these responses, claiming that there
is a reason to do the action in so far as it is pleasurable, and a
reason against in so far as it involves pleasure-in-evil.2 This move of
extending or revising principles in the face of counter-examples
again seems quite respectable. Indeed it might be suggested that
z
zyxw
zyx
zyxwv
zyxw
*
zyxwvuts
Thomas Hurka suggestcd this option to mc
@ Basil Rlackwpll Ltd. 19%
186
zyxwvut
zyxw
ROGER CRISP
Dancy himself engages in it, when, for example, he attempts to
make sense of the notion that there is an absolute constraint on
killing by refining what might be taken to be a substantive principle
forbidding killing into a ban on killing the innocent and then into a
ban on killing the innocent and unwilling (to allow for voluntary
euthanasia) (pp. 229-30).
Dancy has other examples to support his holistic conception of
reasons (the view that the reason-giving force of properties depcnds
on which other propcrties are present). One concerns my borrowing
a book from you and then finding that you have stolen it from the
library. Usually the fact that I have borrowed the book from you
would be a reason to return it, but here it is no reason at all.3 A
second example concerns a family game which involves lying. In
the game, the fact that an action is a lie is a reason to do it, despite
the fact that a n action’s being a lie is usually a reason against doing
it.
There is a distinction between non-ultimate and ultimate reasons
for action. Ultimate reasons provide ultimate grounds for actions,
and they can be expressed in the form of general principles. A
hedonist, for example, will take the property of pleasurableness as
constituting the only ultimate ground for doing anything. But this
is not to say that a hedonist cannot accept non-ultimate reasons
which justify particular actions in one case and count against them
in others. Indeed these reasons can support or count against
different actions at the same time. Dancy claims that such cases are
the ‘most extreme’ he can offer (p. 62).
His example is that of sending the third article in a series to a
journal which has already published the first two. T h e fact that the
journal has already published two in the series can be said to count
as a reason for publishing, and as a reason against publishing. Nonultimate reasons, even for the hedonist, the paradigm generalist,
clearly are holistic. But this providcs no reason to assume holism
goes all the way up. Indeed quite the opposite is the case. For nonholism at the ultimate level will provide us with a plausible account
of how and why reasons function as they do at the non-ultimate
level. In the case of the article, the hedonist can argue that the
general principle that pleasurableness always speaks in favour of an
action explains how the fact that the journal has already published
two articles in the series can function both as a reason for and a
z
:’ A perhaps pedantic objection to this example would be that one can borrow only what
loaned, and that one can loan only what one owns.
@ Basil Blackwell Ltd. 1993
z
IS
MOTIVATION, UNIVERSALITY AND THE GOOD
zyx
187
reason against. Publishing, perhaps, would give pleasure to those
who enjoyed the previous two, while not publishing would make
room for some other article which people not interested in the
original two articles will enjoy reading. A pluralist at the ultimate
level might account for this case with the suggestion that, say,
variety and unity are both valuable.
Now consider the library case. The first response the generalist
might make is analogous to the first made in answer to the wormcrushing: I do have some reason to return the book. The second
possible response is, however, more plausible, viz., that the claim
that one should return something one has borrowed is too loose (as
was suggested in the first book of Plato’s Republic). Consider, as an
alternative, the more plausible principle that property rights
should be respected. This will explain how the non-ultimate fact
that you have borrowed something can sometimes function as a
reason for returning the borrowed object, and sometimes not.
Similar strategies can be employed in generalist responses to the
lying case and the other examples Dancy uses in support of
particularism.
Dancy allows that a consequentialist can be a particularist
(p. 232, n. 15). It should be noted also that a generalist who allows
for G. E. Moorean ‘organic unities’ can agree with much of
Dancy’s argument about the holistic nature even of ultimate
reasons and argue that explicit substantive principles are nevertheless available on the basis of reflective analysis. One principle
might be, for example, that pleasurableness combined with sadism
is worthless. These principles will be arrived at by honing
principles in the light of reflection on particular cases.
It might be suggested that our primary task in moral philosophy
is to identify our ultimate reasons and state them in the form of
principles. If this were correct, to assume that there are no such
reasons would be to doom one’s project from the start.
zyxwv
3. Agent-relativity
The blurb on the back cover of Dancy’s book says that he will argue
‘that there are more sorts of reason around than consequentialists
can even dream o f . But in fact Dancy clearly hopes to give
consequentialists nightmares. He argues that we have reasons not
to maximize the good, allowing us to pursue our own interests
instead and forbidding us to disobey certain non-consequentialist
moral principles. I agree with Dancy about reasons to promote
@ Basil Blackwell I.td. 1993
z
188
zyxwvut
ROGER CRISP
one’s own interests, though his arguments fail to distinguish
adequately between reasons and values. Because of this, I am
unpersuaded by his argument for moral constraints on maximization.
By ‘consequentialism’, Dancy says that he means a theory that
assesses the moral worth of an action by reference to the difference
it makes to the world that the action was done (p. 167). He argues
that consequentialism cannot make sense of certain reasons that we
take ourselves to have. His definition of consequentialism is, as he
says, ‘extremely flexible’, so flexible in fact that such a theory can
incorporate any kind of reason - including those that Dancy claims
to be inconsistent with consequentialism - by stipulating that
acting on such reasons has value. But what Dancy usually has in
mind when he speaks of consequentialism seems to be any theory,
such as utilitarianism, which advocates solely the impartial
maximization of value.
Dancy accepts the contemporary othodoxy that there is agentrelative value. This arises out of the Nagelian tension Dancy feels
between his belief that his own personal projects matter a great
deal to him and yet do not matter much considered impartially (in
the sense that the world will not be much richer if they succeed).
And what matters impartially often conflicts with Dancy’s personal
concerns: he has both agent-neutral and agent-relative reasons to
act, and there is tension between them.
However we understand the notions of such reasons, we can do
without that of agent-relative value. What makes an object
valuable is its possession of certain properties. There should be a
single fact of the matter whether any object possesses such
properties. Let us assume that Dancy’s personal projects represent
a great accomplishment. Then they are valuable, and this gives
him - and anyone else - a reason to promote them. But Dancy is
clearly right to say that his projects matter more to him than to
others, and that practical rationality requires of him greater
attention to his projects than it requires of others to those projects.
But what gives him the reason to devote attention to his own
projects is that they are his projects, not that they possess some
special value. There is a tension between different reasons here, but
there is nothing mysterious about reasons’ conflicting, and certainly
nothing which requires us to introduce agent-relative value or
different Nagelian ‘perspectives’.
One might try restating Dancy’s arguments without using the
notion of agent-relative value. At times, he himself comes close to
zyxwv
zyxwvutsrq
@ Rasil Blackwell Ltd 1993
zyxw
zyxwv
zyxwv
zyxw
MOTIVATION, UNIVERSALITY AND THE GOOD
189
recognising that agent-relative reasons are grounded in the
separateness of persons and not in a separate sphere of value. He
defends ‘options’ - that is, moral permissions to pursue one’s own
good at the expense of the overall good - by reference to the cost to
the agent. This cost will of course be counted in calculating the
overall good.
But that the cost is to be his cost makes a great difference to [the
agent], and here the identity of the payer makes a difference.
(P. 213)
In the very last paragraph of the book Dancy recognises that not
everything that functions as a reason is a value. Unfortunately,
however, Dancy allowed himself to be argued out of distinguishng
reasons and values (p. 214). This distinction is required to make
sense of the claim that the fact that this is my life in itself grounds a
reason for my giving that life some priority in my practical
reasoning.
There is one place where the argument cannot be restated. This
is in Dancy’s defence of agent-relative ‘constraints’ on our freedom
to promote the good. H e wishes to deny Samuel Scheffier’s thesis
that the rationale for options is distinct from that for constraints.
Dancy argues that the ‘moral cost’ to the agent grounds constraints
in a way analogous to that in which cost seen as agent-relative
value ground options. But once we see that the ground for options
is not the agent-relative value of the cost but the fact that that cost
is to be paid by the agent, an analogy used to support constraints
breaks down.
Dancy is right to claim that consequentialism cannot make room
for agent-relative reasons. But his criticism of consequentialism
also extends to those forms of ‘split-level’ consequentialism
advocating agent-relative reasoning which is itself grounded on
agent-neutral reasons.
Dancy discusses Parfit’s version of such a theory, and suggests
that it is self-contradictory. For Parfit’s theory gives us an aim that outcomes be as good as possible - which we ought to pursue,
and then tells us not to pursue it, but to give some priority in our
practical thinking to ourselves and those close to us.
Though Dancy is right to doubt the coherence of Parfit’s notion
of ‘blameless wrongdoing’, the appearance of a contradiction here
is illusory. Parfit’s consequentialism does give us the ultimate aim
that outcomes be as good as possible, and it does tell us to pursue it.
But it also tells us how to pursue it, viz., by giving priority to
0Basil Bla~kwellI.td
1993
zyxwv
zyxwvu
zyxwv
zyxwvu
zyxwvutsr
zyxwvu
zyxwvutsr
190
ROGER CRISP
ourselves and those close to us. We could express this general claim
without using the notion of an ultimate aim. Consequentialism
states that we have the strongest reason to live that life which will
maximize overall value. It says also that such a life is most likely to
be lived by those who do not try consciously a t all times to
maximize value, but give some priority to themselves and those
close to them .4
St Anne's College
Oxford
' I am gratcful to Brad
@ Basil Blackwell Ltd. 1993
Hooker and Thomas Hurka for comments on an earlicr draft.