Date post: | 18-Apr-2015 |
Category: |
Documents |
Upload: | sanquintin1557 |
View: | 20 times |
Download: | 0 times |
North American Philosophical Publications
The Paradox of LoyaltyAuthor(s): Philip PettitReviewed work(s):Source: American Philosophical Quarterly, Vol. 25, No. 2 (Apr., 1988), pp. 163-171Published by: University of Illinois Press on behalf of the North American Philosophical PublicationsStable URL: http://www.jstor.org/stable/20014235 .Accessed: 09/10/2012 00:07
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .http://www.jstor.org/page/info/about/policies/terms.jsp
.JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range ofcontent in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new formsof scholarship. For more information about JSTOR, please contact [email protected].
.
University of Illinois Press and North American Philosophical Publications are collaborating with JSTOR todigitize, preserve and extend access to American Philosophical Quarterly.
http://www.jstor.org
American Philosophical Quarterly Volume 25, Number 2, April 1988
THE PARADOX OF LOYALTY
Philip Pettit
INTRODUCTION
HTHE ideal of loyalty is at the heart of common -^ sense morality. That morality highlights our
special obligations to ourselves and one another and
duties of personal loyalty to other people bulk large
amongst these; they are the duties that we owe to
other people in virtue of more or less intimate
bonds, whether bonds of family, friendship, com?
radeship, collaboration, or whatever.1
Yet the ideal of loyalty generates a paradox, for it
requires that the loyal agent satisfy two apparently
conflicting assumptions. He must be sensitive to
considerations of a sort that we can all find com?
pelling; that is why it is an ideal. And he must
be sensitive to considerations relating distinctively to the welfare of a particular person; that is why it
is an ideal of loyalty. The first assumption casts
the loyal agent as praiseworthy from an impartial
point of view; the second presents him as the very
examplar of partial concern.
The paradox of loyalty will not be unfamiliar.
One relative of the antinomy lies behind the long? standing challenge to Christian beliefs, that no one
is loved who is loved just for Christ's sake; to be
loved one must be loved for one's own. And another
related antinomy is deployed against Kantian ideals
when it is said that the friend who sticks by a
colleague out of a sense of duty fails to display any sense of friendship.2 In each case we have a
tension between the impartial and partial aspects of the ideals invoked. Love is partial in a way that seems to conflict with the Christian motivation;
fidelity to a friend is partial in a manner that appears to conflict with the Kantian.
The ideal of loyalty is less exotic than the Chris? tian or Kantian ideals but it gives rise to just the same sort of problem. The trouble in those cases
is that if you really love someone you cannot love
them for Christ's sake and that if you really think
as a friend you cannot act out of a sense of Kantian
duty. The trouble in this is that if you are really loyal to a person, it appears that you cannot be moved by considerations that recommend themselves gen
erally. To be loyal is to be dedicated to a particular individual's welfare and that seems to conflict with
the idea that the loyal agent is idealistic or dutiful.
It is now common among philosophers to distin?
guish between considerations which play a critical
role in regard to our reasoning about what to do and
considerations which play a practical or operational role.3 Critical considerations are those we invoke
when we are under pressure to justify or revise our
habits of decision and action. Operational con?
siderations are those we deploy in the exercise of
those habits. This distinction suggests an easy resolution of the
paradox of loyalty. The resolution would be that
loyalty in operational practice means acting on
considerations that are relativized to particular
people but that such a practice can be justified at the
critical level by reference to considerations of a
wholly impersonal kind. The loyal agent is dedicated in practice to the welfare of a particular individual or individuals. But that sort of dedication can be
represented at the critical or reflective level as an
ideal worthy of universal admiration. This resolution will not do. If the operational
considerations deployed by the loyal agent were
purely partial, as the resolution suggests, then he could not invoke them in an attempt to justify him? self. Challenged about his favoritism towards a
friend or child, all that they would enable him to
do is make a confession of bias. But the consider? ations deployed by the loyal agent do serve him in a justifying role. It is not merely an admission of
bias, even a bias capable of critical vindication, if
he responds to a challenge by remarking that the
beneficiary is his child or his friend. It is a way, in itself, of legitimating his selective concern.
We are returned then to our paradox. How can
the reason on which a loyal agent acts serve him
163
164 AMERICAN PHILOSOPHICAL QUARTERLY
in two quite different roles? How can it mark his
partiality towards his principal and still give an
impartial cast to his behavior?
I. The Source of the Paradox
We need to develop a more exact statement of
the source of our paradox, for otherwise there is
no hope of identifying what is necessary to resolve
it. In particular we need a more specific account
of the apparently conflicting requirements on the
reason with which a loyal agent acts. Before
offering such an account, however, it may be useful
to provide also a more specific account of a reason
for action.
My notion of a reason for action is abstracted
from the traditional concept of the practical syl?
logism. That syllogism is meant to represent an
agent's deliberation at the operational level, whether explicit or tacit. It begins with considera?
tions in virtue of which a certain state of affairs is
seen as desirable and worthy of realization: these
are the materials assembled in the major premiss. It proceeds in the minor premiss to considerations
that link?at the limit indeed, identify?that state
of affairs with an option which the agent can realize.
And it issues by way of conclusion in the choice
of the option or, more plausibly, in the judgment that the option is fit to be chosen, ought to be
chosen, or whatever.
A reason for action, as I use the notion here, is
the sort of proposition which may appear in the
major premiss of a practical syllogism. It consists
in the ascription of certain motivating properties to
the state of affairs it supports. These may be
intrinsic or relational, universal or particular, gen? eral or specific. All that is required of them is that
once recognized and appreciated they should be
such as to make the preference for the state of
affairs intelligible. The apparently conflicting demands on the reason
with which a loyal agent acts are that in some sense
it must be at once partially orientated and impartially
justifying. My more specific account of those
requirements amounts to a hypothesis. I claim that
the mutually antagonistic demands are, on the one
side, that the reason has to be particularized and, on the other, that it has to be universalizable.
A reason is universalizable if and only if it moti?
vates in virtue of just the non-particular or universal
features of any situation where it applies: that is, in
virtue of the features which can be realized across dif?
ferent places, times, persons, and other particulars.4 The properties of the situation to which the agent is
required to be responsive are all such that, if the rea?
son is a good reason, then the concurrence of those
properties would equally command the response of
any other agent faced with a similar situation.
A reason is particularized, on the other hand, if and
only if, whatever the properties in virtue of which it
motivates, it cannot present those properties for the
agent's consideration without directly referring him
to some particular: for our purposes, to some other
person. A reason directly refers to another person if it fixes him, for example, by name or demonstra?
tive. It refers indirectly to him if it is addressed, say
by definite description, to whoever satisfies or best
satisfies certain properties; it is addressed, not to
Fred, not to this person, but to the most deserving
party around, to the agent's oldest relative, or
whatever.5 Since the particularized reason cannot
present the motivating properties without directly
referring to someone, it cannot be replaced without
motivational loss by a consideration which employs
only indirectly referring expressions. Any such re?
placement would miss out on some of the moti?
vating properties; it would not be motivationally
equivalent to the original consideration.
Where universalizability requires that the moti?
vating content of a reason should be universal,
particularization requires that the conditions under
which the reason motivates an agent?the conditions
under which he is engaged with the motivating content?should involve his being focussed on a
given individual. The requirements of universaliz?
ability and particularization are not strictly incon?
sistent, but they are in tension with one another.
Suppose that a reason is universalizable and motivates
an agent to act suitably towards anyone displaying the universal properties F, G and H. We would
expect the reason to be presentable in a formula
which says that there is an x such that Fx, Gx
and Hx. And such a reason would not be particu? larized, for it would not refer directly to x or to any?
thing else. A directly referring expression might be
needed in the minor premiss of a practical syllogism
THE PARADOX OF LOYALTY 165
where the reason figures as major premiss; the
minor premiss would say that this or that person is an x of the motivating kind. But the expression
would not appear in the reason itself and so the
reason would not be particularized. Because loyalty is considered as a justifiable
ideal or duty, even by loyal agents themselves, the reason with which a loyal agent acts must be a
universalizable one. Were it non-universalizable, then the properties in virtue of which it motivates
would involve a particular person or other entity. And in that case it would be incapable of justifying the behavior to those who have no special relation?
ship with that individual or entity. If the fact that
the beneficiary is Dick is essential to the agent's
being motivated to help Dick, then the reason with
which the agent acts can hardly be invoked in an
attempt to justify his favoritism. In order to serve
in that role the reason must motivate in virtue of
properties whose concurrence would have had the same motivational hold on anyone else, in any other
company or context. In short, it must be univer?
salizable.
Because loyalty is seen as a form of dedication, however, and not just as fidelity to an ideal or the
exercise of a duty, the loyal agent is required to
act on a particularized reason which directly involves his principal. The notion of loyalty is that of a disposition to prefer, not anyone who satisfies certain universal conditions, but this or that contex
tually given person. This requirement will be ful?
filled so far and only so far as the agent's reason
depends for its motivational power?that is, for its
capacity to present the relevant motivating proper? ties?on referring him directly to the principal. That is to say, it will be fulfilled if and only if the reason is particularized.
My hypothesis as to the source of the paradox of loyalty is that it is difficult to identify a reason which is at once particularized and universalizable. For the moment I have nothing more to say in
defense of the hypothesis, but I hope that it may become more plausible in the light of the following two sections. In the next section it will become clear that there is a difficulty about finding reasons that display both of these characteristics. And in the section after that it will become apparent that once we have identified such a reason we are in a
position to resolve the paradox of loyalty.
II. The Materials for Resolving
the Paradox
If the hypothesis is correct then we shall only be able to resolve our paradox if we can identify reasons that are both universalizable and par? ticularized. In this section we review the different sorts of reasons for action in an attempt to find
suitable candidates.
The main distinction among reasons for action
is between agent-relative and agent-neutral consid?
erations.6 Any reason, as we know, ascribes
motivating properties to a state of affairs. With the
agent-relative reason the motivating properties ascribed include some which can be expressed only by pronomial back-reference to the agent who is
to act. With the agent-neutral reason this is not so.
Suppose that Tom is the agent. An agent-relative reason for Tom to prefer a state of affairs might be that it is in his interests; that it will save the
honor of his family; that it will confer some benefit on his dependants; and the like. The crucial,
motivating property in each case is that the agent, whoever he is, is served in a certain way by the state of affairs. It is not that it will benefit Tom's
dependants, for example; after all, the agent might not even be aware that he is Tom. It is that the state of affairs will benefit his, the agent's, depen? dants. The sort of reason on offer essentially involves pronomial back-reference; it is essentially indexical in form.
Agent-neutral reasons, by contrast with agent relative ones, are free of this relativization to the
agent. An agent-neutral reason for Tom to prefer a state of affairs might be that it is just; that it will
please Mary; that it will promote human happiness; and the like. When such a reason argues for action there will have to be an indexical involvement of the agent, as he is the one to be moved.7 But that
involvement will come only in the minor premiss of the deliberation, not in the major.
We want to identify reasons that are at once uni?
versalizable and particularized. We are unlikely to find any in the agent-neutral category. A particular? ized reason is bound to refer directly to some indi?
vidual, say by name or demonstrative; it cannot just refer to whoever satisfies certain properties. How can such a mode of reference be motivationally
166 AMERICAN PHILOSOPHICAL QUARTERLY
necessary in an agent-neutral reason, if the reason
motivates an agent just in virtue of universal features
of the situation where it applies? The necessity of the reference is impossible to explain, if the moti?
vational properties in question do not involve the
particular individual referred to. Thus while "Jack is in trouble" may seem to be particularized and uni?
versalizable, the appearance withers once we see
that, not depending on the reference to Jack for its
motivational power, it is motivationally equivalent to "There is someone in trouble." The particularized consideration can be replaced without motivational
loss by the non-particularized surrogate; the two
formulations serve equally well to direct the agent to the relevant motivating properties.
We must turn to the category of agent-relative reasons if we are to have any hope of success in
our quest. But without one further distinction this
category is too coarse to be of help. The distinction,
ignored in the literature, is between agent-relative reasons that are context-free and agent-relative reasons that are context-bound.
Take an agent-relative reason which bears on the treatment of a person other than the agent; this is
the only sort that will be relevant for our purposes.8 Such a consideration will refer back to the agent,
involving an egocentric description of the person on whom the action bears as his child, his friend, his client, or whatever.9 The difference between
context-free and context-bound reasons relates to
the role which the egocentric description plays in
determining who is to be given special treatment.
The agent-relative reason will be context-free if the
intended referent is that person, whoever it may be, who satisfies?or perhaps best satisfies?the descrip? tion in question. It will be context-bound if the
intended referent is fixed, or partly fixed, by context:
fixed, for example, by whoever is actually present in some demonstrative context, or by whoever is at
the causal origin of the referential expression used.
The context-bound reason refers directly to the per? son; the context-free reason refers only indirectly to
him. Thus the egocentric description plays a dual role
in the context-free reason, serving to determine the
intended beneficiary and helping to explain why he
makes a claim on the agent. It plays only the second
role in the context-bound reason, for the identity of
the beneficiary is fixed with the help of context.
Consider the case of someone who contemplates what he owes a friend. If he thinks in the context
free way, the thoughts which move him will be that there is a friend of his who needs this or that,
who would benefit from one or another sort of
action, and so on. In those thoughts the identity of the particular friend is not relevant; it does not
matter to the agent's motivation. He is targeted on
whoever fits the description of being his friend. That it is this person rather than any other who does so is a matter only of incidental interest.
If the agent thinks in the context-bound way however, then the identity of the friend will be
taken as given independently of his satisfying the
egocentric description. The considerations which
engage the agent will be that this, his friend stands
in need of so and so or stands to benefit by such
and such. Considerations of that kind, as the demonstrative signals, presuppose that the identity of the friend is fixed with the help of context. The
agent is targeted then on that particular person, not on whoever happens to fit the egocentric descrip? tion. The description serves to present the person in a way which helps to make the claim on the
agent salient; he is a friend. It does not play the
further role of determining its own reference.
The contrast between the two sorts of reason comes out nicely in a familiar type of thought experiment. Suppose that Tom is subject to a con?
text-bound consideration for favoring Dick: This, my friend, is in need. Now compare his situation with that which he would be in if everything were
the same except for the absence of Dick. In the actual situation Tom has an agent-relative reason
of a context-bound kind for favoring Dick. In the
imagined situation he hasn't; Dick isn't there to
provide the referent for "this, my friend."
This does not hold of the context-free counter?
part: There is a friend of mine in need. If Tom is
subject to such a reason in the actual situation?a reason to take special account of anyone in Dick's
sort of position?then he may be equally subject to it in the imaginary situation where Dick does
not exist. All that is required for it to apply is that
he has some friend there who is in need. It happens in the imaginary world that Dick does not exist, but the consideration which is expected to move
Tom in the actual world may still be there to move
him in the imaginary.
THE PARADOX OF LOYALTY 167
Agent-relative reasons are particularized only if
they are context-bound, for it is only in that case
that they refer the agent directly to an individual
or individuals. Our search for reasons that are at
once universalizable and particularized must be
concentrated, therefore, on considerations in this
category. Do such considerations simultaneously
display the two characteristics?
If an agent-relative context-bound reason is uni?
versalizable, then it motivates in virtue of just the
non-particular features of any situation where it
applies. So what then are the universal features
made salient for an agent in the consideration "This,
my friend, is in need?" And are they such as can
be presented to the agent only if the reason refers
him directly to the potential beneficiary? At first sight there seem to be just two universal
features involved: that the person is in need and
that he is a friend. But while those features are
non-particular, they cannot be the only motivating ones. If they were, there would be no motivation
ally relevant difference between the consideration
given and the context-free reason expressed by "There is a friend of mine in need." Thus the reason
would not depend on directly referring the agent to the beneficiary for its having motivational power; it would not be a particularized consideration. The
agent might have been equally well moved by the
context-bound or context-free thought: each would
direct him to the same motivational properties. But there is a third, non-particular feature intro?
duced by the context-bound reason. This is
presupposed by the form in which the reason is
given and does not correspond to a predicate used
in the formulation of the reason. It is exemplified
by the formula used, not expressed in it. The feature is that the agent is confronted with
someone who is identified for him by context: some? one who is identified by being, brutely, there. It is this property which distinguishes a situation where
the context-bound reason of friendship is available
from any circumstance where the agent does not have his claimant independently individuated. In
the former situation the agent finds himself faced with the demands of a particular friend. In the latter circumstance he also finds himself confronted
with the demands of friendship, but in abstraction from the individual whom they would benefit.
Because of the three universal features on which
it turns?in particular, because of the third?"This,
my friend, is in need" cannot be replaced without
motivational loss by the consideration expressed in
"There is a friend of mine in need." Does this
mean, more generally, that the reason involved is
particularized as well as universalizable? Unfortu?
nately not. Although it cannot be replaced by "There is a friend of mine in need," the consider?
ation may still be replaceable by some other reason
which fails to refer the agent directly to the potential
beneficiary: in effect, by some other context-free
consideration. We have to consider whether there
is any context-free surrogate which could take the
place of the thought, "This, my friend, is in need," and do so without motivational loss.
Any context-free reason which directs the agent to the fact that the potential beneficiary is contex
tually individuated must register that fact in its
content. As to the example on hand, any prospec? tive surrogate must go something like this: "There is a friend of mine, identified for me by context,
who is in need." The question then is whether the
context-bound reason "This, my friend, is in need" can be replaced without motivational loss by such
a sophisticated context-free consideration. If it can, then it is not particularized: it does not depend on
directly referring the agent to the friend for the
motivational power it has; the agent could be pre? sented with the universal motivating properties
without being directly referred to anyone. But the context-bound reason cannot be replaced
without motivational loss by the sophisticated coun?
terpart. For it cannot be replaced by the counterpart at all. Suppose that the context-bound reason moti? vates an agent, so that he thinks "This, my friend, is in need." If the agent genuinely has that thought, notice that there must be a person there to provide the referent for the demonstrative.10 Consider now
whether the context-bound consideration could be
replaced in the agent's motivation by the context
free surrogate "There is a friend of mine, identified for me by context, who is in need." The answer is
clearly that it could not. If the agent is motivated
by the context-free consideration, and if there really is a person identified for the agent by context, then
things must be such that he is motivated also by the context-bound thought. They must be such that
168 AMERICAN PHILOSOPHICAL QUARTERLY
there is a person of whom he thinks in the
demonstrative mode, or at least in some directly referential manner, and they must be such that he thinks of that person as a friend in need. If the context-bound consideration motivates an agent, then there is no way that it can be displaced in favor of the context-free surrogate; the surrogate
will motivate him only so far as he continues to
be motivated by the original. I conclude that context-bound, agent-relative
reasons satisfy the two requirements with which we have been concerned. They are particularized reasons which have their motivational power only so far as they refer the agent directly to another
person or persons. And yet they are universalizable considerations which motivate the agent just in
virtue of the universal features of any situation where they apply.
If this conclusion is counter-intuitive, that may be because it displaces an assumption which is
widely but uncritically made. The assumption is
that if an agent acts on a universalizable reason, then there must be a universal reason for what he
does which is available in common to him and his
counterparts in relevantly similar situations. But
that assumption is a mere prejudice, as the category of context-bound, agent-relative reasons shows. The counterparts who each act on such a reason act on distinct considerations in their different situ? ations: namely, those which are particularized in favor of their particular principals. And yet the reasons motivate in each case on the basis of the same universal properties. There is no universal reason available in common to those agents, though the reasons which are available motivate on the same universal basis.
III. The Resolution
The first section put forward the hypothesis that the source of the paradox of loyalty is the difficulty of
identifying reasons that are at once particularized and
universalizable; it seemed that such reasons must be
available to motivate the loyal agent, if he can espouse
loyalty as a duty or ideal. In the last section we identi?
fied reasons that have the two marks required, and it
remains now to show that those reasons do indeed serve to resolve the paradox. We are required to show
this, if the original hypothesis is to be vindicated.
There are two things in particular that we need to
show. The first is that there is nothing implausible about the idea of a loyal agent being motivated by a
context-bound agent-relative reason. And the second is that there is nothing implausible about the idea of
that agent's principal being suitably gratified by attentions that are motivated in such a manner.
If we are to show these things we are required in particular to be able to resist the thought that
loyalty presupposes motivation by a non-univer
salizable, particularized reason. Consider two
agents, one of whom is moved by the thought "This,
my friend, is in need," the other by the thought "Dick is in need"; we assume that these expressions are precisely tailored to the moving considerations.
What we have to demonstrate is that the giver and the taker of loyal attentions, the agent and the prin? cipal, can each be comfortable with the fact that the universalizable sort of motivation is that which
holds sway. We deal first with the perspective of
the agent, then with that of the principal. The agent who is moved by the thought "Dick
is in need" makes much of the particular identity of the person in question. The non-particular facts about Dick do not fully account for the claim he
represents. That claim turns essentially on the fact
that he is Dick. His thisness or haecceitas matters. The agent who is moved by the consideration
"This, my friend, is in need" does not make any?
thing of the particular identity of the person before him. He makes something only of the non-par? ticular facts that he is a friend, that he is in need,
and, as we might put it, that he is there.
This second story is surely a more plausible account of what counts with the agent. Loyalty can
hardly require the agent to prize the individual
essence of the principal; on the contrary, it is com?
patible with severe reservations about his particular worth. What it demands is rather susceptibility to
the fact that the principal is, relatively speaking, one of his own; he is one of the family, one of the
gang, one of the network, or whatever. And that
demand fits with the universalizable sort of motiva?
tion, not with the singular kind.
What now of the principal's perspective? Will
the beneficiary of putatively loyal attentions be suit?
ably gratified if the basis on which he receives
those attentions is universalizable? Will the friend
THE PARADOX OF LOYALTY 169
in need recognize loyalty in the agent who is moved
by the context-bound agent-relative reason, and not
by the consideration involving him by name? The ground on which the friend is singled out for
preferred treatment in each case contrasts with that
invoked when an agent acts on the non-particularized consideration "There is a friend of mine in need."
In this case the principal is benefitted for the more
or less incidental reason that he happens to fit the
description which activates the agent's concern. He
is not himself the target of concern; he falls on the
line of favor determined by that target. If the beneficiary is picked out by name or in
the demonstrative manner of the agent-relative, context-bound reason, then it is not incidental in the same way that it is he rather than anyone else
who is benefitted. He can feel in either case that
he matters in his own person, for the consideration is focussed on him in particular, not on anyone
fitting a certain bill; it is particularized in his favor.
But still there is a difference between how the
beneficiary matters in the two cases, and the ques? tion is whether this tells against the suitability of
the universalizable reason.
Take the case where the principal is picked out as "This, my friend." The principal knows that
were the agent to find himself in a world where
he, Dick, was replaced by another individual, then
though he would lack access to the reason for which
he benefits Dick in the actual case, still the agent would be moved by a replica consideration: the
consideration that that other person, his friend, was
in need. Take now the case where the principal is
picked out by name. Here he has no ground for
the same thought, for in the world imagined the
agent might be unmoved by the plight of the other
individual, just because that individual was not
Dick. Is this extra individuation of concern neces?
sary for the enjoyment of loyalty? Or is the univer? salizable individuation enough?
I think that the extra individuation is not required, however flattering and gratifying it might be. The
only ground on which a person can expect loyalty is that he has the good fortune to connect with the
agent; he counts in some sense as one of the agent's own. It would be comic or megalomaniacal of the
person to expect loyalty on the ground of who he
is; no one can reasonably think himself so special.
Everything suggests, then, that the principal can fully
enjoy the fruits of loyalty while recognizing that the
agent is moved by a universalizable consideration. The availability of context-bound, agent-relative
reasons is thus sufficient to resolve the paradox of
loyalty. We can coherently identify it as an ideal or duty for an individual that he be loyal to these, his children or those, his friends. The reason offered
to him thereby is universalizable, for it is allowed that for any agent and any corresponding claimants, that agent too should be loyal to those, his princi?
pals.11 Yet the reason offered is also, as required, a particularized consideration. It is a reason to care
for his principals which refers him to them directly and not just so far as they satisfy certain identifying descriptions; it is a consideration which makes them
matter in their own right, and not as opportunities for furthering some independent cause.
One further word may be useful. Although the
possibility of loyalty turns on the availability of
context-bound, agent-relative reasons, there is a
role too for context-free considerations. This is
worth noting, since otherwise it may be a source
of confusion. The primary demands of loyalty relate to
principals already recognized or owned by the
agent as children, friends and the like. This
recognition or owning is a social fact that usually comes to obtain beneath the level of agent choice.
But sometimes it can happen that an agent fails to own someone as a principal of a certain kind, even when it is appropriate for him to do so. In
such cases there are secondary demands of loyalty and these are naturally expressed by context-free
considerations.
The sort of demand in question is that the
loyal agent ought to own as a child or friend or
whatever any individual who meets suitable
equirements. Although it is a legitimate obli?
gation of loyalty, such a context-free demand
does not compromise the claim that loyalty re?
quires context-bound reasons. This kind of demand is an obligation to recognize or own
certain people in the role of principals. The primary sort of demand is an obligation consequent on the fact of recognition: not an obligation to own, but an obligation of
owning.
170 AMERICAN PHILOSOPHICAL QUARTERLY
Conclusion
There are at least two sorts of misgiving that may be raised about this resolution of our paradox and in
conclusion I wish to indicate why I do not think that
they have to trouble us. They are both misgivings about the status which the resolution gives to agent relative, context-bound reasons. One turns on the
fact that those considerations are agent-relative, the
other on the fact that they are context-bound. The
first is a misgiving in moral theory, the second a
misgiving in philosophy of mind.
The misgiving about agent-relativity is that it of?
fends against standard consequentialist assumptions. Those assumptions are that goodness is evaluator
neutral; that the Tightness of an option is a function
of the goodness it promises; and that moral agents
ought to train their sights therefore on the pro? motion, in the manner determined by that function, of goodness.12 You clearly reject that conclusion
if you endorse the idea that agents ought to act, even
act in restricted circumstances, on considerations
of an agent-relative kind.
If Tightness is a function of the impersonal good, then the consequentialist must certainly wish for the
promotion, in the appropriate manner, ofthat good. But that does not mean that the only sort of con
ideration he can allow an agent to be moved by is
the agent-neutral consideration of what maximizes
the prospect of good. It may be that by acting on
another type of consideration, at least within
certain limits, the agent will actually succeed
in better promoting the good. This will be so in
particular if acting on such non-consequentialist reasons produces a significant benefit which is
compromised by acting on consequentialist considerations.
Agent-relative context-bound reasons are certainly
non-consequentialist considerations. But if my
argument in this paper is sound, then agents pro? duce the benefits associated with loyalty if and only if they act on such reasons in appropriate circum? stances. On the assumption, then, that the benefits
associated with loyalty are important, it may well
be that the consequentialist should endorse such
a course of behavior. It may well be that he
ought to counsel a restriction of consequentialist
reasoning in favor of a dispensation where
agent-relative reasons have an important place.13
The misgiving about the context-bounded
aspect of the reasons I associate with loyalty comes up in the philosophy of mind. It is a common
dogma of psychologists, philosophical and scien?
tific, that properly psychological states must be
fixed by how things are within the boundary
assigned to subjects: this may be the boundary of the body, the central nervous system, the res
cogitans, or whatever.14 The dogma would mean that the state of being responsive to a
context-bound consideration cannot be properly
psychological. It cannot count among the springs of a person's action.
There are a number of possible views about what
the true spring of action must be when it is appro?
priate to ascribe a context-bound reason to an agent. Under all of these views however, a question must
be raised about how satisfactory our resolution of
the paradox of loyalty is. And that is why the
dogma creates a misgiving about the resolution.
We need surely not be troubled by this misgiving. The reason is that there is an alternative view of the mind available which fails to support the dogma; I have committed myself elsewhere to it.15 According to that view the mental states which explain our
actions may have contents that are context-bound, even though the states explain actions in virtue of
those contents. The springs of our actions may be tied
significantly to our surroundings. If one adopts this
broad view of mental states, then there is no difficulty with the fact that my resolution of the paradox of
loyalty requires people to be moved by context
bound reasons.16 We do not have to defend the view
in order to gain relief from the difficulty; it is suf?
ficient for our purposes that context-bound reasons
have not been shown to be psychologically impossible or suspect. But even if defense were necessary, we
would not be without resources. It will be defense
enough for many that the broad view of mind is
required for the possibility of loyalty.17
Research School of Social Sciences
Australian National University Received June 11, 1987
THE PARADOX OF LOYALTY 171
NOTES
1. See Derek Parfit, Reasons and Persons (Oxford: Oxford University Press, 1984). See also Philip Pettit and Robert Goodin, "The Possibility of Special Duties" in Canadian Journal of Philosophy, vol. 16 (1986), pp. 651-676. For more general, related
considerations see Michael Ignatieff, The Needs of Strangers (Harmondsworth: Penguin Books, 1986).
2. See Peter Railton, "Alienation, Consequentialism, and the Demands of Morality" in Philosophy and Public Affairs, vol. 13
(1984), pp. 134-171.
3. The distinction is invoked in Parfit, op. cit., in Railton, op. cit., and, perhaps most notably, in R. M. Hare, Moral Thinking
(Oxford: Oxford University Press, 1982). It is also deployed in Philip Pettit and Geoffrey Brennan, "Restrictive Consequentialism" in Australasian Journal of Philosophy, vol. 64 (1986). We make use of it in the conclusion below.
4. See for example R. M. Hare, op. cit. and Wlodzimierz Rabinowicz, Universalizability (Dordrecht: D. Reidel, 197?). See too
my paper "Universalizability without Utilitarianism" in Mind, vol. 96 (1987), pp. 74-82.
5. For a careful account of the sort of distinction I am drawing?the difference between descriptionally and nirpctly referring
expressions?see Nathan Salmon, Reference and Essence (Oxford: Blackwells, 1982), Chapter 1.
6. See Derek Parfit "Prudence, Morality and the Prisoner's Dilemma" in Proceedings of the British Academy, vol. 65 (1979),
pp. 555 ff.; Thomas Nagel "The Limits of Objectivity" in Tanner Lectures on Human Values, Vol. 1 ed. by S. M. McMurrin
(Cambridge: Cambridge University Press, 1980), pp. 75-139; and Amartya Sen "Rights and Agency" in Philosophy and Public
Affairs, vol. 11 (1982), pp. 03-39. See too Pettit and Goodin, op. cit..
7. See John Perry "The Problem of the Essential Indexical" in Nous, vol. 13 (1979), pp. 03-21.
8. Even with reasons which bear on an agent's treatment of himself, there is a parallel distinction to that introduced here. There
is a difference between having a reason to be true to that particular self one is and having a reason to be true to whoever one
happens to be.
9. Robert Pargetter has made the point to me that it is likely that the description required to express all the relevant aspects of
the potential beneficiary will be much more personalized than those given in these examples.
10. On related matters see Gareth Evans, The Varieties of Reference (Oxford: Oxford University Press, 1982).
11. This is to support the line taken in John Cottingham "Partiality, Favouritism and Morality" in The Philosophohical Quarterly, vol.
36 (1986), p. 359. It is to reject the claim of Andrew Oldenquist in "Loyalties" in The Journal of Philosophy, vol. 79 (1982),
pp. 185-86. I suspect that Oldenquist is swayed by the assumption mentioned at the end of section 2.
12. See Philip Pettit and Geoffrey Brennan, "Restrictive Consequentialism" in Australasian Journal of Philosophy, vol. 64 (1986),
p. 438.
13. See Pettit and Brennan "Restrictive Consequentialism" ibid. See also my papers "Social Holism and Moral Theory" in
Proceedings of the Aristotelian Society, vol. 86 (1985-86), pp. 173-198, and "The Consequentialist Can Recognise Rights" in
Philosophical Quarterly, forthcoming. One of the challenges to which this restrictive version of consequentialism is addressed is
the requirement that a moral theory should not undermine our spontaneous motives of love, friendship and the like. Thus the
present paper fits with that program. The challenge is best posed in Michael Stocker, "The Schizophrenia of Modern Ethical
Theories" in Journal of Philosophy, vol. 73 (1976), pp. 453-466.
14. For an overview of the issue see the introduction in Subject, Thought and Context, ed. by Philip Pettit and John McDowell
(Oxford: Oxford University Press, 1986).
15. See my "Broad-minded Explanation and Psychology" in Subject, Thought and Context, ibid.. See also my "A Priori Principles and Action-explanation" in Analysis, vol. 45 (1986), pp. 39-45.
16. In "Social Holism and Moral Theory" op. cit.. I argued that the acceptance of such a view of the mind facilitates the recognition of agent-relative reasons. It was as a result of discussion of that paper at the Aristotelian Society that I came to see that, more
specifically, it facilitates the recognition of agent-relative, context-bound reasons. I am particularly grateful for comments by Mark Sainsbury and Richard Sorabji.
17. I am grateful for written comments on earlier drafts from Ted Bond, Jeremy Butterfield, Frank Jackson, Mark Sainsbury and
Michael Stocker. The paper was written during my tenure of an Overseas Fellowship at Churchill College, Cambridge and I am
also grateful to the College for the use of its facilities.