Law and Philosophy (2015) 34: 241256

DOI 10.1007/s10982-014-9222-9

 Springer Science+Business Media Dordrecht 2014

PHILLIP MONTAGUE

SPECIFICATION AND MORAL RIGHTS

(Accepted 3 November 2014)

ABSTRACT. In this paper, I offer objections to an approach to formulating
principles referring to moral rights that has come to known as specification.
These objections (which are directed at all forms of specification, including one
recently defended by Hallie Liberto) focus on rights-principles in their role as
premises of inferences to conclusions regarding the moral rights of individuals in
particular situations. I argue on practical grounds that specified principles have no
useful role to play in such inferences, and on theoretical grounds that the specificationist position is self-defeating. This latter argument also suggests an interpretation of rights principles that avoids the objections to which specification is
vulnerable.

Certain philosophically interesting and important questions about

principles that refer to moral rights are, at least broadly speaking,
logical in character. These questions focus on rights-principles in
their role as premises of inferences to conclusions regarding the
rights of individuals in particular situations. Consider, for example,
the principle that everyone has a right to life, and hence a right to
not be killed. The principle appears to entail that each particular
homicide is a moral rights-violation, and is therefore morally
impermissible. Yet certain homicides certain self-defensive homicides for example seem to be morally permissible. Assuming that
no particular action can possibly be both permissible and impermissible, there is a consistency problem here one that has parallels
for all principles that refer to moral rights. These problems arise

242

PHILLIP MONTAGUE

from a certain interpretation of the logical form and behavior of

rights-principles.1
My purpose in this essay is to examine an alternative interpretation of rights-principles as a possible approach to avoiding the
aforementioned consistency problems. This interpretation is commonly referred to as specification. It was brought into philosophical
prominence by Judith Thomson, who distinguished two types of
specification moral and factual.2 The two differ importantly
from each other; each is problematic in its own way, although they
also have problems in common. In the interest of thoroughness,
both will be discussed here, but with primary emphasis placed on
factual specification. It is much more interesting than moral specification which, when clearly formulated and closely examined, is
revealed to be something of a non-starter.
As its name suggests, factual specification interprets rights-principles as highly specific as incorporating all of the morally significant factual features of the situations to which the principles are
applied.3 I will argue that this approach to the consistency problems
alluded to above doesnt succeed. Its principles are of little or no
practical use and, more importantly, it is self-defeating. That is, it
could be made to work only by employing a particular kind of very
general rights-principles principles whose existence renders specificationist principles entirely superfluous.4 In developing this argument, I will highlight Hallie Libertos recent defense of specification.5
Libertos account incorporates features that are essential components
of the specificationist position, and that are often overlooked in other
discussions of specified principles.
1
I am not assuming here that the rights-principles in question are foundational. They might, in fact,
be conclusions from more basic premises. What I am assuming is that inferences to conclusions about
the rights of individuals in particular situations must include principles that refer to rights as premises.
2
Judith Jarvis Thomson, Self-Defense and Rights, The Lindley Lecture (Lawrence Kansas: University
of Kansas Press, 1977).
3
Note that the difference between specificationists and generalists isnt that they arrive at different
conclusions about particular cases; it is rather that specificationists interpret rights-principles so that only
one such principle can apply to any given situation. Generalists, according to whose views multiple
principles can apply to particular situations, take very different approaches to avoiding inconsistencies.
4
I discuss these general principles at greater length in When Rights are Permissibly Infringed,
Philosophical Studies 53 (1988), pp. 347366). However, this latter discussion doesnt include the extended
critique of specification that I present here.
5
Hallie Liberto, The Moral Specification of Rights: A Restricted Account, Law and Philosophy, 33
(2014), pp. 175206. Liberto refers to her account as moral specification but, as will become evident
later, it actually corresponds to the position that I am calling factual specification.

SPECIFICATION AND MORAL RIGHTS

243

The contexts in which specification is most commonly employed, hinted

at in the preceding remarks, are exemplified by two scenarios depicted
by Thomson. In one, a (presumably culpable) Aggressor will kill a
(presumably innocent) Victim with his tank, unless Victim uses her antitank gun to kill Aggressor in self-defense. In the other, a person can
prevent a protein-deficient child from starving to death if and only if she
takes someones steak without the owners permission. It is reasonable
to infer from the descriptions of these cases that Victim is morally
permitted to kill Aggressor, and that the person in a position to save the
childs life by taking the steak (call her Rescuer) is permitted to do so.
Although reasonable, however, these inferences lead to inconsistencies
under certain assumptions; and specification is viewed by its proponents
as the proper way in which to avoid these inconsistencies. Determining
whether specification can indeed do the trick requires precise formulations of the problems surrounding cases like Thomsons, and of specification itself. While it might seem that both requirements are routinely
satisfied in discussions of specification, quite the contrary is true.
For example, consider how Thomson characterizes what a specificationist would say about her self-defense example:
You only think theres a problem here because you think that Aggressor has a
right to life entails Aggressor has a right to not be killed. But it doesnt. We all do
have a right to life, but that right to life is a more complicated business than it at
first may appear to be. In particular, having a right to life doesnt include having a
right to not be killed. Indeed, nobody has a right to not be killed: all you have is
and here there are two ways in which the speaker may go on. I will call the first
moral specification: all you have is the right to not be wrongly, unjustly
killed. I will call the second factual specification: all you have is the right to
not be killed if you are not in process of trying to kill a person, where that person
has every reason to believe he can preserve his life only by killing you.6

Thomsons examples of specified principles obviously dont explain the

concept of specification; and her examples give only rough indications of
6
Thomson, p. 5. Although in this discussion Thomson argues against both moral and factual
specification, she subsequently espouses two other forms of specification that differ from each other,
and neither of which corresponds either to moral or to factual specification. On the one hand, she
claims that rights are always rights other things being equal (Self-Defense, Philosophy and Public Affairs
20 (1991), pp. 283310). On the other hand, however, she applies specification to the normative
constraints implied by rights rather than to the rights themselves, claiming that the constraints are
qualified by ceteris paribus conditions (Thomson, The Realm of Rights (Cambridge, Mass.: Harvard
University Press, 1990), pp. 23.

244

PHILLIP MONTAGUE

how to understand moral and factual specification. Based on what

Thomson says, it appears that factually specified rights-principles refer to
particular factual features of actions that are relevant to the possession of
particular rights, while morally specified principles treat only certain
moral considerations as directly relevant to the possession of rights.7
In addition to being rather sketchy, Thomsons remarks are also
mistaken in certain respects.
She seems first of all to be suggesting that a factual specificationist
would deny that Aggressor has a right to not be killed on the ground
that all you have is the right to not be killed if you are not. But
this latter principle doesnt imply that Aggressor has no right to not
be killed. To derive this latter conclusion, Thomson would need
something like all you have is the right to not be killed only if. Yet
this latter principle would fail to serve a central purpose of rightsprinciples, which is to provide bases for inferring that people do
possess particular rights for inferring, for example, that Victim has
a right to not be killed. Parallel difficulties are associated with what
Thomson says about moral specification if the principle you have a
right to not be unjustly killed is equated either with you have a
right to not be killed if killing you would be unjust or with you
have a right to not be killed only if killing you would be unjust.
The preceding remarks presuppose that the principles you have a
right to not be killed if and you have a right to not be killed only
if are ordinary universally quantified conditionals (universalized
conditionals for short), and therefore entail ordinary conditionals. The
antecedents of the entailed conditionals of one principle and the consequents of the conditionals entailed by the other have the form P has
a right to not be killed. Yet according to Thomson, specificationists
claim that nobody has a right to not be killed, which implies that P has

According to John Oberdiek,

I do not believe this distinction [between moral and factual specification] warrants significant discussion,
as any plausible conception of moral specificationism will swallow factual specificationism, thereby
overriding the distinction (Specifying Rights Out of Necessity, Oxford Journal of Legal Studies 28 (2008),
pp. 127146, at p. 128, n3).
If Oberdiek is right about this, then my critique of factual specification applies as well to moral
specification.

SPECIFICATION AND MORAL RIGHTS

245

a right to not be killed is false for all values of P. The antecedents of the
conditionals entailed by one of the suggested specificationist principles,
and the consequents of the conditionals entailed by the other, would
therefore be false; and neither principle would be of much use in
dealing with Thomsons self-defense scenario. If nobody has a right to
not be killed, then Victim doesnt although she clearly does; and no
additional principle specified or not would be needed in order to
conclude that Aggressor doesnt have this right.
So what should we make of Thomsons claim that, according to
specificationists, nobody has a right to not be killed? I suspect that
Thomson has confused Nobody has a right to not be killed i.e., It
isnt the case that anyone has a right to not be killed with the
negation of Everyone has a right to not be killed. The idea would
then be that specificationists are denying that, given any x (or given
any x that is of a certain kind), x has a right to not be killed. In any case,
I will assume that this is how specification should be interpreted, and
that its principles can therefore be interpreted as ordinary universalized conditionals or, rather, as ordinary universalized biconditionals.
The need for biconditionals becomes evident on carefully considering the problems that specification is supposed to solve, and the
contexts in which those problems arise. To cope with Thomsons
self-defense case, specificationists need a principle on which to base
the conclusion that Aggressor has no right to not be killed, bearing in
mind that a principled basis is also required for concluding that
Victim does have a right to not be killed. The natural way in which
both moral and factual specificationists might attempt to achieve
these ends is by interpreting their positions as delivering principles
that provide conditions that are necessary and sufficient for the
possession of particular rights.
The idea that specificationist principles should be interpreted in
this way tends to be overlooked in discussions of specification, but is
implicitly endorsed by Liberto. She states that:
According to Moral Specificationism, the very description of any right includes all of
its exceptions, those morally justifiable circumstances in which it is permissible for
someone to act in a way that, ordinarily, would appear to be at odds with the right.8

As it applies to the right to not be killed, Libertos characterization of

specification implies that the relevant principle would be something like
8

Liberto, p. 176.

246

PHILLIP MONTAGUE

A person has a right to not be killed if and only if killing the person isnt
self-defense against a culpable aggressor, and isnt, and.9 Libertos
reference to all exceptions, those morally justifiable circumstances
in her characterization of what she calls moral specificationism explains
why I noted earlier that her position actually corresponds to factual
specificationism as the latter is being interpreted here.
I will have more to say about this biconditionals idea as it applies
to factual specification a bit later in the discussion. First, however, Ill
look briefly at moral specification, interpreted so that it yields the
following principle: x has a right to not be killed if and only if killing
x is unjust. The problem for moral specificationists is that of justifying the use of these principles in reasoning to conclusions about
the possession or nonpossession of rights in cases like that of Victim
and Aggressor.
Thus, suppose that we reason in this way: x has a right to not be
killed if and only if killing x would be unjust; killing Aggressor
wouldnt be unjust; hence, Aggressor has no right to not be killed.
But how would we show that killing Aggressor wouldnt be unjust?
Presumably, by means of an argument that includes the premise that
killing Aggressor wouldnt violate his right to not be killed a
proposition that would be inferred from the proposition that
Aggressor has no right to not be killed. The entire inference is
therefore question-begging as would be the case for all similar
inferences from morally specified rights-principles. Hence, such
principles cant serve a purpose that anyone who believes in rightsprinciples specificationists in particular must assume they do

9
Unfortunately, Libertos examples of specificationist principles dont seem to fit within either of
Thomsons categories. Thus, she counts the following (which pertains to Thomsons case of the
starving child) as referring to a morally specified right:

You have a right that no one breaks into your freezer unjustifiably, and to justifiably break into you
freezer means, among other things, that one provides compensation (Liberto, p. 180).
This proposition obviously doesnt include all the factual exceptions to the relevant right, but neither
does it correspond to the simple form of Thomsons moral specification.

SPECIFICATION AND MORAL RIGHTS

247

serve in reasoning about individual rights.10 This result provides

ample justification for setting moral specification aside and focusing
on factual specification. Let us now look more closely at the problem
that specification is supposed to avoid.
II

As it arises in connection with her self-defense case, Thomson states

the problem in this way:
I think that most people would say that it is morally permissible for Victim to use
that anti-tank gun: surely it is permissible to kill a man if that is the only way in
which you can prevent him from killing you.
On the other hand, one of the things we are firmly wedded to is the belief
that human beings have a right to life, and this presumably includes the right
to not be killed. Aggressor is a human being; so he, like the rest of us, has a
right to life, and presumably, therefore, the right to not be killed. So how can
Victim kill him? Precisely why is it permissible for Victim to use that anti-tank
gun on Aggressor?11

10

This objection, which focuses on the role played by specified principles in inferences, is reminiscent of (although distinct from) the circularity objection to moral specification that Thomson raises.
The latter, which concerns explanations rather than inferences, is dismissed by Liberto, largely on the
basis of a reply to the objection presented by William Parent and William Prior (William Parent and
William Prior, Thomson on the Moral Specification of Rights, Philosophy and Phenomenological Research
56 (1996), pp. 837845). As raised by Thomson and criticized by Parent and Prior, the objection is that
explanations of the possession (or non-possession) of rights that rely on morally specified principles are
bound to be circular. The crux of Parent and Priors reply (an analogue of which might be regarded as
undermining my question-begging objection to moral specification) consists in questioning the
explanatory value of rights-principles specified or not:
But does the appeal to moral rights whose content is not specified explain what we may or may not do?
In the Aggressor-Victim case one might at first think that Victims right [to] not be killed explains
why Aggressor may not run him down with his tank. But it does not. The reason Aggressor may not
proceed is that Victim has done nothing whatever that would justify such an assault and there is
nothing else in the circumstances that could justify it. Sometimes, however, killing a human being is
permissible. In every case the moral work of explanation will be borne by the goodness of the ends
being sought and the appropriateness of the means chosen to realize those ends (Parent and Prior,
p. 844).
But Parent and Priors first explanation of why Aggressor isnt permitted to kill Victim (Victim has
done nothing whatever that would justify) presupposes that Aggressors assault needs justification.
And this presupposition would in turn presuppose that Victim has a right to not be killed. Hence, Parent
and Priors own explanation implicitly appeals to a moral right contrary to their claim that such
appeals cant explain what we may or may not do. Assuming that Aggressor has done something that
justifies Victims killing him, his actions must affect whether he has a right to not be killed. Hence, if
Parent and Priors explanation were modified so that it applied to Victims killing Aggressor, the result
would again implicitly appeal to a moral right. Moreover, this explanation is incompatible with Parent
and Priors second, largely consequentialist, explanation of why Aggressors killing Victim would be
impermissible.
11
Self-Defense and Rights, p. 3.

248

PHILLIP MONTAGUE

However, while Thomsons characterization of the problem in her

case is correct as far as it goes, it doesnt go nearly far enough.
We can infer from Thomsons remarks that she views these two
propositions as conflicting with each other:
(1) Victim is permitted to kill Aggressor; and
(2) Aggressor has a right to not be killed.

And these propositions are logically incompatible given the

plausible assumption that, if x has a right to not be treated in
some way, then others arent permitted to treat x in that way.
Less obvious is Thomsons implicit assumption that (2) is deducible
from
(3) All human beings have a right to not be killed and Aggressor is a
human being.

This last assumption presupposes a certain view of the logical form

of rights-principles, namely, that they are universalized conditionals.
Given this presupposition, the nonspecified principle contained in (3)
would have this form: for all x, if x is a human being, then x has a
right to not be killed.12
Factually specified rights-principles could also be formulated as
universalized conditionals but, for reasons given above, they must
instead be interpreted as universalized biconditionals. Accordingly
(and continuing with the idea that specified rights-principles refer to

12
So formulated, the right referred to in (3) seems to belong to the category of general rights
discussed by John Oberdiek. He states that

On this view, for example, a persons possession of a right to life may entail a moral protection against
being intentionally killed that remains standing in all cases (Specifying Rights Out of Necessity,
p. 127).
But Oberdiek also claims that
Defining a right, according to the general conception implies nothing whatever about the normative
effect of the rightabout whether the right will, at the end of the day, prohibit what it stands against
(Specifying Rights Out of Necessity, p. 128).
Here Oberdiek appears to be interpreting general rights not on the model suggested above for the right
referred to in (3), but rather as something like a prima facie right. I defend a prima facie interpretation
of rights in the papers referred to in note 4, although I now think that referring to them as presumptive
rights is preferable. These are the rights referred to in the presumptive principles that I discuss below.
This interpretation of rights-principles provides a way in which to avoid contradictions without
employing specified principles, and that requires abandoning the deductive model of principle-based
inferences.

SPECIFICATION AND MORAL RIGHTS

249

exceptions to rights), these principles would look something like this:

for all x, x has right R if and only if x isnt A, and X isnt B, and.
Like inferences from nonspecified principles such as the one contained in (3), inferences from specified principles would be deductive.
I emphasize this last point because I believe that a tacit commitment
to the idea that principle-based inferences are deductive is a principal
source of the idea that rights-principles must have all their exceptions
included which leads to the formulation of these principles as
universalized biconditionals.
One result of interpreting specified rights-principles in this way is
that no more than one such principle can apply to any given situation.
Specificationists would thereby avoid the contradictions that can
apparently be generated in situations to which multiple nonspecified
principles apply. What specificationists of all varieties must do is make
a plausible case for the proposition that, for every right, there is a
principle that contains conditions that are necessary and sufficient for
the possession of that right. Specificationists dont do this, at least
partly because they overlook the need to interpret their principles as
biconditionals. Although Liberto seems implicitly to acknowledge this
need, she too fails to argue that all rights-principles or any rightsprinciples, for that matter can be formulated so that they include all
exceptions to the rights to which they refer. Liberto does, however,
propose an explanation of how to derive these conditions.
Liberto applies this explanation to her own Restricted Account of
specification, but I believe that it should be embraced by all factual
specificationists:
The Restricted Account holds that if the role of rights within a broader ethical
theory is also what justifies the exception to a particular right, then that exception
should be built into the right itself; however, if the exception to the right is
justified by a moral consideration that is unrelated to the role that rights play
within the ethical theory, then the exception is an infringement of the right.13

I say that all factual specificationists should accept what Liberto says
here about her Restricted Account because I dont see how
exceptions could be genuinely justified if they werent derivable from
ethical theory. How such derivations might work can be seen by
returning to Thomsons self-defense case.
13

Liberto, pp. 194195.

250

PHILLIP MONTAGUE

Although we focused previously on the right to not be killed, the

more basic right is the right to life. Hence, the relevant principle
would be: x has a right to life if and only if; and this principle must
provide bases for concluding both that Aggressor has no right to life
and that Victim does have a right to life and, ultimately, bases for
concluding that Aggressor isnt permitted to kill Victim but Victim is
permitted to kill Aggressor. Whether Victim is permitted to kill
Aggressor, however, depends on the general conditions under which
self-defensive homicide is permissible. The precise nature of these
conditions is debated by self-defense theorists, none of whom (at
least none that I know of) rest their positions on rights-principles that
purport to provide necessary and sufficient conditions for possessing
the right to life.14
It is worth bearing in mind, moreover, that because the requisite
principle is supposed to provide such conditions, it would be an
essential component of inferences in situations where lives are at
stake, but where the lives arent necessarily those of culpable
aggressors and their intended victims. These situations might
involve, for example, self-defense against innocent aggressors, or selfdefensive actions that threaten the lives of innocent bystanders, or
self-preservation (as in some of the famous runaway trolley cases),
or capital punishment.15 According to specification, then, determining the permissibility of killing culpable aggressors in self-defense
would require a principle that also provides a basis for determining
whether executing murderers is permissible, or whether killing
innocent people is permissible in the many and varied situations in
which innocent lives are at stake. It seems to me incumbent on
Liberto and other proponents of specification to provide at least the
first approximation of a rights-principle that refers to the right to life,
14
See, for example, Jeff McMahan, The Basis of Moral Liability to Defensive Killing, Philosophical
Issues 15 (2005), pp. 386405, and Suzanne Uniacke, Permissible Killing (Cambridge: Cambridge University Press, 1994). Thomson does propose a principle that appears to state a condition that is sufficient
for possessing a right to not be killed (Self-Defense, p. 300). However, nowhere does she refer to
principles that contain conditions that are necessary and sufficient for possessing particular rights, even
though her arguments here and (as was pointed out in the text) in her earlier discussion require such
principles.
Of course, this hardly counts as an objection to specification, but it does raise questions about the
usefulness of specified principles in moral theorizing.
15
If capital punishment is claimed to be permissible, then arguing that it is would require a specified
right-to-life principle that includes an exception that applies to individuals who engage in especially
serious wrongdoing. And if capital punishment is claimed to be impermissible, then showing that it is
would require a specified principle according to which even those who commit the most egregiously
wrongful actions have a right to not be killed.

SPECIFICATION AND MORAL RIGHTS

251

and that can yield conclusions regarding whether individuals possess

that right in all of the enormously varied situations in which questions about the permissibility of killing people arise.
Parallel remarks apply to other rights, such as the property right
that is at issue in Thomsons example of the protein-deficient child.
The problems associated with appealing to specification in this area
are clearly illustrated by Libertos application of specification in her
discussion of property rights. Her view implies that inferring that,
say, Rescuers taking the steak is permissible requires a principle
(derived from broader ethical theory) that includes all the exceptions
to property rights, and that is therefore applicable to situations that
are unlike that depicted by Thomson. Yet some of Libertos examples paint a very different picture of the nature of inferences to
conclusions about individual rights.
Liberto describes a situation (Case 1) in which she imagines
herself using person Bs phone without his permission to summon
emergency assistance for person A. She then says this:
Imagine that a version of ethical pluralism is true such that autonomy and choice
and true respect for persons requires a strict set of rights, but that human welfare
needs to be promoted as best as possible. Now, assuming all of this, let me explain
what takes place [in the described situation] given the Restricted Account.
Person Bs right against others using his phone uninvited plays the role in my
ethical theory of protecting person Bs choice and autonomy. In Case 1, where I
use person Bs phone, I am justified not by considerations of Bs choice and
autonomy, but by considerations of As welfare.16

In the described circumstances, using Bs phone uninvited is surely

permissible. Yet Libertos explanation of why this is so is open to
question and this on grounds that are unrelated to the distinctive
features of her Restricted Account.
Note first of all that nowhere in Libertos analysis of Case 1 does a
specified rights-principle that contains all the exceptions to respect
for property rights come into play. The only indication of an
exception is Libertos reference to considerations of welfare. Perhaps she believes that, given an appropriate conception of rights,
considerations of choice, autonomy, and welfare could be bundled
together into a specified principle that is derivable from some (pre16

Liberto, p. 200.

252

PHILLIP MONTAGUE

sumably pluralistic) ethical theory, and that entails that using the
phone is permissible. With all the exceptions to property rights included, the principle would also apply to the protein-deficient child
case and, indeed, to all cases in which property rights are at stake.
But Liberto seems quite comfortable in concluding that using Bs
phone is permissible without appealing to any principle that includes
all the exceptions to property rights just as she seems comfortable
in concluding that Victim is permitted to kill Aggressor without
relying on a principle that includes all exceptions to the right to life.
And Libertos inferences from the factual features of these cases
are quite sound. The fact is that drawing conclusions about the rights
of individuals in particular situations requires identifying the morally
significant features of those particular situations, and without regard
for how other, importantly different, situations should be analyzed.
In Thomsons self-defense case, for example, the relevant assumptions are that Aggressor will kill Victim if and only if she doesnt kill
him in self-defense, that Aggressor is culpable, and that no innocent
bystanders are at risk. We conclude correctly that Victim is
permitted to kill Aggressor on the basis of these assumptions, and
without the aid of a principle that applies to, say, innocent aggressor
or runaway trolley cases. Similarly, we correctly conclude that
Rescuer is permitted to take the steak by accounting for the morally
significant features of that case not by relying on a principle that is
applicable to situations in which property rights seem to conflict
with other moral considerations.
III

The preceding discussion provides reasons for doubting that people

ever rely on rights-principles, with all exceptions built in, when they
reason to conclusions about individual rights. Ill now shift focus
from this practical realm to the theoretical, and explain why the
specificationist position is self-defeating.
All moral rights-principles implicitly affirm that individuals have
moral discretion to make decisions within particular areas of activity,
and that others are morally prohibited from arrogating those decisions
to themselves. So, for example, if one has a right to life, then it is up to
her not others to make certain decisions regarding the course of her
life; and if one has property rights, then it is up to him not others to

SPECIFICATION AND MORAL RIGHTS

253

determine the disposition of his belongings. However, there are conditions under which people dont have discretion to make these
determinations conditions under which others are permitted to make
them. In Thomsons cases, for example, Victim and Rescuer are permitted to make certain decisions that affect Aggressors life and the
steak-owners property respectively decisions that would normally
be Aggressors and the steak-owners to make.
Constructing a specified principle that refers to a particular right R
would require identifying the conditions that are necessary and
sufficient for individuals to have discretion to make decisions in the
area of activity associated with R, and under which others are permitted to arrogate such decisions to themselves. Now, the sorts of
decisions that can be made within the area of activity associated with
a particular right are enormously varied as was pointed out earlier
in noting the widely differing situations in which rights to life or to
property are at stake. If a specified rights-principle contained conditions that are in general necessary and sufficient for possessing
some right, then it would include conditions that are necessary and
sufficient for possessing that right in all of the situations in which
appealing to the right is appropriate.
As we have seen, specification is a response to problems that can arise
in situations like those depicted in Thomsons two cases. While it might
be natural to characterize these problems as arising from conflicts between rights and permissions, this characterization would be disputed by
specificationists, according to whom such conflicts are only apparent.
Yet there is no denying that cases like Thomsons are problematic precisely because they do involve some sort of conflict. In these cases,
considerations that are relevant to the moral status of individual actions
point in opposing directions or (more perspicuously, I think) they
correspond to opposing moral presumptions. Thus, there is a presumption that Victim is morally permitted to preserve her life, and also a
presumption that she isnt permitted to commit homicide presumptions that conflict because Victim can preserve her own life only by
killing Aggressor. And there is a presumption that Rescuers saving the
child is permissible, as well as a presumption that taking anothers
property is impermissible presumptions that conflict because Rescuer
can save the child only by taking the steak. Reconciling such opposing

254

PHILLIP MONTAGUE

moral presumptions presumptions that are expressed by presumptive

principles is a central motivation for the specificationist position.
Although this reconciliation process is aimed at constructing specified principles, its raw materials are presumptive principles principles
that are themselves nonspecified. Assuming, moreover, that the reconciliation process must be rational, presumptive principles would
function as premises in inferences whose conclusions are specified
principles; and such inferences would clearly be nondeductive. The
idea, then, is that inferring that some person P possesses right R
in situation S (in which appealing to R is appropriate) is a two-step
process. First, a specified principle referring to R is nondeductively
inferred from the presumptive principles that refer to the considerations relevant to the possession of R in all situations in which
appealing to R is appropriate. Then the conclusion that P possesses (or
doesnt possess) R in S is deduced from the specified principle.17
If this process could be made to serve its intended purpose,
however, then specified rights-principles would be unnecessary. If
presumptive principles could provide the basis for inferring a specified principle referring to right R that entailed conclusions about the
possession of R in all the situations in which appealing to R was
appropriate, then these conclusions could be inferred directly from
the presumptive principles, without the aid of the specified principle.
If, for example, there were specified principles that reconciled the
opposing moral presumptions in Thomsons two cases, then these
reconciliations could be accomplished by relying on certain of the
presumptive principles from which the specified principles were
respectively inferred.
Indeed, this is precisely whet happens when, on examining
Thomsons two cases, we conclude that Victim is permitted to kill
Aggressor and that Rescuer is permitted to take the steak. It is hard
even to imagine anyone drawing these conclusions by relying on
principles that purport to provide the conditions that are necessary
and sufficient for possessing a right to life, or for having property
rights. I strongly suspect that similar remarks apply to Libertos
17
According to Oberdiek, Specificationism maintains that a right should be designated only after
the final interaction of all of the reasons bearing upon the justifiability of a given action (Specifying
Rights Out of Necessity, p. 135). My claim is that the reasons to which Oberdiek refers are expressed
in principlespresumptive principles, that is. If the final interaction of the reasons is a rational process
that results in a specified principle, then the latter is inferred nondeductively from presumptive principles.

SPECIFICATION AND MORAL RIGHTS

255

handling of her Case 1. As was noted above, there is no indication

that she relies on a specified principle that includes all the exceptions
to the possession of rights in particular situations. Rather, she seems
to base her analysis on principles that refer to factors that are relevant to that particular case.
The specificationist position is self-defeating because specified
rights-principles can serve their intended purpose only if there are
(nonspecified) presumptive principles principles that can exist
without specified principles and whose existence eliminates the need
for specified principles in reasoning to conclusions about individual
rights. Specified rights-principles are also unnecessary for the more
specific purpose of avoiding the logical problems that can arise when
rights-principles are interpreted as nonspecified universalized conditionals. Although presumptive principles are nonspecified, they
arent universalized conditionals, and inferences from them arent
deductive. In particular, it isnt possible to infer from presumptive
principles that Victims killing Aggressor is permissible and also that
Aggressor has a right to not be killed. Of course, the concept of a
presumptive principle needs a more extensive explanation than I
have provided here.18 The fact remains, however, that without
presumptive principles there is no explaining how specified rightsprinciples could ever be derived. Neither is there any way precisely
to characterize the exceptions to rights that, as Liberto points out,
are rooted in broader ethical theory.
I argued early in this discussion that specificationists must regard
their rights-principles as providing conditions that are necessary and
sufficient for possessing particular rights. It is this interpretation
(given the deductive model of principle-based moral inferences) that
allows specificationists to deny that Aggressor has a right to not be
killed by Victim, while affirming that Victim does have a right to not
be killed by Aggressor, and also that Rescuer is permitted to take the
steak without its owners consent.
The interpretation of specified principles with which I have been
operating is also central to my claim that the specificationist position
faces serious practical difficulties. However, the theoretical reasons
18
Donald Davidson presents an account of moral principles and nondeductive moral reasoning that
is adaptable to rights-principles and inferences from them to conclusions about individual rights (How
is Weakness of the Will Possible?, in Joel Feinberg (ed.) Moral Concepts (New York: Oxford University
Press, 1969), pp. 93113).

256

PHILLIP MONTAGUE

for rejecting the specificationist position apply to it regardless of

whether its principles are interpreted as containing conditions that
are necessary and sufficient for possessing particular rights. For it is
in the nature of specified rights-principles that they are tailored to fit
particular situations, and they must therefore refer to features of
those situations that are in general relevant to the possession of the
rights to which the specified principles refer. Hence, there must be
nonspecified principles from which the specified principles are inferred. And, for reasons given above, the existence of these nonspecified principles obviates the need for any specified principles.
The specificationist position is therefore self-defeating regardless of
whether its principles are interpreted as including all the exceptions
to the rights to which the principles refer.
Department of Philosophy,
Western Washington University, Bellingham, USA
E-mail: phillip.montague@wwu.edu