A Liberal Argument for Slavery
Stephen Kershnar
In this paper, I argue that state recognition and enforcement of slavery
contracts is consistent with the value of autonomy and with liberalism. By
“liberalism” I mean the notion that unjust harm to others (and perhaps pejorative exploitation) is a good reason, and the only good reason, for state coercion.1 My strategy is to show that formation and enforcement of a slavery
contract does not necessarily infringe upon anyone’s moral rights or lead to
pejorative exploitation and hence may not be disallowed on liberal grounds.
This argument thus challenges the compatibility of liberalism and the current
international ban on slavery at least insofar as the ban rests on considerations
of justice.
Slavery is the legal, quasi-legal, or coercive relation whereby one person
owns another. A common result of this relation is that the slave lacks certain
significant legal rights, such as a right of access to the courts, rights against
seizure, detention, and violence, and the liberty to go where he wants to go
and to work where he chooses to work.2 Ownership does not entail the
absence of these legal rights, since the rights that comprise legal ownership
may vary in different contexts. For example, merely because a person has a
legal right against torture, this alone is not enough to show that he is not the
property of another.
I. The Argument for Slavery Does Not Apply to Most Cases of
Slavery in the Contemporary World
The argument in this essay does not apply to the most cases of slavery in
the contemporary world (e.g., the enslavement of minors captured in a civil
war), since they are the result of unjust coercion. Still, the notion of a slavery
contract is relevant for a couple of reasons. First, certain contractual relations
(e.g., membership in the armed forces) resemble temporary slavery contracts
insofar as they involve the waiver of many legal rights. Second, there are cases
in which the desperation of certain persons or their families leads them to
enter into slavery contracts. One such type of case would involve persons in
desperate situations in the third world who wish to trade their liberty, either
temporarily or permanently, for benefits for themselves or those they love
(e.g., family members). In the early years of the United States, some immigrants agreed to indentured servitude in order to get to the United States. It
is also similar to cases in which young women from third-world countries
enter into the sex trade as a way to benefit their family. Another type of case
involves criminals who have been convicted and received extremely severe
JOURNAL of SOCIAL PHILOSOPHY, Vol. 34 No. 4, Winter 2003, 510–536.
© 2003 Blackwell Publishing, Inc.
A Liberal Argument for Slavery
511
penalties (e.g., the death penalty or a life sentence). Where agreed to by the
relevant parties, substituting slavery for the death penalty or life sentence
results in a transaction that might be mutually beneficial. And it could be used
to provide the means by which some compensation can be provided to the
family of a murder victim.
These cases are quite different from one another. In particular, they differ
in terms of the physical possibility of escape and permanence (e.g., indentured servitude has a time limit that might not be present in slavery in lieu
of the death penalty). The physical possibility of flight is true of ordinary
types of property (e.g., fowl) and hence does not seem to raise any new issues
in this context. The second difference raises the issue of whether temporary
enslavement is possible, since the rights against a person need not include
within them all the rights that characterize the full liberal concept of private
property.3
Here a brief account of private property rights is helpful. A property right
is a collection of particular rights that relate to control of an object. When held
by an individual, the right is a private one. The essential feature of the right
to control an entity is the moral standing by which to exclude others from the
use of it and to determine within limits whether and how it is used, changed,
or destroyed. Exclusion itself is a multifaceted relation that relates to the type
of object in question.4 For example, one might be able to exclude another
person from walking on or salting one’s grape fields, but not from enjoying
the sight and smell of them from a nearby hill or from viewing them from
the sky. A different sense of exclusion may be relevant where the entity is a
chemical structure, a dog, or a person’s body. That the exclusion relation has
many parts does not prevent the central aspect of it, the standing by which
to exclude others, from picking out the central moral relation. Where the
rights of exclusion are distributed between multiple persons over time (e.g.,
irrevocable leases), the identity of the owner becomes less clear. My argument
need not address this issue, because it attempts to establish that the permanent transfer of ownership of a person is in some cases permissible. If this
argument succeeds, the same reasons support the notion that in some cases
a temporary transfer is permissible.
II. The Slavery Contract and Moral Rights
My discussion makes two assumptions about injustice. First, I assume
that all moral rights, especially the claim elements, mark off the boundaries
of autonomy, and that since justice is concerned with the protection of autonomy, the infringement of a right is a (and probably the) litmus test for injustice.5 It is controversial whether pejorative (or wrongful) exploitation is a type
of injustice or provides a liberal reason for state coercion.6 For the purposes
of this argument I will assume that it is.
A. The Moral Rights of the Potential Slave
I am assuming that all rights rest on the value of autonomy. Hence, a
person has the right to enter into a legally enforceable slavery contract only
512
Stephen Kershnar
if such a right is grounded by autonomy. The first step in showing this is to
provide an account of autonomy.
1. An Account of Autonomy
The model of autonomy I adopt involves a variant on Harry Frankfurt’s
theory of a person and freedom of the will.7 On his account, a person has firstorder desires, which are desires to do or not do a particular action, and
second-order desires, which are desires to have or not have a particular firstorder desire.8 A parallel structure holds with regard to beliefs. On this variant
of his theory, an agent is morally responsible for her actions only if she
has the capacity to form second-order desires (and beliefs) that determine
which first-order desires (and beliefs) bring about her actions.9 Frankfurt
calls these effective second-order desires, second-order volitions. The basic
idea of my variant on Frankfurt’s theory is that autonomy involves two
elements. First, it involves the capacity of an agent to choose, on the basis of
higher-order desires (and higher-order beliefs) with which she identifies,
which desires (and beliefs) guide her actions (and thoughts). Second, it
involves the capacity of the agent to effectuate the chosen first-order desires
(and first-order beliefs) into action (and thought).10 For the purposes of
simplicity, I will leave out the discussion of beliefs in the following discussion of autonomy.
The capacity of an agent to choose on the basis of higher-order desires
with which she identifies which desires guide her actions, allows an agent to
constitute herself through the process of adopting and rejecting desires. There
is thus a close link between a person’s actions and her identity. On my theory,
it is in part the capacity to control her actions through her higher-order desires
that makes a moral agent responsible for her actions.
The first-order desires must cohere with the higher-order desires (and
perhaps also many of the other first-order desires). This coherence relation is
analogous to that posited by coherentist theories of epistemic justification.11
There are several advantages to this coherentist account of autonomy. First,
such an account can avoid the problem of an infinite regress of desires. In
particular, it need not posit an infinite regress of successively higher-order
desires, each of which evaluates the immediately lower-level desires. Nor
need it posit an arbitrary level at which higher-order desires are no longer
evaluated. Second, it can provide a powerful account of self-governance, a
notion that lies at the heart of autonomy. On this theory, a person is selfgoverned if her actions result from the coherent set of desires with which she
identifies (or would identify were she to think about it). Third, this theory can
account for a person’s identification with her desires (and beliefs). A person
identifies (or would identify) with her desires insofar as she has authorized
(or would authorize) them by both adopting them and ensuring that they
cohere with her body of desires and beliefs. Fourth, the coherence account
recognizes the close relation between rationality and autonomy, since a coherent body of desires will likely come about through a rational authorization
process. This is important, since rationality is an important condition of moral
responsibility as well as autonomy.12
A Liberal Argument for Slavery
513
This account of autonomy thus helps to make sense of the notion of a
person as self-determining being, because it makes sense of a how a person
constitutes herself by gradually choosing (or abstaining to choose) her desires
and beliefs in such a way as to form her character. This theory allows us to
thus develop a theory of a person’s character. A person’s character is her relatively stable and coherent body of lower-order and higher-order desires and
beliefs. It also allows a person’s self to be identified as the holder of the character. A person then determines her own actions by producing actions that
are (or can be) endorsed and effectuated by her character. A person is thus a
being that is self-determining insofar as she shapes (or can shape) her self,
character, and actions.
An autonomous person must not only be able to evaluate first-order
desires but must also be able to put them into effect.13 This can be seen via
thought experiments in which persons lack this capacity. So, for example, if
a person has the capacity to evaluate her first-order desires but cannot effectuate them because of paralysis or an inadequate range of options, we would
not say of her that she is autonomous. For example, imagine a person who is
trapped in a very tight pit that allows her no limb movement. She can merely
eat and drink the food and water that periodically falls into it. She also
has no contact with sentient beings. We would not say of her that she is
autonomous. Similarly, imagine a person who is unable to think or do
anything other than escape a fierce carnivore that constantly hunts her,
where escaping it requires every mental and physical resource the person has.
Intuitively, this person is not autonomous. Hence, autonomy is the capacity
of an agent to choose on the basis of higher-order desires (and beliefs) with
which she identifies which desires (and beliefs) guide her actions and
thoughts.14
The concept of autonomy might be thought to involve a set of capacities,
such as the one I have sketched above, or a condition whereby the agent
actually determines her self, character, and actions through her evaluative
processes.15 An agent has the condition of autonomy only if through her own
reasoning process she adopts certain desires and beliefs (both moral and nonmoral), initiates her projects on the basis of these desires and beliefs, and acts
in a manner that is in general consistent with them.16
The type of autonomy that we are normally concerned with is the
capacity version. There are a couple of arguments that support this conclusion. First, intuitively we attribute autonomy to a person with the relevant
capacities even if he does not exercise these capacities. For example, a person
who simply adopts the moral views of the local moral authority and thereby
adopts the views conventionally thought to be appropriate for his class
or station, without any critical reflection on these views, fails to authenticate
his moral views. However, we would not judge such a person to be less
than fully autonomous, thus suggesting that we view autonomy in the
capacity sense. Second, to the extent that we judge that moral responsibility
presupposes autonomy and we judge that the above person is fully responsible for his actions, despite his not having reflected upon his beliefs
and desires, we have another reason to view autonomy in the capacity
sense.
514
Stephen Kershnar
2. The Value of the Exercise of Autonomy
Beings that have autonomy as a capacity intuitively seem to have intrinsic value. Support for this comes from two sets of intuitions. First, we have
the general intuitive sense that more weight should be given to the interests
of fully functioning adult human beings than nonhuman animals. Second,
where the nonhuman animals or alien beings have the same autonomy as a
capacity as do humans, this intuition disappears. For example, we intuitively
consider the interests and rights of a pig with genius level IQ or an alien being
with the same intellectual and emotional capacities to be equal that of human
persons.17 The likely explanation for this is that autonomy is the primary
ground of intrinsic value in both human and nonhuman persons.
The exercise of autonomy (i.e., autonomy as a condition) is valuable
because of its relation to knowledge, pleasure, and freedom of the will.18 The
exercise of autonomy is instrumentally valuable because it leads to knowledge. This is because the justification element of knowledge requires critical
consideration of evidence and reasons and the exercise of autonomy forces a
person to give just such critical consideration.19 Also, the exercise of one’s
autonomy often is pleasurable, and this is another instrumental reason to
value the exercise of autonomy.20 In addition, a person who actively shapes
her character and actions through her selection of beliefs and desires is
less subject to environmental forces and more the author of her own
life. Since, on some accounts, the exercise of autonomy is a part of freedom
of the will, it has contributory value (i.e., it is part of an intrinsically valuable
thing).
3. The Value of Options
If the having of autonomy as a capacity is intrinsically valuable, then the
presence of certain minimal options upon which it can be exercised is also
valuable. This is based on the notion that autonomy presupposes the presence of certain minimal options that enable a person to shape herself via the
endorsement of various desires (and beliefs).21 The value here is contributory,
that is, the value of a part in relation to an intrinsically valuable whole.22 In
addition, the more the options a person has, the more opportunities she has,
other things equal, to exercise her autonomy. The other-things-being-equal
clause rules out cases in which additional choices eliminate future opportunities to act in an autonomous manner.23 The idea here is that a greater range
of options will lead to a greater degree of selection and prioritization among
alternative desires (and beliefs). For example, if my only two choices are to
become a highly ethical and capable judge and a highly ethical and capable
physician, then I need not make very many choices in terms of the principles
to govern my life. If I also have to choose between the above options and the
choice of being a promiscuous guitarist for a rock supergroup, then more
selection and prioritization must be done.
Even though there is instrumental value in having good and bad options,
the option to eliminate future options is valuable. Such an option encourages
A Liberal Argument for Slavery
515
me to determine for myself the value and priority to be placed on future
options. This consideration and prioritization of such future options has considerable value in my shaping my life.
In addition, in some cases a person may rationally act on the basis of
values that lead to the elimination of her autonomy (either as a capacity or
as a condition). For instance, a person suffering from the earliest stages of a
debilitating neurological disease may place great disvalue on humiliation and
physical degradation, and thus desire death.
Also, a woman who enjoys the security and the warmth of a permanent
family relation may, particularly outside of the United States, choose to enter
into a type of polygamy that differs from slavery only in degree. Since the
loving and secure relations presuppose the irrevocability of everyone’s status,
this new wife’s value system may make the quasi slavery rational. In fact,
many of the decisions we make constrain our future act options both morally
and in the sense that we are now subject to coercive state laws (e.g., the decision to have children). An important case occurs in which a person trades off
her autonomy as a condition in order to benefit others with whom she has a
special relationship. For example, a Thai woman who sells herself into temporary sexual slavery in order to provide her parents and sister with a way
to escape subsistence farming and a substantial chance of starvation might
make such a trade-off.
Another reason such options are valuable is that in some cases a person
may shape her life in a desired manner only where she temporarily gives up
some of her act options. For instance, alcoholism may disrupt a person’s
ability to have a certain type of life. Where an alcoholic temporarily sacrifices
various options in order to exercise autonomy (e.g., by committing herself to
a detoxification program) and in return reasonably expects to eliminate this
disruption, the sacrifice may be an effective strategy by which to gain control
over her life.
The waiver of future act options may end up decreasing the degree
of self-determination a person has over her lifetime. How then can selfdetermination justify the waiver of act options? On my account, act options
and the exercise of autonomy (i.e., autonomy as a condition) are valuable
because they allow us to shape our life in accordance with self-chosen
principles. The value of the waiver rests on the value of having complete
control over one’s life (at least within the boundaries of one’s duties to
others).
The degree of control can be a function of a number of factors. One factor
is the level of control. Control can be first-order, that is, ranging over the
agent’s identity and her relation to the world (e.g., her mental states, actions,
and relation to external objects), or second-order, that is, ranging over firstorder control. The highest-order level is also reflexive, since a person has the
ability to end every level of control (e.g., by killing herself). I leave aside
whether persons can have still higher orders of control. A second factor is the
significance of control (i.e., how central is the object of control to the person’s
identity and the shape of her life). This shape of her life is itself a factor of
her experiences and relation to the world. For example, a person’s control
516
Stephen Kershnar
over whether she has a child is likely more significant than control over
whether she wears Nike running shoes. A third factor is the duration of the
object of control and its effects. For example, control over one’s identity and
shape over one’s life at age twenty is ordinarily more significant than similar
control at age ninety-five, in part because the events or states that are
controlled last longer. There may be still other factors. In valuing complete
control, I have in mind a state that in the present includes unfettered secondorder control. Respecting complete control may result in the failure to protect
maximal control, which is the product of the significance of control (whether
first-order, second-order, or both) and its duration. This is because a person’s
decision to kill herself or become a slave may result in her not having control
for as long a period of time or not addressing many significant issues (e.g.,
her relation to God).
The argument for valuing complete control over one’s life is that what is
valuable is the control that allows a person to be the author of his own life.
A person is the author of his life to the extent that he is responsible for the
length and shape of his life (e.g., his identity, experiences, and relation to
external objects). This requires that his preferences be formed, prioritized, and
applied. This authorship can involve the decision not to continue with life or
to have a life shorn of control. It can also be had despite a person’s failure to
consider his options systematically. In contrast, since maximized control may
involve restricting a person’s choices, it may prevent a person from being the
author of his own life to the same extent. These restrictions might increase
the amount of control by increasing the significance and duration of the
objects of control. This can be done, for example, by preventing persons from
killing themselves or by forcing them to think in a systematic way about their
value system and their relationships.
Consider an analogy. A writer of fiction is the author of his characters so
long as he develops them in the way and to the extent that he prefers. This
is true even if he does not write very much about them, does not consider
many aspects of their lives, and does not write more than one novel about
them. The writer would have a greater amount of control over his characters
if he was forced to write many long novels in which he systematically laid
out each character’s life and their relations to his other characters. The former
case intuitively seems to respect more appropriately the author’s creative
choices, even though the latter likely involves a greater amount of creative
activity. This is because the former case involves the person applying his
creative abilities to the direction and extent of character development.
Similar reasoning suggests that complete control more appropriately respects
a person’s autonomy even though maximized control likely involves a greater
amount of autonomous activity.
Given this view of autonomy, the next step then is to determine the structure of the waiver of autonomy.
4. The Structure of Autonomy-Related Waiver
The notion that a person waives her autonomy is ambiguous. It might
mean any of the following:
A Liberal Argument for Slavery
(1)
(2)
(3)
(4)
(5)
(6)
517
A person refuses to exercise the capacities that constitute autonomy.
A person refuses to act on the basis of her analysis of the balance of
reasons for or against an act.
A person waives or transfers some, if not all, of the moral rights that
protect her autonomy as a condition or capacity.
A person waives or transfers some, if not all, of the legal rights that
protect her autonomy as a condition or capacity.
A person waives or transfers some, if not all, of the moral rights that
protect her autonomy as a condition.
A person waives or transfers some, if not all, of the legal rights that
protect her autonomy as a condition.
I shall argue that waiver consists of both (3) and (4), although these entail (5)
and (6).
Version (1) involves a person giving up her autonomy as a condition.
Specifically she gives up deciding to hold a given belief/desire set, analyzing the reasons for or against an act, or acting on the basis of this analysis.
In contrast, (2) involves the agent merely giving up the third element. For
example, consider an Air Force bombardier who is deciding whether to
release an incendiary bomb onto an undefended city. A person who has
waived her autonomy in accordance with (1) does not deliberate over
whether or not she ought to release it. A person who has waived her autonomy in sense (2) follows the order to release the bomb no matter what
outcome results from her deliberation. Neither (1) nor (2) captures what we
mean by waiver in the context of the slavery contract, because neither sense
results in the person who waives her autonomy having a changed moral relation with regard to another that is characteristic of, if there is such a thing,
legitimate slavery.
In addition, an agent may not waive her moral duties, because these are
not her property.24 So, for example, a man cannot waive his duty not to rape
a twelve-year-old girl, for he does not own the duty (i.e., he is not the claim
owner). Waiver in the sense of (2), and in some cases of (1), will sometimes
result in the person refusing to do what she knows she has a duty to do,
because the person either will not deliberate or will deliberate and recognize
the balance of moral reasons for an act but then have no commitment to act
on the basis of this recognition. This is morally impermissible, since it
involves her exercising a moral power that she does not have. Versions (3)
through (6) involve the person waiving some, if not all, of her rights. This
does not involve a person directly avoiding her duty. She may know that indirectly a person who is the owner of these rights may use them to coerce her
into doing wrongful actions, but then she will not be culpable with regard to
those actions. As a moral matter, a person may sometimes perform actions
that have a significant chance of leading to a right infringement where she
does not intend that the right infringement come about and where the risk is
outweighed by other results.
Thus, an agent may waive her moral rights with regard to autonomy
as a condition or both forms of autonomy only if she retains her intent to
518
Stephen Kershnar
deliberate over and to act on her analysis of the balance of reasons for and
against an act. A second condition on permissible waiver occurs where the
waiving agent might be coerced into infringing upon the rights of others. This
condition requires a consideration of the consequentialist threshold. The consequentialist threshold is satisfied where the sum of the product of the value
and likelihood of the positive consequences occurring and the autonomybased value of the agent having act options exceeds the product of the disvalue and likelihood of negative consequences occurring. The idea here is that
the value of autonomy requires that a person take into account the likelihood
that he will be transformed into one who unjustly infringes on others’ rights
or does other wrongful acts. For example, where it is 95 percent likely that a
voluntary slave will become an assassin who targets innocent political leaders
(e.g., via hypnosis or neurosurgery) with no offsetting benefits, the choice to
become a slave would exceed this threshold. I leave aside the issue of whether
this threshold is to be judged from an ex ante perspective, as this raises murky
issues of the deontic status of unlikely but actual consequences an act.
An objector might assert that if a person retains her intent to deliberate
and act on the result of her deliberation, then she has not surrendered any
part of her autonomy.25 However, an important feature of autonomy is the
right-protected options that allow one to exercise the relevant capacities
without interference, and these may be waived. To the extent that significant
options are waived, part of the agent’s autonomy is lost. Nor is a waiver of
the claim against interference with an act inconsistent with a person having
a duty to perform that act so long as other values (e.g., the value of selfshaping life) override or undermine the probability-discounted disvalue of
the duty being unsatisfied.
Two other conditions on valid waiver should also be noted. First, an act
of waiver is valid (i.e., morally binding) only if an agent was not wrongfully
coerced into doing it. For if she was wrongfully coerced, then her waiver does
not reflect her autonomous choice. Second, the condition that the agent’s
rights are her own prevents her from waiving rights that she has already
given up. We thus end up with the following account of rights waiver:
(7)
A person may permissibly waive or transfer some, if not all, of the moral
and legal rights that protect her autonomy as a condition or capacity if
and only if the rights are hers, such a waiver or transfer is not the result
of wrongful coercion, and the unjust results of this waiver are unintended and do not exceed the consequentialist threshold.
A number of assumptions underlie this conclusion. First, I shall assume that
all duties to oneself are waivable. While an agent who waives her rights may
end up harming herself, this is not an impermissible infringement upon her
claims against herself if, as I shall assume, these claims are waivable.26 Second,
the deontic status of an act is independent of the agent’s motive. For example,
if waiver is morally wrong where it exceeds the consequentialist threshold,
the agent’s belief with regard to this aspect of it goes toward his blameworthiness but not toward the act’s deontic status. Here I am assuming that the
intention with which an act is performed is distinct from its motivation.
A Liberal Argument for Slavery
519
It should also be noted that there might be cases in which a morally valid
act of waiver ought not to receive legal recognition (e.g., when this threatens
to produce race riots). Nevertheless, morally valid waiver creates a moral
reason to respect a person’s choice to waive or transfer her rights. Furthermore, the Constitution probably allows for legal recognition of the
slavery contract, at least in the case where it is a convicted criminal who
agrees to slavery in return for a lessening or elimination of his scheduled
punishment.27
5. The Notion That One Right Is Included in Another
Moral rights protect autonomy. This is in part why we think that there
are rights to shape one’s life in a number of important areas (e.g., intimate
relations, the expression of ideas, religion).28 It is also in part why we think
that there is a right to do wrong.29 If rights protect autonomous decisions
about one’s life, and if one can autonomously decide to become a slave, then
a right protects this decision. And if persons have a right to become a slave
and the right to give a benefit to another, these rights together generate a right
to create slave contracts.
In general, one right generates another where the first contains the second
or where the second is instrumentally valuable in protecting the first. Containment is a conceptual notion.30 For example, the right not to be assaulted
at any time includes the right not to be assaulted on Tuesdays. A second type
of generation occurs where one right is instrumentally valuable in protecting
or, if this is different, interpretive of another right. In this type of case, the
second right generates the first. For example, the right to privacy may include
the right against state officials wiretapping a person’s telephone without
reasonable cause.31
In this context, consider a contract with regard to which an individual,
Joan, has broad moral rights (e.g., to her labor). Such a set of rights entails
the right to enter into a contract whereby Joan waives some of the elements
of her rights (e.g., certain claims and privileges) in return for something else
she values (e.g., money). The inclusion relation is probably a conceptual one,
because the power in this case is the standing by which to demand satisfaction of or waive a duty or privilege for nearly any reason, and this includes
reasons such as financial gain. An apparent exception to this rule occurs
where the reward that Joan seeks is itself an infringement of another’s right
(e.g., having her ex-husband shot dead). However, this is not a real exception,
because in such a case Joan’s powers do not extend to actions that she does
not have a privilege to perform.
As a matter of justice, if a person has the right to X or not to X and another
has the right to Y or not to Y, then, other things being equal, the first may
contractually waive his right not to X in return for the other granting him Y.
The other-things-equal clause relates to other rights or moral entities that
make such an exchange impermissible. In this case, a person may waive her
moral claim not to be a slave in return for payment or something of value,
since she has the relevant moral powers over the claims and liberties that in
part constitute the right not to be a slave. Similarly, the owner may waive her
520
Stephen Kershnar
right to wealth or other benefit in return for the potential slave waiving one
of her rights.
An objector might claim that this notion of inclusion is mistaken. She
might argue that since legal and moral rights have the same basic structure,
if the inclusion relation does not hold with regard to legal rights, then it does
not hold with regard to moral rights. She might cite the following type of case
in support of her claim that the inclusion relation does not hold with regard
to legal rights.32 James has a legal right to release or not release lawfully
obtained information about Trudy. Trudy has the legal right to give or not
give money to James. It does not follow from these assumptions that James
may trade his right to release information about Trudy’s having had an adulterous affair in return for her giving money to James. The problem with this
objection is that one component of the legal right, the power, is narrower than
it may at first appear. In the context of legal rights, the state determines the
scope of the relevant power. The legislature may narrow the scope of the
power so that it does not include the ability to waive the right to release
injurious information in return for money. This is not to say that the state’s
narrowing of the scope of a person’s power is morally permissible, just, constitutional, or even within the limits of its authority. I would argue that in the
case of the U.S. government, a ban on blackmail is none of these. Rather, it is
merely to say that under a positivist account of the law, the state’s (impermissible) narrowing of its citizens’ powers explains how it can restrict blackmail. Hence, this type of case does not present a counterexample to the notion
of rights inclusion.
6. Involuntariness and Contract Validity
An objector might assert that the slavery contract is invalid because it was
involuntarily entered into.33 He might argue that the lack of reasonable
options for the potential slave resulted in the choice being coerced (or,
depending on one’s terminology, the result of an analogue of coercion) and
that an act of consent (or promise) that is coercively obtained is morally
invalid. The problem with this argument is that the involuntary nature of a
person’s consent to (or promise connected with) a contract does not as a moral
matter invalidate the contract. An exception exists where the other contractor does not cause the lack of reasonable options. Consider the following case.
During an expedition into Africa, a highly venomous black mamba bites a
wealthy scientist. He is quickly taken to the house of a local doctor, who offers
to sell him the doctor’s only portion of mamba antivenin for the market price.
The scientist quickly agrees and signs a contract. He is then given the
antivenin. After a month of lying near death, the scientist recovers. He then
refuses to pay, arguing that the contract is invalid since his consent was
coerced.
Absent an exorbitant price that might indicate pejorative exploitation, it intuitively seems that the scientist has an obligation to pay for the antivenin even
though his consent was coerced (or quasi-coerced). If so, then either volun-
A Liberal Argument for Slavery
521
tary consent is not a necessary condition for a morally valid contract or
voluntariness does not require the availability of reasonable alternative
actions at the time of the consent.34 Either move is also available in the
case of the slavery contract. Either the alternative of death makes the wrongdoer’s consent involuntary, but this does not morally invalidate the contract,
or the lack of reasonable alternatives, given the historical sequence
that brought about this situation, is not sufficient to make his consent
involuntary.
In general, a paradigmatic voluntary act is one performed by a competent individual who is not subject to duress, coercion, or ignorance. In some
contexts, such as the slavery contract, an offer is coercive in that there may
be a huge objective difference in payoffs between the proposed outcome and
the outcome that would otherwise occur. This difference may produce enormous psychological pressure to choose the best option (e.g., the least of a pair
of horrendous alternatives). However, where the offeree is responsible for the
unreasonableness of his other alternatives, coercive force need not undermine
valid consent, because valid consent is a moral notion that is in part a function of the way in which a person’s situation became a desperate one.35 Consider, for example, a person who is a gambling addict and who feels he must
choose to sell some of his blood in order to pay for treatment before he loses
all of his house. He subjectively views the offer as having coercive force, and
from an objective perspective, the payoff difference is great enough to constitute coercive force. Since the offeror did not contribute to the absence of
other reasonable alternatives, the offeree’s consent is valid, even though the
coercive force in effect dictates his decision. The same sort of analysis should
be given with regard to slavery contracts (e.g., where the agents of murder
victims offer to waive execution in return for a murderer waiving his right
against slavery).
B. The Moral Rights of Third Parties Are Not Infringed Upon by the
Slavery Contract
It might be claimed that the legal recognition and enforcement of the
slavery contract may be banned on liberal grounds, since it harms the public.36
The harm to the public in this case might take a number of different forms.
First, the presence of such a legally enforced institution might make the public
increasingly insensitive to the suffering of others and dampen the socially
useful virtue of benevolence.37 Second, the spectacle of slavery might produce
psychological harm to persons, harm similar to that of seeing people starving in the streets. Third, widespread opposition to the institution of slavery
might lead the government to protect the institution of slavery against an
unsympathetic public via incursions into the rights, especially privacy rights,
of citizens. These incursions would resemble those that have accompanied
the U.S. government’s attempts to decrease recreational drug use. Fourth, a
contract with such grave consequences requires a high threshold of voluntariness in order to be valid. The costs to the citizenry of ensuring that this
threshold has been met might be sufficiently high as to constitute a public
harm.
522
Stephen Kershnar
The problem with this objection is that not all setbacks to the public interests are harms of the type that constitute a reason for state coercion. For
instance, the promotion of racist ideas may on the whole bring about a setback
of the public interest insofar as it exacerbates racial tension in the community.
However, speech promoting these ideas might be thought to be protected
against state coercion because the harm does not come about in a way that
infringes upon any person’s (or the community’s) rights. This is because the
speech brings about a right infringement only via the deliberation and decision of morally responsible agents.38
What the public has a right to depends on the proper sphere of public
concern. On a liberal theory, there seem to be certain areas relating to an individual’s ability to shape her life that lie outside the public realm. This sphere
includes concerns relating to a person’s body, her landed and chattel property, and her vital life decisions. The activities that fall squarely within one of
these categories are ones that do not directly cause right infringements and
harm to others. The concern for setbacks to the public interest in benevolence
and other socially useful attitudes is the sort of broad-based interest that if
recognized will almost entirely eliminate the private realm, because the
aggregate presence of an unconventional activity often dampens socially
useful attitudes.39 Similarly, the notion that the public’s psychological discomfort with certain activities should be given legal recognition (and weight)
will also almost entirely eliminate the private realm. For example, this discomfort could have been used in the past to support laws banning public
interracial affection or private gay conduct. If however, a private sphere
should be protected against coercion based on these reasons, then these harms
do not justify criminalizing or refusing to enforce slavery contracts. Furthermore, if there is a concern that the government will trespass upon the rights
of citizens in the protection of unpopular practices, then the solution is to constrain the government. This is especially true in the context of liberal values.
If the public’s psychological discomfort with certain activities should be given
legal recognition but minimal weight, then it becomes unclear whether this
reason can override the autonomy-based reasons in this or other contexts.
The concern for the costs of ensuring voluntariness in slave contracts and
the costs of enforcement may turn out to be sizable, and this may weigh
against putting large amounts of resources into an area. However, the voluntariness requirement may fairly be placed, in the form of an administrative fee, on the prospective contractors. In addition, the costs of enforcing
such a contract may weigh against committing state resources to their
protection. However, this does not provide a reason for the state to refuse to
recognize, to disallow private enforcement of such contracts, and to prevent
wronged owners from receiving compensation when such contracts are
breached by third parties.
An objector might note that since the state is asked to enforce the contract, it is in effect a party to the contract, and thus the fully voluntary slavery
is within the public realm.40 If granted, the objection will result in the content
of all enforceable contracts being an area for potential state intervention. The
upshot is that a private realm for individuals cannot involve enforceable
contracts, thus restricting the sort of arrangements that can be made within
A Liberal Argument for Slavery
523
certain physical areas (e.g., one’s own house), activities (e.g., procreation), and
the dissemination of certain information (e.g., confidentiality agreements).
An alternative approach is for the state to enforce any valid contract that
does not constitute or directly bring about right-infringing harm. The latter
approach better protects autonomy in that it allows persons to have the option
of contractual relations in the private realm.
Thus, the slavery contract probably does not involve a infringement of
the rights either of the slave or the public. Still, however, it might be objectionable on the liberal grounds that it involves pejorative exploitation.
III. The Slavery Contract and Pejorative Exploitation
There are cases of pejorative exploitation that are wrongful but neither
right-infringing nor harmful. Perhaps the slavery contract is impermissible
on liberal grounds because it involves this pejorative type of exploitation. To
explore this notion, let us begin by considering the following case:41
The Lecherous Rich Man
A rich man offers to pay for the very expensive surgery that alone can save
the life of Helen’s child provided that Helen becomes for a period his mistress. Helen values the life of her child much more than she values avoiding
the sexual arrangement. The man is from a different country and does not
know Helen other than by description. He has no antecedent duty to Helen
and is considering spending his money to save other children in the third
world. The surgery is so expensive that no state or charitable agency will pay
for it. Thus, if the rich man does not pay for the expensive surgery, the child
will die. Helen has sex with the man in order to save her child’s life.
My background assumption here is that prostitution does not in itself involve
a right infringement. The rich man does not infringe upon Helen’s rights,
since he has rights to give or not give the money to Helen and to have or not
have consensual sex with her. Hence, he has the (moral) power to waive his
claim to the money in order to exercise another right. Nor is Helen harmed
by this transaction. She is in a better position than she would be in under the
expected and a just course of events. Assuming her choice is an informed one
that is based on her rational assessment of her circumstances, she consents to
the arrangement.
An act of exploitation is wrong in virtue of its being exploitative where
one side takes unfair advantage of the other side.42 The “in virtue of” clause
is designed to screen out cases in which an act is exploitative and wrong but
the act’s wrongfulness is not a result of its being exploitative. For example, if
a thief steals a loaf of bread from a store and then outruns the heavy-set store
owner, the thief exploits the store owner’s lack of speed and acts wrongly, but
there is no connection between the exploitation and the wrongfulness of his
act. If an objector denies that this last case is an example of moral exploitation, she probably does so on the basis that the moral notion has the in-virtueof condition. This, of course, is exactly my point.
524
Stephen Kershnar
The feature of pejorative exploitation that makes it unfair is that the probable benefits of the interaction accrue disproportionately to the person (or
group) with the stronger position because of their relative position. In the case
of transactions, such an unfair advantage must result from an asymmetry in
bargaining potential between the two parties. That an outcome conflicts with
a person’s self-image, religious principles, or values does not make a transaction exploitative. For if this were so, then many choices reflecting hard personal or economic circumstances would be exploitative, and this does not
seem intuitively correct. So, for example, a socially unattractive individual
who has to choose between being single (which she abhors) and being with
men she finds conflict with her self-image does not seem to be exploited if
she chooses one of those men. This is so even though this is precisely the sort
of hard choice that often causes a person to feel that she is between a rock
and a hard place. A party’s bargaining potential is primarily a function of
her external characteristics (e.g., resources and circumstances), whereas her
bargaining ability is primarily a function of her internal characteristics (e.g.,
information, toughness, patience, perceptiveness).43 Can this model account
for the wrongfulness of the slavery contract?
We need an account of when a party receives a disproportionate share of
the transactional benefits.44 Such an account rests in part on the notion of a
reservation value. This is a value at which a party’s gain from the transaction
is equal to her best alternative to a negotiated agreement. The reservation
value is then used to set the reservation price. This is the minimum price she
is willing to accept to enter into the agreement. The space between the parties’
reservation prices constitutes the bargaining range. Any outcome within their
bargaining range involves a transactional surplus that is the difference
between the buyer’s and seller’s reservation prices. This surplus is the gain
by the two parties from the contract. A reservation price may be judged as
the party’s actual reservation price (determined by a counterfactual about the
minimum price that a party would accept) or a moral reservation price (the
minimum price that a party ought to accept). The latter may take into account
the subjective preferences of the relevant party, but it is not fully subjective,
since the party may be mistaken about what is in her interest. So, for example,
if a worker will not (or ought not to) take a particular job for less than $10
and an employer will not (or ought not to) pay more than $15, the transactional surplus is $5.
The claim of exploitation in a mutually beneficial transaction involves the
stronger party using the vulnerability of the weaker party to take a disproportionate share of the transactional surplus relative to each party’s moral
reservation price. In this context, the notion of proportionality is a function
of the relation between the relative contribution of and benefit of the transactional surplus taken out by each party. This feature cannot be a consequentialist one, since unfairness is thought to be a moral relation between the
two parties independent of its effects on others, whereas plausible consequentialist analyses do not screen out third-party effects.
An account of disproportionate gain should focus on whether the
stronger party is taking unfair advantage of the weaker party, not on whether
the stronger party takes advantage of an unfairness to the weaker party. In
A Liberal Argument for Slavery
525
the latter case the stronger party is not the cause of the unfairness to the
weaker party. Focusing on the latter would make most transactions, no matter
how rational and appropriate given the background conditions, unfair if the
background conditions reflect injustice or unfairness, and this seems counterintuitive. Unfairness in the context of a disproportionate gain is a property
of particular transactions, not a property of macrostate distributions of wealth
(or other resources). This distinction is useful because pejorative exploitation
can take place within a just economic system and because a fair transaction
can take place within the context of an unjust economic system. Also, since
the wrong-making features of distributive injustice and economic exploitation may differ, the two should be kept separate for the purposes of analyzing the permissibility of different acts.
In the slavery context, in some cases the potential slave faces an enormous loss if the contract is not formed. To see this, consider the case in which
the expected (and just) treatment that a murderer expects to receive is the
death penalty but in which he instead is released from the death penalty in
return for his agreeing to become a slave. One might be tempted to say that
volunteering for slavery is against even a condemned man’s interests, but on
standard theories of interest, this does not appear to be so.
There are three theories of what is in a person’s interest.45 Hedonic theories identify a person’s interest in terms of pleasure or the experience of desire
satisfaction. Desire fulfillment theories identify a person’s interest in terms of
the satisfaction of his desires. Objective-list theories identify a person’s interest in terms of factors that are independent of the having of various experiences or desire satisfaction. These objective-list theories identify a person’s
interest in terms of such things as meaningful relationships with others, moral
goodness, and knowledge. On any of these theories, a person’s death will
usually result in a great setback to his interests. In contrast, slavery will allow
for many moments of pleasure and the satisfaction of some desires and
objective-list elements. Here the form of slavery is important, but even under
chattel slavery as practiced in the U.S. South, there were in some cases significant opportunities for meaningful relationships with others, moral goodness, pleasure, and desire satisfaction. Since a transaction is viewed from an
ex ante perspective, the murderer’s expected gain is significant.
The parties waiving the right to capital punishment gain considerably less
from the slavery contract than does the murderer. I am assuming here that
the right to execute a murderer is held by the victim (via her agent or the
state).46 The murder victim, having lost her life, at most gains a form of compensation that can be used to benefit her family, friends, or valued projects.
The value of such a gain is considerably less than that of the murderer, since
very few persons would want to trade such benefits for one’s life, whereas
many people would want to trade being killed for enslavement. The state
probably does not gain much. Since the state is not a conscious entity, it has
no interests and is thus not capable of receiving a benefit. The collection of
persons (perhaps in some complex interrelationship) that comprise the state
may gain something, namely, the tax dollars that would be spent to execute
a murderer, but again this seems considerably less than the murderer’s gain.
This can be seen in that very few persons would want to trade the money the
526
Stephen Kershnar
state gains for their lives. Other benefits and costs of the transaction (e.g., a
change in deterrence, racial division, and the benefits of the work done by
the slave) are harder to assess, because they are the sort of factors that are
unclear in the case of the death penalty and even more speculative in the case
of the reintroduction of slavery. The best that can probably be said here is that
the murderer gains more than the victim; whether he gains more than the
victim and state combined is unclear.
Each party’s contribution is even harder to assess. How much does the
victim (or the state) contribute? One measure is to consider the case in which
a victim is informed that a fully culpable murderer will kill him and ask how
much he would pay to have that murderer executed rather than merely
enslaved, because in some cases the victim’s contribution is the value to her
of having a proportionate punishment imposed as compared to a specific
lighter punishment. Here much depends on the identity of the murderer, the
surrounding situational factors (e.g., the amount of money the victim has or
ought to have at her disposal), the nature of the slavery (e.g., chattel or a lesser
form), and the victim’s preferences (e.g., the punishment that she would have
preferred). By requiring the victim’s or her agent’s consent, we at least make
it likely that the transaction is expected to produce a net gain. The murderer’s
(and the state’s) contribution is even murkier and harder to assess than the
victim’s contribution. At best, the relative weight of the contractors’ contributions seems murky.
In general, where the contribution and benefit of a transaction are largely
dependent on the preferences of the transacting parties and a wide range of
other empirical factors, the best that can be said is that the case for such a
transaction being exploitative is unclear. Given the general priority of liberty
in a liberal state, the burden should rest on those who seek to disqualify a
mutually beneficial transaction on the basis of its being pejoratively exploitative. Unless this burden is met, and in cases such as the death-penalty waiver
I doubt that it will be, this objection to the slavery contract fails.
In summary, the slavery contract does not infringe upon anyone’s moral
rights. Nor is it likely to be shown to be pejoratively exploitative. I now turn
to some general objections to the slavery contract and to the structure of my
argument.
IV. Objections to Deserved Slavery
A. Liberty Does Not Include the Option to Alienate One’s Liberty
John Stuart Mill argues against slavery on the basis that the justification
of liberty is the value of voluntary choice. Depending on one’s interpretation
of Mill, the value of voluntary choice is grounded by the value of utility,
autonomy, or both. Mill argues that liberty does not include within its scope
the liberty to alienate one’s future liberty. His argument for this is that such
an option would thwart the justification of liberty.47 The problem with Mill’s
argument is that it does not follow if we assume that the value of liberty rests
on the value of autonomy. The more valuable options (i.e., those that are
A Liberal Argument for Slavery
527
important to shaping one’s life or that are gateways to a large number of other
valuable options) are on average more important to the process of shaping
oneself. The option of selling oneself into slavery is one of these more valuable options in part because it forces a person to consider the nature and value
of self-determination and in part because it is an option that can close off
many other valuable options.
B. The Slavery Contract Is Invalid Because It Is Irrational
Jean-Jacques Rousseau argues that a contract whereby a person alienates
autonomy is so irrational as to invalidate the contract, and that the irrationality of it relates to the impossibility of proper compensation.48 It is not
clear, however, why payments to third parties or other benefits that might not
otherwise occur could not make such contracts rational. The preferences
expressed by the person who may be entering slavery do not appear irrational, since such preferences may relate to an agent’s own welfare or the
welfare of third parties and may warrant the sacrifices accompanying slavery
given the other options. So, for example, the Thai woman who agrees to
become a sexual slave temporarily so that her family does not starve to death
does not act on the basis of irrational preferences or choose irrational means
by which to promote those preferences.
Arthur Kuflik, in a related objection, argues that it is irrational to abdicate the right to guide one’s beliefs and actions on the basis of one’s assessment of the balance of moral reasons relating to them,49 because there is a
possibility of potentially unlimited disvalue in the action that the guardian
or owner might require of the person who abdicated her autonomy. Kuflik
might provide the example of a person who abdicated her moral autonomy
to save her child from starvation. But after having waived her autonomy, her
master might require of her that she not feed or even kill the child. A slave,
however, cannot waive the moral duties that she owes to others, since they
are not her property. Hence, any order to do that which is a violation of a
duty owed to another does not generate a corresponding obligation. The presence of these duties also prevents a slave from refusing to deliberate over the
balance of reasons for and against an act. The owner might take legitimate
steps that may manipulate or destroy the slave’s capacities, thereby making
her into a tool of injustice. For example, the owner may impose without
legitimate complaint on behalf of the slave such mind-control techniques
as psychosurgery, hypnosis, and mind-altering drugs and thereby turn the
slave into a brainwashed agent of injustice. It is not irrational, however, to
take this risk into account when deciding whether to enter into the slavery
contract and decide that in some cases the expected benefits outweigh the
risk.
The objector might still claim that remaining an active moral agent
requires the ability (and the opportunity) to deliberate and act on one’s moral
duties to others. He might further claim that if one is a slave, then one is completely subject to the will of another and hence unable to deliberate and
respond to one’s duties to others. Thus, the objector might conclude, being a
528
Stephen Kershnar
slave excludes being an active moral agent, and one has a duty to be a moral
agent. An initial problem with this is that it is empirically false. Some types
of slavery (e.g., sexual servitude) do not always undermine a slave’s ability
to deliberate and act on her duties to others. However, even if this were not
the case, it still would not follow that such a choice would be impermissible
let alone irrational. This is so for two reasons. First, in some cases it is permissible to choose to no longer be a moral agent. For example, it seems intuitively permissible to choose euthanasia in some cases. If this is correct, then
there is no duty simpliciter to retain one’s moral agency. In some cases, it is
also permissible to continue to exist without moral agency. For example, consider a case in which a person with extraordinarily valuable antigens suffers
from a horrible illness that causes him constant pain. It is intuitively permissible for him to choose to continue his biological life (in order to benefit
others) but to so damage his thought process that he no longer can feel pain,
even where this results in the loss of his moral agency. Second, we need to
distinguish the choice to not satisfy his duties to others and the choice to
risk losing the ability to satisfy his duties to others (where these are other-thingsequal duties). Consider the case of Odysseus. He had his men lash him to
the mast as a way of avoiding a future choice to go the Sirens. In so doing,
one can imagine that he risked losing the ability to prevent his men from
committing acts of sexual violence against the weaker members of the crew
and thus risked losing the ability to satisfy his duties to protect them.
Nevertheless such a risk was permissible under the circumstances. In contrast, it would not have been permissible for Odysseus to permit or order
such attacks. A person who chooses to be a slave similarly risks losing the
ability to satisfy his duties, but in so doing has not chosen to act against his
duties to others.
C. The Slavery Contract Is Invalid Because It Involves a Duty to
Bring About an Impossible State of Affairs
Baruch Spinoza argues that a person may not agree to waive his autonomy, since the agreement is impossible to abide by. Since a duty cannot be
generated with regard to the bringing about of an impossible state of affairs,
such an agreement will not generate a duty.50 However, in a slavery contract,
a person need not, and in fact may not, give up her tendency to make an independent judgment of the balance of reasons with regard to a prospective
action. Rather, all she may do is agree to waive her protective moral and legal
rights. Such an agreement can be satisfied, since it requires of the slave only
that she try to satisfy the owner’s preferences and permit legal and nonlegal
sanctions should she fail to satisfy them. Neither result is impossible even as
a practical matter.51
Randy Barnett provides a variant of this objection. He notes that absent
physiological changes, a person cannot transfer partial or complete control
of her body to another. He then argues that given that control cannot be
transferred, the right to control cannot be transferred.52 Again, however, this
objection succeeds only if what is transferred is the control rather than the
protective rights.
A Liberal Argument for Slavery
529
D. While Right-Based Accounts of Justice and Liberalism Allow for
Slavery Contracts, Other Versions Do Not
An objector might reject my assumption that the infringement of rights
is the only test of justice. She might assert that there are institutional arrangements that make society less just without violating any rights. Her example
involves the assumption that children have no right to education. She then
asserts that a government that failed to ensure the availability of education
intuitively seems to make the society less just (e.g., because people would be
less likely to know what their rights are and what they can do to defend them)
without violating any particular person’s rights.
However, if we view justice as reflecting the duties owed to and liberties
(absence of duties) had by autonomous beings, then it is hard to see how this
objection is supposed to work. If neither the children nor their parents have
a claim to the state provision of such benefits, then the persons’ general liberty
would seem to make the lack of educational opportunities just. The background notion here is that persons have a sphere of act options within which
they are not obligated to promote the projects or interests of others. Even if
an omission makes it more likely that injustices will occur in the future, this
does not by itself show that the omission is unjust. For example, consider my
refusal to buy a gun for my neighbor who is being pursued by an insanely
jealous ex-husband. This refusal is not unjust, even if it makes it considerably
more likely that her ex-husband will victimize her. The objector might also
assert that the state should provide children with educational benefits on the
basis that these benefits would be agreed to under a hypothetical contract
designed to track the demands of fairness or on the basis of their preinstitutional desert.53 If the former account is successful, then the children or their
parents would likely have a claim to the educational benefits, and hence the
objection does not get off the ground.54 The desert-based account may be
capable of supporting the objection, but only at the expense of making justice
part of a theory of goodness rather than rightness, and this conflicts with the
notion that the demands of justice consist of defeasible side constraints.
A related objection is that my argument may work for autonomy-based,
rights-oriented Millian liberalism, but it will not work for the sort of egalitarian liberalism found in the work of Ronald Dworkin and William
Kymlicka.55 On Dworkin’s account, the government must treat all those in its
charge as equals, that is, as entitled to its equal concern and respect. One
implication of this requirement is that the government cannot impose any sacrifice or constraint on any citizen in virtue of an argument that the citizen
could not accept without abandoning his sense of self-worth.56 This in turn
leads to the requirement that the government be neutral in matters of personal morality, that is, that persons should be free to live as they choose, so
long as they do not harm others.
This form of liberalism may prevent the sort of material deprivation that
would likely give rise to some slavery contracts, because the right to equal
concern and respect leads to a second requirement, which is that each person
be given an equal share of resources. However, some slavery contracts would
still occur in the absence of deprivation. For example, some contracts would
530
Stephen Kershnar
likely result from transactions between prisoners and either the victim or the
state (depending on which one has the right to punish), and others would
involve persons who wish to be slaves. The equality-of-resources requirement
would not eliminate the motivation for such contracts.
Egalitarian liberalism requires that the state permit and enforce these contracts. In these cases, the contractors are validly consenting to these contracts
and not harming third parties, hence the state’s interference would likely be
motivated by its view of the moral status of the transactions. However, egalitarian liberalism requires that the state be neutral toward matters of personal
morality and hence does not allow it to act on such a motivation. Another
way to see this is to notice that the state could not interfere with such voluntary arrangements without treating the contractors as incompetent, and
such treatment could not be accepted by a person with a proper sense of
self-worth. The Dworkinian liberal might object that state enforcement of the
contract involves the state imposing a burden that violates the equality of
concern or respect. Hence, on this objection, the state must permit, but may
not enforce, such contracts. This takes us to the last objection, which is that
a liberal state does not have an all-things-considered reason to enforce such
contracts.
E. The State Refusal to Enforce a Slavery Contract Does Not
Violate Liberal Principles
An objector might argue that my argument does not succeed in showing
that the liberal state has an all-things-considered duty to enforce some of the
slavery contracts in question and hence that I have not shown that on liberal
grounds such contracts should be legally enforceable. The objector begins by
noting that the above argument justifies the state’s duty to enforce the contract on the basis of the following premises. The first premise is that the prevention of unjust harm to others (plus maybe pejorative exploitation) is an
other-things-equal reason for state coercion. This is an implication of liberalism. The second premise is that with regard to slavery contracts that do not
involve unjust harm or pejorative exploitation, the failure of one contractor
to comply with the contract unjustly harms the other. The objector claims that
the premises do not support the conclusion because other reasons, including
liberal reasons, might override the other-things-equal reason for the state to
enforce the contract or require that a remedy be paid for its breach. More significantly, he claims that the state’s refusal to enforce a contract or collect a
remedy for its breach is not itself an unjust harm that the state has a duty,
under liberalism, to prevent.
Both of the objector’s claims should be rejected. If a contract does not
involve a right infringement or pejorative exploitation of either contractor or
any third party, then it creates rights in the contractors. On liberal grounds,
a right infringement that produces (or constitutes) harm, such as one that
accompanies a contract breach, provides a reason for state enforcement. And
where this reason is neither overridden nor undermined, it is an all-thingsconsidered reason for state coercion. Other conflicting duties (e.g., the effi-
A Liberal Argument for Slavery
531
cient use of judicial resources) may override the state’s duty to enforce the
contract, but these are often not liberal duties.
In addition, since contracts create rights in citizens, the state’s refusal to
enforce a contract, slavery or otherwise, is unjust except where the contract
involves an injustice or pejorative exploitation or where the contract itself fails
to meet certain conventional conditions (e.g., a writing, offer followed by an
acceptance, and consideration). If in the state of nature a citizen has a liberty
to enforce a contract or collect a remedy for its breach, and it seems that within
limits she does, then she can transfer this permission to her agent, which in
this case is the state. The state may, via its agency contract with the citizen,
incur an obligation to enforce contracts or collect remedies when they are
breached. In the absence of an overriding or undermining liberal reason, this
obligation is sufficient to warrant state protection. The injustice of the state
refusing to enforce a slavery contract is analogous to the injustice of the state
omitting to protect a prostitute from rape based on the disapproval of his or
her profession. Not every state failure to enforce a contract is a culpable one,
but an omission need not be culpable in order to be unjust. Nor does the
presence of certain conventional requirements for a contract undermine this
result, because contracts are justified in part by the obligations generated by
promises, and promises are conventional acts that must satisfy certain conventional rules.57
V. Conclusion
The slavery contract is not a right violation, since the right not to be
enslaved and the right not to give out a benefit are waivable, and the conjunction of their voluntary waiver is not itself a right violation. The case for
the contract being pejoratively exploitative is not clear. Hence given the
general presumption in favor of liberty of contract, such a transaction ought
to be permitted. The contract is also not invalid on the grounds that the
wrongdoer’s consent to it necessarily reflects involuntariness or irrationality
or on the grounds that it is logically impossible for the wrongdoer to satisfy
the contract. Allowing such contracts may not have desirable consequences,
but for the liberal, this by itself is not a sufficient reason against allowing
them. One case in which the slavery contract might be relevant is that in
which a victim’s agent or the state agrees to waive its claim to harsh punishment in return for the offender agreeing to be enslaved.
I am grateful to Michael Davis, Neil Feit, Michael Levin, and Dale Tuggy for their
helpful comments and criticisms of this paper.
Notes
1
Some liberals (e.g., Joel Feinberg) accept the offense principle (i.e., the prevention of
offense to others is a good reason for state coercion) and soft paternalism (i.e., the prevention of harm to self via substantially nonvoluntary conduct is a good reason for
state coercion) in addition to the harm principle. Joel Feinberg, Harm to Self (New York:
Oxford University Press, 1986), 14. My account of liberalism contains the soft paternalist principle but not the offense principle.
532
2
Stephen Kershnar
The idea for this account of slavery comes from R. M. Hare, “What Is Wrong with
Slavery?” in Applied Ethics, ed. Peter Singer (New York: Oxford University Press, 1986),
165–83, at 167–68. Hare distinguishes these two criteria, whereas I view the second as
resulting from the first. Also, I leave aside the tricky conceptual issue of whether temporary enslavement is possible, since my interest is in the rights of ownership over a
person that need not include within them all the rights that characterize the full liberal
concept of ownership.
3
The standard list of the rights that constitutes ownership occurs in A. M. Honore, “Ownership,” in Oxford Essays in Jurisprudence, ed. A. G. Guest (Oxford: Oxford University
Press, 1961), 107–47. On my account, an owner is the person who has the property
rights over a thing.
4
The exclusion relation is a structurally complex moral relation that often involves a claim
against others not to interfere with the possession or use of an object, a power over that
claim, and an absence of a time limitation on these elements. A person has a claim
against another if the second owes the first a duty. A power is the standing to waive
or demand the satisfaction of another Hohfeldian element (e.g., a claim). The standing
is moral or legal depending on the type of right that is involved. The elements of the
exclusion relation are usually held against all others. There are often other elements
that are part of this complex, such as a liberty (absence of a claim in others) to use the
property as one sees fit. Also, these structural complexes often include the right to use,
derive income from, transfer to another, and so on.
5
A claim is a duty owed by one person to another. A claim is often part of a right.
The notion that autonomy is the justification for justice relates to the notion that
justice involves two types of reasons for action or omission. One type of reason is concerned with the ways in which one person is permitted to treat another. Reasons of this
type are sometimes referred to as “deontological constraints.” A second type of reason
limits obligations that apply to a person by protecting a sphere of moral and political
freedom whereby a person may pursue her various projects, commitments, and personal relationships. Reasons of this type are sometimes (misleadingly) referred to as
“autonomous reasons.” This distinction comes from Thomas Nagel, “Autonomy and
Deontology,” in Consequentialism and Its Critics, ed. Samuel Scheffler (New York: Oxford
University Press, 1988), 142–72. Both of these types of reasons are probably grounded
in the value of autonomy, for it alone can provide a unified account of them while at
the same time accounting for the tendency of both types of reasons to permit actions
that do not maximize the good. The rival interest theory of rights asserts that a right
is a claim designed to protect the interest of the claim holder. For example, Joseph Raz,
“On the Nature of Rights,” Mind 93 (1984): 194–214, and “Legal Rights,” Oxford Journal
of Legal Studies 4 (1984): 1–21; Neil MacCormick, “Rights in Legislation,” in Law, Morality, and Society: Essays in Honour of H. L. A. Hart, ed. P. M. S. Hacker and J. Raz (Oxford:
Clarendon, 1977), 100–130. I shall assume but not defend the superiority of the will
theory. However, it should be noted that the interest theory would likely produce far
different results on this issue.
6
Hillel Steiner argues in support of the notion that exploitation involves a rights violation
between one transactor and a third party and hence can provide a reason for government coercion under classical liberalism. Hillel Steiner, “A Liberal Theory of Exploitation,” Ethics 94 (1984): 225–41.
7
Harry Frankfurt’s account is found in “Freedom of the Will and the Concept of a Person,”
in Free Will, ed. Gary Watson (New York: Oxford University Press, 1982), 81–95. The
importance of the person’s identifying with these higher-order desires is defended in
Harry Frankfurt, “Identification and Wholeheartedness,” in Responsibility, Character,
and the Emotions, ed. Ferdinand Schoeman (New York: Cambridge University Press,
1987), 27–45.
8
First-order desires trivially produce instrumental second-order desires (e.g., the desire to
eliminate competitor first-order desires). Dennis Loughrey, “Second-Order Desire
Accounts of Autonomy,” International Journal of Philosophical Studies 6 (1998): 216–21. By
“second-order desires,” I mean to focus on ones other than these trivial second-order
instrumental desires.
A Liberal Argument for Slavery
9
533
The model is a minor variant on Frankfurt’s theory, because his theory does not explicitly include a relation between beliefs parallel to that for different levels of desires, does
not explicitly require the capacity to form second-order volitions as a necessary condition for autonomy, and does not use a coherentist account to cut off the regress of
higher-order desires and beliefs.
10
This account of autonomy is quite common in the literature, for example, Gerald
Dworkin, The Theory and Practice of Autonomy (New York: Cambridge University Press,
1988), chap. 1.
11
A coherentist version of autonomy that is very similar to my own can be found in Laura
Waddell Ekstrom, “A Coherence Theory of Autonomy,” Philosophy and Phenomenological Research 53 (1993): 599–616. Robert Noggle points out that the coherentist account
has another advantage in that it can account for the centrality of some first-order beliefs
and desires to a person’s identity. This results because some first-order beliefs and
desires may occupy central roles in the coherent web of beliefs. Robert Noggle,
“Kantian Respect and Particular Persons,” Canadian Journal of Philosophy 29 (1999):
449–78, at 464–68.
12
The idea for this last point comes from Michael Davis.
13
The idea for this point and the two examples that follow it come from Joseph Raz, The
Morality of Freedom (New York: Clarendon Press, 1986), 373–77.
14
I am assuming here that beliefs can sometimes be indirectly chosen via the selection of
a search procedure. The idea for this point comes from Michael Levin. This is a controversial assumption, but a necessary one if autonomy is to be a large part of what
makes a person morally responsible for her actions.
15
This distinction between autonomy as capacity and autonomy as condition can be seen
in Feinberg, Harm to Self, 27–44.
16
This process whereby the agent forms herself by authorizing certain desires and beliefs
via other desires and beliefs (higher-order ones or ones that already form a coherent
desire/belief set) occurs gradually and must begin with some externally imposed
desires and beliefs (including some rules of rational reflection). Feinberg, Harm to Self,
33–35. Also, this process may take into account both empirical considerations (e.g., the
feelings of one’s spouse) and abstract moral considerations as well (e.g., the duty to
respect the dignity of others). Here I am explicitly rejecting the Kantian notion that
autonomy (and autonomous action) must involve self-legislation focusing on principles derived from a priori reason. The Kantian account leaves out too much in life that
seems important both for morality and for a rich personal life. This point comes from
Thomas May, “The Concept of Autonomy,” American Philosophical Quarterly 31 (1994):
133–44, at 138–39.
17
These intuitions can be seen in how we view certain movies, such as Coneheads (1993),
Star Wars (1977), and Babe (1995). In such movies we not only identify with but also
attach greater value to those creatures who have autonomy as a capacity.
18
The idea for this point comes from Thomas Hurka, “Why Value Autonomy?” Social Theory
and Practice 13 (1987): 361–82. Hurka then grounds the value of knowledge and freedom
of the will in the value of being properly connected to reality (372). Robert Nozick presents an intuitive case for this last value in his famous experience machine hypothetical. Robert Nozick, Anarchy, State, and Utopia (New York: Basic, 1974), 42–45.
19
I leave aside accounts, such as reliabilism, that deny that evidential justification is needed
for knowledge.
20
The move from something being a good for a person to its being an objective good is
defended in Thomas Nagel, The Possibility of Altruism (Princeton: Princeton University
Press, 1970), chaps. 10–12.
21
In the free-will literature, there is a debate as to whether responsibility requires that the
agent could have done otherwise and thus whether she had options. See, for example,
Harry Frankfurt, “Alternate Possibilities and Moral Responsibility,” Journal of Philosophy 66 (1969): 828–39. There is also an attack on the notion that autonomous acts need
involve choices. See, for example, Meir Dan-Cohen, “Conceptions of Choice and Conceptions of Autonomy,” Ethics 102 (1992): 221–43. I will assume but not defend the
claims that responsibility for the agent having done an action on his own requires that
534
Stephen Kershnar
he could have done otherwise and that autonomous acts must have a connection to
choices that occurred at some point in time.
22
Assuming that autonomy as a capacity is an intrinsic good, there is an interesting issue
as to whether act options, such as the option to be a slave, are an instrumental or contributory good, where the latter is a part of an intrinsically valuable thing. Assuming
the options presupposed by autonomy as a condition are only the extremely limited
ones that are necessary for self-shaping, further options (e.g., the option of slavery) are
instrumentally valuable.
23
Also, consideration of some options may close a person off from certain forms of good
that she might otherwise prefer, thus closing her off from some valued options. George
Sher cites cases in which consideration of some options prevents one from being able
to enjoy certain lifestyles. For example, a woman who gives serious consideration to a
feminist perspective may be unable to have a simpler and less self-conscious commitment to domestic life. George Sher, “Our Preferences, Ourselves,” Philosophy & Public
Affairs 12 (1982): 48–49. A nice account of different kinds of cases in which the having
of options interferes with the exercise of autonomy is found in Dworkin, The Theory
and Practice of Autonomy, chap. 5.
24
The idea for this point comes from Feinberg, Harm to Self, 386 n. 34. This point can also
be seen in Randy Barnett, “Contract Remedies and Inalienable Rights,” Social Philosophy & Policy 4 (1986): 179–202, at 186–88. Barnett uses this point to argue that the right
to control the future use of one’s person is inalienable (186–88, 194).
25
I owe this objection to Michael Levin. This point can also be seen in Arthur Kuflik, “The
Inalienability of Autonomy,” Philosophy & Public Affairs 13 (1984): 271–98, at 286–87.
26
The issue arises as to whether this waiver is morally wrongful. I shall assume that an act
is morally wrong only if it infringes a perfect or imperfect duty. Given the waiver of
the claim, it seems that a perfect duty has not been infringed. I leave open the issue of
whether it might infringe on an imperfect one, although it is hard to see how one might
have imperfect duties to oneself, since such duties would seem to have an identifiable
beneficiary and relatively specific content.
27
The Thirteenth Amendment provides for the punishment of slavery, for example,
“Neither slavery nor involuntary servitude, except as punishment for crime . . . , shall exist
within the United States . . .” (emphasis added). Also, the Eighth Amendment does not
permit disproportionately harsh punishments, for example, Rummell v. Estelle, 445 U.S.
263, 281–84 (1980), but neither the text nor the case law suggests that a disproportionately light punishment is unconstitutional. See, for example, “Note: Disproportionality in Sentences of Imprisonment,” Columbia Law Review 79 (1979): 1119–67. Since the
Constitution allows states to impose slavery as a punishment and since they may
impose a disproportionately light punishment, it is hard to see why a system that in
some cases allows criminals to choose the disproportionately light punishment of
slavery would be unconstitutional. Although this is less clear, I would argue that the
role of the victim or her agent does not make it so arbitrary as to be unconstitutional.
This is in part because the lesser punishment relates to an important factor, namely,
whether a victim’s agent relinquishes her claim to have the victim’s murderer killed
and thus does not arbitrarily impose the sentence. The relevance of arbitrariness, wantonness, and similar factors in the context of punishment to the Supreme Court’s Eighth
Amendment analysis can be seen in Jurek v. Texas, 428 U.S. 262, 276 (1976).
28
For an autonomy-based argument in support of the right to free speech, see T. Scanlon,
“A Theory of Freedom of Expression,” in The Philosophy of Law, ed. R. M. Dworkin (New
York: Oxford University Press, 1977), 153–72.
29
For an autonomy-based argument that there are rights to do wrong, see Jeremy Waldron,
“A Right to Do Wrong,” Ethics 92 (1981): 21–39.
30
Michael Kramer makes a similar point where he sets out two ways in which abstract
rights can generate concrete rights. Michael Kramer, “Rights without Trimmings,” in
A Debate about Rights, ed. Matthew Kramer, N. E. Simmonds, and Hillel Steiner (Oxford:
Clarendon, 1998), 7–111, at 43–44. It may be the case that right generation consists
purely of entailment relations but that interpretative generation occurs as a result of
epistemic difficulties. I leave this digressive issue aside.
A Liberal Argument for Slavery
31
535
Ibid.
The inclusion relation between rights in the context of blackmail is mentioned in Joel
Feinberg, Harmless Wrongdoing (New York: Oxford University Press, 1988), 238–76.
33
Feinberg, Harm to Self, 80.
34
There are different conceptions of coercion and threats as they relate to a contract. One
conception is a moralized notion that views these entities as involving the other contractor acting in an immoral (or perhaps unjust) manner. For example, Robert Nozick,
“Coercion,” in Philosophy, Science, and Method, ed. Sidney Morgenbesser (New York: St.
Martin’s Press, 1969), 447–53. The other is independent of the other contractor and
instead looks at the reasonableness of the alternatives for the contractor for whom a
proposal might be coercive or a threat. A third uses different approaches for each of
the two notions. I favor the first account but leave aside this digressive issue.
35
The notion that hard choices do not necessarily invalidate consent is nicely set out in
Jeffrey G. Murphy, “Consent, Coercion, and Hard Choices,” Virginia Law Review 67
(1981): 79–95.
36
The ideas in this section come from Feinberg, Harm to Self, 75, 79–81.
37
Ibid., 75.
38
For a more in-depth defense of this claim, see Scanlon, “A Theory of Freedom of Expression,” 153–71.
39
The concern for socially useful attitudes is put forth as a rationale for antislavery laws in
Feinberg, Harm to Self, 80–81.
40
I owe this objection to Neil Feit.
41
The idea for this example comes from Daniel Lyons, “Welcome Threats and Coercive
Offers,” Philosophy 50 (1975): 425–36, at 430.
42
Given this account and given that Helen benefits more than the rich man in the above
case, this would seem to entail that the rich man does not act wrongly toward Helen
(leaving aside any duty not to do things likely to insult another). This does not necessarily follow, since the rich man might violate an imperfect duty via his action.
However, the case can be modified so as to screen out such a violation. I would suggest
that the rich man does act in a permissible manner. Our intuitions to the contrary
weaken considerably if we consider cases in which Helen takes the initiative in offering sex for the money. Since it is morally arbitrary who makes the offer (again leaving
aside the duty not to insult another) and since the rich man has no ex ante perfect duty
to benefit Helen, our intuitions here are mistaken.
43
Alan Wertheimer, Exploitation (Princeton: Princeton University Press, 1996), 64.
44
This account comes from Wertheimer, Exploitation, 20–22, 211.
45
Derek Parfit, Reasons and Persons (Oxford: Clarendon, 1984), 493–502.
46
The victim’s right to execute her murderer is consistent with the will theory if one
assumes that autonomous choices include posthumous events and include a right to
punishment (although the power must be held by the victim during her lifetime or by
her agent).
47
John Stuart Mill, On Liberty, ed. Elizabeth Rapaport (Indianapolis: Hackett, 1978), chap.
V, sec. 11, pp. 101–102. Mill states: “But by selling himself for a slave, he abdicates his
liberty. . . . He therefore defeats, in his own case, the very purpose which is the justification
of allowing him to dispose himself. . . . The principle of freedom cannot require that he
should be free not to be free” (emphasis added). My interpretation is a controversial
one. A helpful discussion of the specific interpretation of Mill occurs in Richard J.
Arneson, “Mill versus Paternalism,” Ethics 90 (1980): 470–89; Feinberg, Harm to Self,
75–78.
48
Jean-Jacques Rousseau, The Social Contract, trans. Maurice Cranston (New York: Penguin,
1968), 55.
49
Kuflik, “The Inalienability of Autonomy,” 284–96.
50
Benedict De Spinoza, Theologico-Political Treatise (New York: Dover, 1951), chaps. 17 and
20. The idea for this objection also comes in part from Steiner, “A Liberal Theory of
Exploitation,” 234 n. 7.
51
Other arguments worthy of note are John Locke’s argument that God has given human
beings a claim to self-determination, but not the power to alienate it, and hence a
32
536
Stephen Kershnar
slavery contract would involve humans alienating claims in a way that exceeds their
legitimate powers. John Locke, “Second Treatise,” in Two Treatises of Government, ed.
Peter Laslett, 2nd ed. (New York: Cambridge University Press, 1960), secs. 22 and 23,
pp. 324–25. However, in the absence of an argument as to why God did not give us the
power of alienation, Locke’s argument is unsuccessful. Immanuel Kant argues that
the slavery contract would result in a loss of contractual powers, thereby invalidating
the contract itself. Immanuel Kant, The Metaphysics of Morals, trans. Mary Gregor (New
York: Cambridge University Press, 1991), sec. 49, pp. 139–40. Kant’s argument fails
since it does not take into account the differences in rights held before and after such
a contract. These responses come from Kuflik, “The Inalienability of Autonomy,”
279–80, 282–83.
52
Barnett, “Contract Remedies and Inalienable Rights,” 188–89.
53
The idea for these accounts of justice comes from Fred Feldman, “Adjusting Utility for
Justice: A Consequentialist Reply to the Objection from Justice,” Philosophy and Phenomenological Research 55 (1995): 567–85, and John Rawls, A Theory of Justice (Cambridge:
Belknap Press of Harvard University Press, 1971).
54
The objector might claim here that the child or parents have a conditional right to educational support (e.g., if the parents can’t provide it, the taxpayers have a duty to pay
for a child’s education). The problem with such conditional rights is that they are probably better understood as a context-specific application of a right to a certain condition
(e.g., the right to an education). If this is correct, then it seems the child has the right
to the underlying benefit.
55
Ronald Dworkin, “Liberalism,” in Liberalism and Its Critics, ed. Michael J. Sandel (New
York: New York University Press, 1984), 60–79, and “Neutrality, Equality, and Liberalism,” in Liberalism Reconsidered, ed. Douglas MacLean and Claudia Mills (Totowa, NJ:
Rowman & Allanheld, 1983), 1–11; Will Kymlicka, Liberalism, Community and Culture
(Oxford: Clarendon, 1989).
56
Dworkin, “Neutrality, Equality, and Liberalism,” 3.
57
An argument for the view that that the moral force of contracts rests on the role of
promises and that this view fits with much of the contract law in the United States is
set out in Charles Fried, Contract as Promise (Cambridge: Harvard University Press,
1981), chaps. 1–4.