INTRODUCTION
1. The issue
It often happens that a
philosopher challenges a commonsense belief. One commonsense belief that some
philosophers have recently challenged is the belief that there are constraints.
A constraint is a particular kind of moral restriction on how we may treat
persons.[1] There is a wide range of constraints within commonsense morality.
One may not break a promise nor lie simply because doing so will bring about a
greater good. Nor may one physically harm, coerce or kill an innocent even if
great good would follow from doing so.
Utilitarians and other
consequentialists have always been critical of constraints as an element in a
basic moral theory. Consequentialists believe that persons ought to bring about
as much good as they possibly can. Bringing about the best outcome may involve
doing what some non-consequentialists will see as violating a constraint.
Consequentialists may
think that outcomes will be better in the long run if people do not always aim
at bringing about the best outcome. Given various human characteristics it may
be that the outcome will be better in the long run if we train ourselves to
observe constraints rather than to harm others whenever doing so seems to us to
bring about the best outcome. It may even be that the outcome will be better in
the long run if we believe that there are moral constraints on action. If so,
then we ought (morally), according to consequentialists, to cause ourselves to
believe in constraints (supposing that to be possible). These sorts of indirect
justifications of constraints will not occupy me here.
As the recent debate
about constraints has shown, one need not be sceptical about constraints
because of one's adherence to a particular moral theory.
Some philosophers (notably
Samuel Scheffler) have argued that constraints appear paradoxical: it seems
that a moral system that includes constraints tells us not to bring about the
outcome which a such moral system presupposes is the best outcome.
Others (notably Jonathan
Bennett) have argued that when one analyzes the non-moral distinctions on which
constraints rest, it becomes apparent that these distinctions have no moral
significance.
However, not all
philosophers reject our commonsense belief that there are constraints. In this
thesis I shall discuss recent attempts by Warren Quinn and Frances Myrna Kamm
to justify constraints. In so far as my thesis suggests an overall verdict
about the justifiability of constraints, it is one of scepticism.
2. Two important distinctions
A large part of the
debate about constraints has been about the exact nature of and the moral
significance of two distinctions: the distinction between doing and allowing
harm and the distinction between intending and merely foreseeing harm. Below I
illustrate these two distinctions without going into any deeper analysis of
them.
Let us, following Kagan,
say that I countenance harm to someone whenever they are harmed and I
could have prevented that from happening (Kagan 1989,84).[2] I can countenance harm to others in (at least [3]) four different ways.
First, compare a
case in which I drown Pierre by holding his head under the water with a case in
which I stand idly by watching Theodor drown. It seems natural to say that the
first case is a case of doing harm, while the latter case is a case of allowing
harm. The morally salient feature in the first case is something I do,
holding Pierre's head under the water, while the morally salient feature in the
latter case is that I do not do something that I could do, namely rescue
Theodor.
Second, compare
the case in which I let Wilhelm drown because I want to inherit his fortune
with the case in which I let Ferdinand drown because my clothes would get wet
if I tried to save him. It seems natural to say that the first case is a case
of intentionally countenancing harm to someone, whereas the latter case
is a case of merely foreseeing that one will countenance harm to
someone.[4] In the first case it is the fact that if I do not save Wilhelm he
will drown that motivates me not to save him. What motivates me not to save
Ferdinand is not that he will drown if I do not save him.[5]
The distinctions between
doing and allowing harm and intending and foreseeing harm are not
coextensional. I can harm someone without harming him intentionally (e.g. I
hold Pierre's head under the water, because I will suffer excruciating pain if
I move). And, as illustrated above (Wilhelm's case), I can intentionally allow
someone to be harmed.
When moralists defend
constraints, they typically defend either a constraint against doing harm (the
Doctrine of Doing and Allowing - henceforth the DDA: e.g. Nagel 1979), a
constraint against intending harm (the Doctrine of the Double Effect -
henceforth the DDE [6]: e.g. Nagel 1986), or both constraints (e.g. Foot 1985,26). Both
the DDA and the DDE have some intuitive support.
One issue that is often
discussed in relation to constraints is whether constraints have thresholds.
Most recent defenders of constraints, Quinn and Kamm included, believe that
constraints have thresholds. That is, they believe that, although it is harder,
say, to justify killing an innocent than to justify allowing an innocent to
die, killing an innocent is permissible when the consequences of not killing
the innocent are sufficiently bad (e.g. it is impermissible to harm one to save
five, but permissible to harm one to save five thousand). The various arguments
for constraints that I discuss in this thesis are all arguments intended as
justifications of threshold (as opposed to absolute) constraints.
The DDA
(understood as a threshold constraint) says that doing harm is harder to
justify than allowing harm. The DDE (also understood as a threshold
constraint) says that intending harm is harder to justify than foreseeing harm.
Someone who accepts threshold versions of the DDA and the DDE believes that
doing harm requires, so to speak, more offsetting benefit than allowing harm in
order to be justified and that intending harm requires more offsetting benefit
than merely foreseeing harm in order to be justified. For example, the defender
of a threshold constraint might think that not spending £1000 to save Alan from
being hit by a car is permissible, while not spending £1000 to avoid hitting
Alan with one's car is impermissible.
Now, defenders of
constraints typically believe that doing and(/or) intending harm is harder to
justify than allowing and(/or) merely foreseeing harm. However, one can accept
this claim and yet not hold a position that one would be inclined to call a
"deontological position". Why is that so?
I shall answer this
question in a moment, but to do so I first need to introduce two distinctions.
First, I shall
distinguish between situations in which an agent can reduce harm and situations
in which the agent cannot reduce harm, but can determine who suffers the harm.
I shall call the former kind of case "cases involving the possible
reduction of harm" and the latter kind of case "cases involving the
possible redistribution of harm".
Second, I shall
distinguish between harm that is caused by the action of a human agent and harm
that is not caused by the action of a human agent but by natural causes. I
shall call the former kind of harm "human harm" and the latter kind
of harm "natural harm".
Consider now a moral
theory, T, according to which:
- It is permissible in cases involving the possible reduction of
harm to conduct oneself in whatever way will minimize harm. In such cases
it is, for instance, permissible to kill some to save more from death.
- It is permissible in cases involving the possible redistribution
of harm to redistribute (but not increase) harm in whatever way one
pleases, except that it is
impermissible to harm some to save the same number of people from (the same
amount of) natural harm. Thus, it is impermissible, according to T, to, say,
kill one to save another from death from natural causes. However, it is
permissible, according to T, to, say, kill one to save another from being
killed by another agent.
Now, according to T doing harm is harder per se to justify than allowing harm.
This is so, because, according to T, doing harm is sometimes harder to justify
than allowing harm [7] and allowing harm is never harder to justify than doing harm.[8]
Defenders of a threshold
version of the DDA obviously would reject T even though T does hold doing harm
to be harder per se to justify than allowing harm. They would reject T, because
T is incompatible with a number of other moral beliefs typically held by defenders
of constraints. Defenders of constraints, for example, typically do not believe
that it is (typically) permissible to harm some to prevent even more cases of
doing harm.[9]
It seems clear then that
defenders of constraints do not simply believe that doing and intending harm
are harder to justify than allowing and foreseeing harm (call the claim that
doing and intending harm are harder to justify than allowing and
foreseeing harm respectively THE BASIC DEONTOLOGICAL CLAIM).[10] Typically, they also make three additional claims.
First, for doing
and intending harm to be permissible quite substantial offsetting benefits are
needed (killing one or intentionally countenancing the death of one to save
five is, for example, considered impermissible - at least in standard cases -
by defenders of the DDA and the DDE). Call this claim THE CLAIM ABOUT
SUBSTANTIAL OFFSETTING BENEFITS. I shall not say anything about precisely
how many offsetting benefits are needed for doing and intentionally
countenancing harm to be permissible. I shall assume that defenders of the DDA
and the DDE more or less accept moral intuitions about the required amount of
offsetting benefits.
Second, defenders
of the DDA and the DDE believe that there are no special reasons that
favour doing and intending good that counterbalance the special reasons that
discriminate against doing and intending harm. If there were such reasons, then
it might after all be justified to, say, harm one to save two from being
violated. The reason that disfavours harming the one in this case would be
counterbalanced by the reason that favours saving the two. So it seems that one
should maximize the number of people not harmed. Call the claim that there are
no special reasons that favour doing and intending good THE SUPPLEMENTARY
DEONTOLOGICAL CLAIM. Someone who affirms THE BASIC DEONTOLOGICAL CLAIM and
THE CLAIM ABOUT SUBSTANTIAL OFFSETTING BENEFITS, but denies THE SUPPLEMENTARY
DEONTOLOGICAL CLAIM might hold that it is permissible not to pay £1000 to save
someone from drowning, impermissible to drown someone to save £1000, but
permissible to drown someone to save two from drowning.
Three, the fact
that a violation of the DDA minimizes the total number of violations of the DDA
does not make that violation easier to justify, let alone permissible. Nor does
the fact that a violation of the DDE minimizes the total number of violations
of the DDE make that violation easier to justify, let alone permissible. Call
these claims THE CLAIM ABOUT THE IMPERMISSIBILITY OF MINIMIZING VIOLATIONS.
A defender of a threshold constraint may, of course, think that a violation
that minimizes the total number of violations is permissible because it
produces substantial offsetting benefits in terms of avoiding harm.
These four claims are
logically independent of one another. One might consistently affirm THE BASIC
and THE SUPPLEMENTARY DEONTOLOGICAL CLAIMS and yet, say, deny THE CLAIM ABOUT
THE IMPERMISSIBILITY OF MINIMIZING VIOLATIONS. As a matter of fact most
deontologists who accept THE BASIC DEONTOLOGICAL CLAIM also accept the three
other claims. At any rate the
deontological position the tenability of which I explore in this thesis is one
which consists in the affirmation of all four claims. Note that this means
that I intend to explore the tenability of a deontological position that
includes an affirmation of the DDA as well as an affirmation of the DDE. I
shall call the one who affirms this position THE DEONTOLOGIST.
I discuss THE BASIC
DEONTOLOGICAL CLAIM primarily in chapters 1-2. THE SUPPLEMENTARY DEONTOLOGICAL
CLAIM is discussed in chapters 3-4. THE CLAIM ABOUT SUBSTANTIAL OFFSETTING
BENEFITS is discussed in chapters 5-6, while THE CLAIM ABOUT THE
IMPERMISSIBILITY OF MINIMIZING VIOLATIONS is discussed in chapter 7.
In addition to a
discussion of these four claims I shall also discuss in chapter 8 how, if at
all, the justification of constraints is connected with the justification of
other elements in a moral scheme, namely options and certain claims about
distributive justice.
1
DOING HARM
The first challenge that
THE DEONTOLOGIST has to face is this: what morally significant properties do
cases of doing and intending harm have, which cases of allowing and foreseeing
harm do not have, that make doing and intending harm harder to justify than
allowing and foreseeing harm? To answer this question THE DEONTOLOGIST needs to
provide a clear analysis of the distinctions between doing and allowing and
between intending and foreseeing.[11]
Suppose that there are
no morally relevant properties that distinguish, say, doing harm from allowing
harm. In that case the reason that disfavours doing harm would be no stronger
than the reason that disfavours allowing harm. One would then, say, have no
more reason not to kill someone by depriving him of his food than one has not
to allow someone to die of hunger by failing to assist him regain his food.
This implication would leave someone who initially defended the DDA with the
following choice. He might say that because harming others is very hard to
justify and because the reason against allowing others to be harmed is as
strong as the reason that disfavours doing harm, it is much harder to justify
allowing others to be harmed than we tend to believe. Or he might say that
because allowing harm is relatively easy to justify and because the reason
against doing harm is no stronger than the reason that disfavours allowing
harm, doing harm is much easier to justify than we think. Neither option is
acceptable to the defender of the DDA.[12]
It is one thing to identify a property that cases of doing or
intending harm have that cases of allowing or foreseeing harm do not have and
which feels morally significant. It is another thing to give an account of why
this property is morally significant. To provide an argument (other than the
fact that the distinction feels morally significant) for the moral significance
of the property in question, THE DEONTOLOGIST must give an account of why the
properties that doing/intending harm have and that allowing/foreseeing harm do
not have are morally significant. This task is the task of providing a
rationale for the DDA and the DDE.
Should we expect the DDA
to have the same rationale as the DDE? It seems not. The fact that we can
intentionally allow someone to be harmed and the fact that we can harm someone
unintentionally suggests that whatever properties cases of doing harm have that
make them especially morally displeasing, there are cases of intending harm
that do not have these properties (and vice versa).[13]
I discuss the DDE in the
next chapter. In this chapter I discuss Quinn's and Kamm's recent analyses of
the doing-allowing distinction and its moral significance.
Quinn attempts to
justify a version of the DDA based on a distinction that is closely related to
the doing-allowing distinction (namely the distinction between what Quinn calls
"negative" agency and "positive" agency). Quinn argues that
a morality that includes this version of the DDA appropriately recognizes us as
independent beings. I argue that moralities that do not include Quinn's or
indeed any other version of the DDA may also recognize us as independent beings
and that Quinn's attempt to justify the DDA, thus, fails.
In the second section of
this chapter I discuss Kamm's defence of the DDA. Central to Kamm's analysis of
the doing-allowing distinction is the concept of interference. Basically, she
claims that doing harm may or may not involve interfering with one's victim
whereas allowing harm necessarily does not involve interfering with one's
victim. I argue that her account of the distinction between doing and allowing
in terms of interference fails to classify cases of countenancing harm in the
way it seems that they should be classified. There are cases of allowing harm which
involve interfering with the victim. Bracketing this problem, I then go on to
discuss Kamm's claim that allowing harm is easier to justify than doing harm
per se because the former does not involve interfering with one's victim. I
defend this account of the DDA against Shelly Kagan's claim that it is
circular. Finally, I suggest that Kamm's account of the DDA may be problematic
for other reasons.
1.1. QUINN ON THE DDA
1.1.1. Direct harmful agency
1. The DDA discriminates
in favour of one kind of agency and discriminates against another kind of
agency. Quinn calls the former kind of agency negative and the latter kind
positive agency. What exactly, then, distinguishes negative and positive
agency? That is, what is the non-moral content of the DDA?
One way to interpret the
non-moral basis of the DDA would be to distinguish between harm that comes from
action and harm that comes from inaction. What does Quinn mean by the
distinction between action and inaction?
Quinn basically accepts
the Anscombe/Davidson view on these matters. To Anscombe/Davidson an action
(unlike an inaction) is a concrete particular, an event, that is intentional
under some description.
If one accepts this view
on the metaphysics of action, one can give the following account of the
distinction between harmful positive and negative agency: positive agency is
agency in which someone's being harmed is most directly explained by the
existence of an action performed by the agent (e.g. "Bill died because
John gave him a lethal injection"). Harmful negative agency, on the other
hand, is agency in which someone's being harmed is most directly explained by
the non-instantiation of some kind of action that the agent had the power to
perform (e.g. "Bill died because John did not inject the vaccine").
However, Quinn is not
satisfied with an account of the non-moral content of the DDA simply in terms
of the distinction between harm that is most directly explained by an action
and harm that is most directly explained by an inaction. He thinks that the DDA
so construed fails to capture certain moral intuitions that it ought to
capture. Consider the following case:
(RESCUE) "We
are off by special train to save five who are in imminent danger of death...
Since the train is on automatic control you need do nothing to keep it going.
But you can stop it by putting on the brakes. You suddenly see someone trapped
ahead on the track. Unless you act he will be killed. But if you stop, and then
free the man, the rescue mission will be aborted" and the five will die
(Quinn 1989a,298).
If the DDA simply
discriminates against conducting oneself in such a way that one of one's
actions is the most direct explanation of the harmful upshot, then your not
stopping the train would be a case of harmful negative agency. The death of the
trapped person would be most directly explained not by something you did
(pouring a drink), but by an inaction of yours (not applying the brakes).
However, according to Quinn your inaction feels like a case of positive agency.
This is why Quinn suggests we substitute a more complex distinction for the
simple distinction between harm that is most directly explained by an action
and harm that is most directly explained by an inaction as the basis of the
DDA.
In response to cases
like RESCUE Quinn distinguishes between two different variables. The first
is whether the harmful upshot is most directly explained by what the agent does
or by what the agent does not do. The second variable is whether or not
the harmful upshot is most directly explained by an action of an object that
the agent intends the object to perform.
An action qualifies as a
case of harmful positive agency, if a harmful upshot of someone's agency is
most directly explained by something he does (rather than by something he does
not do) (positive agency by action), or if the harmful upshot is most
directly explained by an action of an object that the agent controls [14] and that the agent intends the object to perform (positive
agency by inaction). An action qualifies as a case of harmful negative
agency if neither of these two conditions is fulfilled.
What does Quinn mean by
"the contribution that most directly explains the harm"? A
contribution, i.e. someone's action or inaction, "explains harm more
directly than another if the explanatory value of the second is exhausted in
the way it explains the first" (Quinn 1989a,301). What does it mean that
the explanatory value of one contribution exhausts the explanatory value of
another?
I believe that what
Quinn has in mind is something like the following. Take a case in which I could
save A from starvation by donating a sum of money to Oxfam but decide to buy a
Walkman. Clearly, Quinn would want to say that the explanatory value of my
buying the Walkman as regards A's death is exhausted in the way it explains my
not donating the money to Oxfam and, thus, that my not donating the money to
Oxfam is the contribution of mine which most directly explains A's death.
It seems plausible to
say that the explanatory value of the fact about my behaviour, that I did not
donate the money, is not exhausted by the fact about my behaviour, that I
bought a Walkman, since it might have been true that I did not donate the
money, false that I bought a Walkman and true that A died of starvation (I
spent the money in a restaurant).
Need one's most direct
contribution to a certain outcome be either an action or an inaction or can it
be both? If I give someone a poisonous injection with my right hand and do not
simultaneously give him the antidote with my left hand, then neither fact about
my behaviour explains the death of the victim more directly than the other. We
cannot infer from the fact that I do not give my victim the antidote with my
left hand and the environmental facts [15] about the situation, that I give him a poisonous injection with
my right hand. Nor can we infer from the fact that I give the victim a
poisonous injection with my right hand and from the environmental facts about
the situation, that I fail to give him the antidote with my left hand. I think
that this case shows that an agent's action and inaction may be his equally
direct contributions to the same harmful upshot and, thus, that an agent's
relation to a harmful upshot may at the same time be one of positive and
negative agency. It is not as if each time an agent's behaviour is causally
relevant to a harmful upshot, then he exercises either positive harmful agency
or negative harmful agency but not both. This claim does not contradict
anything Quinn says, but, as we shall see, it causes problems for him later on.
I think two questions
immediately arise about Quinn's motivation of the more complex interpretation
of the DDA. First, is
it a good motivation for dropping the simple distinction that cases like RESCUE
"feel like" cases of positive agency [16]? Why not base the DDA on the simple distinction between action
and inaction and then take account of the moral intuitions about cases like
RESCUE by another moral doctrine that accounts for why someone's rights to be
saved are harder to defeat in certain contexts than others' rights to be saved?
Ultimately, the answers to these questions must depend on whether the rationale
for the moral significance of the DDA applies to cases involving positive
agency by action as well as cases involving positive agency by inaction. And as
I said above I argue in the following section that Quinn's motivation of the
DDA does not select either the simple or the complex distinction between
positive and negative agency as the favoured interpretation of the non-moral
basis of the DDA.
Second, even if
it were sufficient motivation for adopting Quinn's complex distinction that
cases like RESCUE feel like positive agency Kamm has persuasively argued that
cases like RESCUE are not in fact morally equivalent to cases involving
positive agency by action (see 5.3.1.).
1.1.2. The moral significance of the distinction
between negative and
positive agency
1. This was Quinn's
account of the non-moral basis of the DDA. Why does Quinn think that the
distinction between positive and negative agency is morally significant?
Quinn distinguishes with
Foot between negative and positive rights. Negative rights (such as the
right not to be killed) proscribe harmful positive agency. Positive rights
(such as the right to be saved from death) proscribe harmful negative agency
(Quinn 1989a,306). The fact that the distinction between negative and positive
agency is morally significant follows from the fact that negative rights are, ceteris paribus, less easily defeated
than positive rights (Quinn 1989a,289-290). That is, more "offsetting
benefit" is needed to justify positive harmful agency than is needed to
justify negative harmful agency. Thus, in order to justify the DDA, one need
not, Quinn contends, argue that "some intrinsic moral disvalue attaches
to" positive harmful agency but not to negative harmful agency.
Does this mean that
Quinn believes that one may affirm the DDA and affirm that killing someone because,
say, avoiding killing would cost one £1000 is intrinsically no worse than
allowing someone to die because saving that person would cost one £1000?
Presumably not, since Quinn would say, I suspect, that the former involves a
violation of the victim's rights whereas the latter does not and a violation of
a right is not only prima facie
impermissible but also morally bad because impermissible.
But if Quinn does not
believe that negative rights are less easily defeated than positive rights
because doing harm is intrinsically worse than allowing harm, why then are
negative rights harder to override than positive rights? Why should it not be
the other way round? Why should either right take precedence over the other? I
shall now present Quinn's argument for why negative rights take precedence over
positive rights. Quinn's argument is based on considerations about the kind of
moral status that different systems of morality grant us. In this respect,
Quinn's argument for the DDA is similar to an argument that Kamm has put
forward (initially) to account for the impermissibility of minimizing
violations (see 3.2 and 7.3).
Suppose neither negative
nor positive rights took precedence over the other. That is, whenever persons'
rights conflict, the rights protecting the greater good should, ceteris paribus, prevail. In that case
morality denies persons a "primary say over what may be done to them. It
does so because if we may rightly injure or kill a person whenever others stand
to gain more than he stands to lose, then his body... is not in any interesting
sense his" (Quinn 1989a,308). A person's body and mind are, morally
speaking, his own only if his say over what may be done to them (and thereby to
him) can override the greater needs of others.
By denying persons a
primary say over what should be done to them morality fails to recognize
persons as independent beings with ends of their own (Quinn 1989a,309). But it
is fitting that morality should recognize us as independent beings. One of the
basic social functions of morality is "to specify the mutual authority and
respect that are the basic terms of voluntary human association" (Quinn
1989a,312). By reflecting on the status that different moral systems would
grant us we throw light on the "(timeless) content of morality"
(Quinn 1989a,310fn).
Quinn's point is not
that it is good for us to believe (whether we are mistaken or not) in a moral
system that recognizes us as independent beings, but that morality in fact
recognizes us as independent beings and that this is valuable in itself. The
value that is realized through morality's recognition of us as independent
beings is a value that is realized through the design of the system of morality
that recognizes us as such, not a value that we can promote through our
actions.
Thus, if morality
recognizes persons as independent beings then it must grant persons a say over
what should happen to them that can override the greater needs of others. But
this in itself does not establish that negative rights take precedence over
positive rights. Persons have a primary say over what should happen to them if either
positive rights take precedence over negative rights, or negative rights
take precedence over positive rights. To see why both sets of norms each grants
persons a primary say over what should happen to them consider the following
case:
(SWIMMERS) You can
save two drowning swimmers only by running over and thereby killing one
bystander.
If negative rights take
precedence over positive rights we (may) get the familiar commonsense morality,
where the negative right of the bystander not to be killed takes precedence
over the swimmers' positive rights to be saved. Such morality recognizes
persons as independent beings, because it gives persons a say over whether they
should be killed that can outweigh the collectively greater needs of others.
Consider now a slightly
modified case:
(SWIMMER) You can
save a drowning swimmer only by running over and thereby killing two
bystanders.
If positive rights take
precedence over negative rights, then the drowning person's positive right to
be saved takes precedence over the two bystanders' negative rights not to be
run over and killed. Such morality grants persons a primary say over whether
they should be allowed to die that can outweigh the collectively greater needs
of others.
A system of rights where
positive rights took precedence over negative rights seems odd. But it does not
fail to recognize persons as having a primary say over what should happen to them.
That is, such a system would grant persons a primary say over what should
happen to them that can outweigh the collectively greater needs of others.
Those who can permissibly be killed in the attempt to save fewer persons (and
indeed anyone else) under a system of morality that gives positive rights
precedence are not denied a status as persons whose say over what should happen
to them can outweigh the collectively greater needs of others. More could
permissibly have been killed to save them. Thus, in order to account for the
precedence of negative rights Quinn needs more premises than simply the premise
that it is fitting that morality recognizes us as independent beings.
What then are these
additional premises? Quinn seems to suggest two different premises each of
which together with the premise that it is fitting that morality recognizes us
as independent beings should get us to the desired conclusion (i.e. the
precedence of negative rights).
First, he
suggests that a morality that allowed us to kill two in order to save one would
be "absurd" and "odious" (Quinn 1989a,307). It certainly
would. But this is beside the point here. What we want to know is why it is any
less absurd to allow two to die because saving them requires killing one than
to kill two in order to save one.
Second, Quinn
argues that "(a) morality that permitted us to run over and kill the two
in order to save the one seems not only odious but incoherent. For once we have
decided to kill the two, we have placed them in at least as much danger as the
one was in originally. And that would presumably activate their positive rights
to be saved from their predicament - rights that would collectively outweigh
the positive rights of the one who is drowning" (Quinn 1989a,307-308). The
incoherence, presumably, is that whose rights take precedence depends on what
the agent decides to do. Should he, having decided to save the one, reason as
Quinn suggests he must, he would now have reason to reverse his decision. But
once he has reversed his decision he deactivates the positive rights of the two
trapped on the track. The agent should now save the swimmer. This could go on
until circumstances determine what happens.
2. In essence, Quinn's
argument for the DDA is this. Morality recognizes us as independent beings.
Assume that persons have negative and positive rights.[17] Either one set of rights takes precedence over the other or
neither takes precedence over the other. If the latter is the case, then it is
simply "the balance of overall costs" that determines what an agent
ought to do, in which case morality does not recognize us as independent
beings. Thus, unless either negative rights take precedence over positive
rights or positive rights take precedence over negative rights, morality does
not recognize us as independent beings. The latter option is formally
incoherent. Thus, negative rights take precedence over positive rights.
I shall now raise six
objections to Quinn's argument for the moral significance of the distinction
between positive and negative agency. I shall argue that Quinn fails to show
that the precedence of positive rights is formally incoherent and that the
requirement that morality recognizes us as independent beings is too blunt an
instrument to justify Quinn's version of the DDA as opposed to other versions
of the DDA, or indeed to justify a moral system that includes some version of
the DDA at all.
First, even if a
morality where positive rights take precedence over negative rights were
incoherent for the reason Quinn gives, it would be quite easy to make it
coherent. That is, one could supplement the doctrine of precedence of positive
rights with the following clause:
(CONFLICT CLAUSE)
If A's positive rights to be saved take precedence over B's negative rights
not to be harmed before C decides whether he will harm B to save A or not harm
B and thereby allow A to be harmed, then A's positive rights take precedence
over those of B's positive rights that C's decision to harm B and thereby save
A activates.
The CONFLICT CLAUSE
ensures that my decision about whom to save does not alter what I should do. A
moral doctrine that held that positive rights took precedence over negative
rights (and the suggested CONFLICT CLAUSE) would give persons a primary say over
what should happen to them and would avoid the incoherence that Quinn believes
such a system would imply. Thus, Quinn has given no reason to reject a morality
that recognized us as independent beings by giving positive rights precedence
over negative rights, when that system incorporates the CONFLICT CLAUSE.
Second, even when
the conflict clause is not incorporated I fail to see why a doctrine of
precedence of positive rights should be incoherent in the way Quinn suggests. I
fail to see that because I fail to see why a decision to go ahead with the
rescue mission in SWIMMER activates the positive rights of the two bystanders
who are trapped on the road as Quinn argues it does.
Positive rights
"are claim rights to aid or support, and would therefore seem to proscribe
harmful negative agency" (Quinn 1989a,306). What negative agency, i.e.
inaction, would be proscribed in SWIMMER if positive rights took precedence
over negative rights? One might have thought that the proscribed inaction here
is, say, not applying the brakes or not removing one's foot from the gas pedal.
But not applying the brakes or not removing one's foot from the gas pedal in
SWIMMER would not count as harmful negative agency, according to Quinn's
definition. Neither would count as negative agency because I deliberately fail
to apply the brakes or remove my foot from the gas pedal to make or to allow
the car to pass over the spot where the two bystanders are trapped in order to
save the one from drowning. In short, I intend an action of the car that will
cause the death of the two and this, according to Quinn, makes my inaction
harmful positive agency. Thus, my decision to go ahead with the rescue mission
does not activate the positive rights of the trapped bystanders. Quinn, thus,
cannot use SWIMMER to argue that a morality that gives positive rights
precedence over negative rights is incoherent.
Here is an objection to
my argument. Quinn says that one's inaction is positive agency by inaction
"(i)n special circumstances ... where harm comes from an active object or
force ... . This ... happens just in case the object harms because the agent
deliberately fails to control it and he fails to control it because he wants
some action of the object that in fact leads to harm" (Quinn 1989a,301).
And we may charitably assume that Quinn meant to say that only in these
circumstances can one's inaction qualify as positive agency by inaction. But if
I am driving the car, then the car is not an "active object" and I do
not fail to control it. Thus, my not applying the brakes is not a case of
positive agency by inaction. Admittedly, if the car were driven by the
autopilot and I deliberately failed to apply the brakes, then my agency would
count as positive agency by inaction. And Quinn could not have used such
a case to illustrate the incoherence of a system of norms according to which
positive rights take precedence over negative rights.
My first (and
less important) reply to this objection is that Quinn's use of the phrase
"control" is unfortunate here, because what he thinks morally
significant in cases of positive agency by inaction is the sense in which I can
either intervene or not intervene to change the movements of the object (see
Quinn 1989a,300). In this sense I do not fail to control an object, when I do
not intervene to change its course of movements.
My second reply
is this. Quinn's reasons for counting deliberate inaction as positive agency in
cases where I need do nothing to bring about the desired outcome through the
action of an object I control is that these cases involve a "combination
of control and intention" that makes for a certain kind of complicity
(Quinn 1989a,300). But this combination exists also in cases where I need to do
something to make the object that I control perform the desired action. Since I
argued above that an agent's relation to a bad upshot may at the same time be
one of positive as well as negative agency, it would be ad hoc to say that my inaction does not count as positive agency by
inaction when the object whose action I intend is one where I need to do
something to make it move in the desired way.
My third
objection is this: suppose I am right that Quinn fails to produce an example in
which the precedence of positive rights generates incoherence. Could he produce
another example that would illustrate the incoherence of a system that gave
precedence to positive rights? He could if my decision to save someone from
their predicament (not brought about by me) can itself make the harmful outcome
in question come not from my positive agency but from my negative agency or at
least also from my negative agency. An action that before my decision to do p
would be an instance of harmful positive agency would after my decision to do p
be an instance of harmful negative agency.
Whether or not Quinn can
produce such a case, the problem for Quinn is that there are cases where the
harmful upshot would come from the agent's negative agency before the agent's
decision to do p and where the harmful upshot comes from the agent's positive
agency after the agent's decision to do p. Thus, there are cases where a system of norms that gives negative rights
precedence generates precisely the kind of incoherence that Quinn believes is
generated by a system of norms that gives positive rights precedence. Here
is one such case:
(MULTIPLE RESCUE)
I find myself on a train off on a rescue mission to save five drowning
swimmers. One person is trapped on the tracks on the way to the beach. On board
the train are three injured passengers who urgently need my assistance. I can
only save one of three different groups of potential victims. Whom should I
save?
Suppose negative rights
take precedence over positive rights as Quinn believes they do. All three
groups have a positive right that I save them, but there are five
drowning swimmers so their claims outweigh the three injured passengers' and
the single trapped person's claims. I decide to rescue the swimmers. But now I
intend the train to perform an action that harms the person who is trapped on
the tracks. This means, on Quinn's account, that his negative rights are
activated and, presumably, take precedence over the positive rights of the
swimmers. I should not go ahead with my decision. I then decide to save the
three injured passengers. Now I no longer intend that the train should pass
over the spot where the bystander is trapped and, thus, do not violate any of
his negative rights. But then it seems that I should change my mind again,
since the positive rights of the three injured passengers are outweighed by the
positive rights of the drowning swimmers. And so I continue deliberating.
I think MULTIPLE RESCUE
suggests the following: if whether a
certain right should figure in an agent's deliberation about what to do
(something I shall question just below) depends
on whether the right is activated (as Quinn's argument against the
coherence of the precedence of positive rights over negative rights
presupposes), then a system where
negative rights take precedence over positive rights is incoherent in the same
way as a system where positive rights take precedence over negative rights.
Fourth, in my
third objection I exploit the fact that the agent's intentions, according to
Quinn's complex distinction between negative and positive agency, play a role
in determining whether someone's involvement in a harmful upshot falls on the
positive or the negative side of the distinction. So perhaps one should see
this objection as a reason to drop Quinn's complex distinction and opt for the
simple articulation of the DDA. That is, one should simply distinguish between
harm that is most directly explained by an action and harm that is most
directly explained by an inaction. One might then go on to argue on this basis
that a system that gives positive rights precedence is incoherent.
I reject this attempt to
restate Quinn's argument for the DDA because I find it hard to understand what
Quinn means by "activating" someone's right. However, I shall assume
that he means at least this: when someone's rights are not activated they do
not count in deliberation about what to do, whereas if they are activated they
do count.
Consider now SWIMMER. We
can save the drowning person by running over and thereby killing the two that
are trapped on the road. Quinn believes that my decision to rescue the swimmer
activates the positive rights of the two trapped on the road to be saved from
their predicament, because my decision to carry out the rescue mission places
them in at least as much danger as the swimmer was in originally. It is crucial
for Quinn's argument which criterion he uses to determine how much danger a
person is in. I can see two different ways in which he can determine how much
danger a person is in, but I do not think either of them gives Quinn what his
argument requires.
First, how much danger a
certain person is in depends on how great the chance is that he will be harmed.
How great the chance is that he will be harmed depends among other things on
what the likelihood is that those who are in a position to save him decide to
save him (and how great the likelihood is that those who are in a position to
harm him decide to harm him). How much danger the swimmer was in before I
decided to rescue him depends (among other things) on the probability of my
choosing to rescue him, the probability of my changing my mind if I decide to
rescue him, as well as the probability of my actually saving the swimmer if I
try. For the two persons trapped on the road to be in at least as much danger
as the swimmer was in before I made up my mind to rescue him (he would drown if
I did not decide to rescue him), it would have to be the case that it was very
likely that I would decide not to rescue him. This would explain why the
positive rights of the persons trapped on the road, assuming that any such
rights are involved at all, were not active before I decided to rescue the
swimmer. Although they would almost certainly be killed if I decided to rescue
the swimmer, it was almost certain that I would not decide to do so.[18]
Suppose that I decide
against all odds to rescue the swimmer (as I ought to do given the precedence
of positive rights and given the minuscule ex
ante probability that I actually rescue the swimmer). It may now be
overwhelmingly probable that I will act according to my decision. In that case,
it seems that incoherence arises (in the way Quinn believes it does), because
now the persons trapped on the road are in great danger whereas the swimmer
will almost certainly be rescued. And that suggests that the positive rights of
the two are activated whereas the positive rights of the swimmer are
"de-activated".
I believe that although
this interpretation of how much danger one is in is certainly what interests
the potential victims, it cannot form the basis of the deliberation of the
agent. It is a mistake to think that whether or not positive rights (or
negative rights) are active depends on the likelihood of harm coming to the
victim, when the likelihood of harm coming to the victim depends on the
likelihood of the different decisions that the agent can make. In deciding whom
to save I cannot take the fact that I am very unlikely to decide to save this
person and that he is, thus, in great danger as a consideration that favours my
saving him. Since I am deliberating about what to do, I must regard it as up to
me whom I decide to save. I think we should take Quinn's argument to show, not
that a doctrine of precedence of positive rights is incoherent, but that one
should reject the belief that rights are activated or deactivated depending on
what one decides to do.
Note in this connection
that the same issue arises with regard to negative rights. That is, does one
have an active negative right not to be harmed if the person who may harm one
is very unlikely to decide to do so? If one thinks not, then I see no reason
why one should not find similar cases where one's decision to do p activates
someone's negative rights that being activated collectively outweigh others'
negative rights that were decisive before one's decision to do p. This would
generate a similar incoherence in a system where negative rights take
precedence.
So far I have argued
that Quinn cannot determine (for the purposes of applying the DDA) how much
danger someone is in simply by asking how likely it is that he will be harmed.
I shall now turn to the
second understanding of how much danger the recipients of one's agency are in.
According to this position how much danger the victim is in (for the purposes
of deliberation about what to do) depends on how likely it is that the victim
would be harmed if the agent, who is deliberating about what to do, had no
power to affect whether the victim is harmed or not. Here is how this
suggestion would work.
Suppose I am deliberating
about whether to run over the bystanders and rescue the swimmer. According to
this proposal the swimmer is in great danger because he would be very likely to
drown if I had no power to affect his situation. The persons trapped on the
road on the other hand are in no danger, since, ex hypothesi, if I had no power to affect their situation, they
would not be run over. Suppose next that I decide to rescue the swimmer and
switch on the car's autopilot. It is now true that if I had no power to
affect the situation of any of the parties involved the persons trapped on the
road are in as much danger as the swimmer (the autopilot will guide the car to
its destination). But that is true only because I earlier on had the power to
affect their situation and did something that would affect their situation. And
I see no reason why a defender of a system where positive rights take
precedence over negative rights should not deal with this situation by saying
that in such cases what matters is that initially only the swimmer would have
been harmed had the agent had no power to affect the situation of any of the
parties.[19] I should not change my decision about whom to save because I at
some later stage act in such a way that it then becomes true that who is in
danger and who is not as a result of my act changes. Adding this stipulation to
a system that gives positive rights precedence over negative rights saves the
system from incoherence.
If it is unacceptable to
revise a system that gives positive rights precedence over negative rights in
this way, then a defender of a system where negative rights take precedence
over positive rights is likely to be in as much trouble as the defender of a
system that gives positive rights precedence over negative rights. This is so
because the defender of the precedence of negative rights has to employ a
similar device to deal with cases in which an agent puts someone in a situation
where he can allow his victim to be harmed (say, I lead someone into the desert
and then allow him to die of thirst).
So far I have argued
that Quinn fails to show that a moral system where positive rights take
precedence is incoherent whereas one where negative rights take precedence is
not.
I shall now bracket this
objection in order to turn to two different kinds of criticism. First, I argue
that even if morality must recognize that persons have a primary say over what
should happen to them by giving negative rights precedence over positive rights
and, thus, include some version of the DDA, there are other articulations of
the DDA than Quinn's which satisfy Quinn's desideratum (my fifth objection).
Second, I argue that even if a morality in which positive rights take
precedence is incoherent it still would not follow that negative rights take
precedence since a system of morality that gives neither negative nor positive
rights precedence could recognize us as independent beings (my sixth
objection).
Fifth, even if
morality grants persons a primary say over what should happen to them by giving
precedence to negative rights over positive rights, this in itself is not a
sufficient reason to accept Quinn's articulation of the DDA (i.e. a version of
the DDA that discriminates against positive agency by action as well as
positive agency by inaction). There are alternative versions of the DDA that
grant us a primary say over what should happen to us. This, of course, does not
imply that there might not be other considerations than those relating to our
moral status as independent beings that motivate Quinn's version of the DDA but
not alternative versions of the DDA. However, Quinn does not discuss the issue
and, thus, does not present any such considerations.
Consider the following
competing version of the DDA, namely one that discriminates simply between harm
that comes from action and harm that comes from inaction. According to this
version the harmful agency involved in RESCUE would count as a case of harmful
negative agency (which it does not on Quinn's account). Quinn is certainly
right that intuitively it seems that I have done the wrong thing by letting the
train continue. But if the motivation for the DDA that Quinn provides is to
motivate his particular version of the DDA rather than the alternative under
discussion here, then the wrongness of my action should be explained by the
fact that no set of norms that recognizes the independence of persons would
allow such inaction. But an articulation of the DDA in terms of the simple
distinction between harm that comes from action and harm that comes from inaction
does entail that persons have a primary say over what should happen to them
that can outweigh others' collectively greater needs.
Quinn's version of the
DDA recognizes persons as having a primary say over what should happen to them.
There may be some respects, Quinn argues, in which it is more important to us
that we have a primary say: "We feel, I believe, most strongly about
assaults on our minds...The idea that against our will we could justifiably be
brainwashed or lobotomized in order to help others cuts deeply against our
sense of who and what we are" (Quinn 1989a,310). Might Quinn not defend
his particular version of the DDA by saying that it corresponds to our beliefs
about "how important the relevant forms of legitimate control are to us -
the extent to which we wish to belong, in the sense under discussion, to
ourselves" (Quinn 1989a,310)? We want to belong to ourselves in cases
involving positive agency by inaction as we do in cases involving positive
agency by action.
I believe that this
argument fails. I have argued that either Quinn's picture of how rights
are activated or deactivated depending on what the agent decides to do is false
in which case a system that gives positive rights precedence over negative
rights is not incoherent (my fourth objection), or Quinn's picture of
how rights are activated is correct in which case both a system that gives
positive rights precedence over negative rights and a system that gives
negative rights precedence over positive rights are incoherent (my third
objection). If either the first or
the second disjunct is true, then Quinn's argument for the DDA is a
non-starter.
Moreover, if the first
disjunct is true, then one cannot justify Quinn's version of the DDA on the
ground that it corresponds to our beliefs about the extent to which we wish to
belong to ourselves in cases involving positive agency by inaction. A system of
morality according to which cases involving what Quinn calls positive agency by
inaction should be seen as falling on the allowing side of the distinction and
according to which positive rights take precedence over negative rights gives, ceteris paribus, persons the same degree
of "legitimate control" over what should happen to them as a system
of morality according to which negative rights take precedence over positive
rights and according to which cases involving harmful positive agency by
inaction falls on the doing side of the distinction.
Sixth, persons
may have a primary say over what should happen to them even if neither negative
rights nor positive rights take precedence over the other. A moral system that
includes either some version of Scheffler's agent-prerogative (see 8.1) or
Quinn's version of the DDE (see 2.1) gives persons a primary say over what
should happen to them.
Here is a case that
suggests a third way in which morality could recognize persons as having a
primary say over what should happen to them without incorporating the DDA:
(TRANSPLANT) A
doctor can kill a patient and use his organs to save five others who will
otherwise die.
If neither negative nor
positive rights take precedence over the other, then one might think that it
follows that the one patient's negative rights not to be killed are outweighed
by the five patients' positive rights to be saved. However, that does not
follow.[20] What does follow is that someone's negative right that others do
not take his organs is precisely as difficult to defeat as his positive
right that others do not allow him to be dispossessed of his organs. But
this is consistent with the claim that someone's positive right to be saved
through the use of others' organs (assuming that one has such rights) is
much easier to defeat than one's negative right that others do not deprive one
of one's own organs.
A system of morality according to which
organs belong to the person in whose body they are recognizes persons as having
a primary say over themselves in Quinn's sense. It may forbid killing the one
patient in TRANSPLANT to save five others and forbid allowing one patient to
die to use his organs to save five others. Such a system of morality grants
persons a certain status with respect to their organs whereby their say can
override the collectively greater needs of others. Yet, this system of morality
does not include any version of the DDA.
If there are other ways
of recognizing us as independent beings than by adopting Quinn's version of the
DDA, then Quinn has not explained why it is not mere "conditioning and
prejudice" that we accept the DDA rather than or as well as some other
element of morality that gives us a status as independent beings. There may, of
course, be reasons why we should also (or only) accept some version of the DDA
(as a way of morality recognizing us as independent), but pending the provision
of such reason Quinn cannot claim to have shown anything else than that we
should not adopt a consequentialist moral system.