Dividing Harm
Draft
Gerhard Øverland
Those who argue against the permissibility of killing innocent aggressors take mere causal
contribution to harm to be morally insignificant.1 In order to be a permissible target for
defensive force, according to this view, a person must be morally responsible for a threat to
others. In this paper I argue that mere causal connection to the threat of harm is morally
significant, and explore the degree to which it is.
The paper has two parts. In the first I look to demonstrate the intuitive plausibility of what
I call “the asymmetrical fair share procedure”. According to this proposal, innocent
contributors have a duty to take on a fair share of the harm if dividing it is possible, and a
fair share of the risk of being harmed if redistribution of harm is impossible. In the second
part, I develop a contractual account of why causal connection is morally significant.
Preliminaries
Contributing to harm is an elusive notion. My interest is in simple ‘paradigmatic’ cases. For
instance, a person fires a gun and shoots someone, or flips a switch that empties water into a
container, drowning the person trapped inside, and so forth.2 Mere contribution is
contribution to harm without moral responsibility, and is what we find among innocent
contributors. A person might, for instance, be an innocent contributor when the agent is
excused for believing that harming another is permissible, perhaps because he or she has
good grounds for believing so, has been manipulated into believing so, or has been drugged,
1
There is a growing literature on the topic and an increasing number of philosophers seem to support this view,
see N. Zohar “Collective War and Individualistic Ethics: Against the Conscription of ‘Self-Defense’” Political
Theory 21 (1993): 606–622; Michael Otsuka “Killing the Innocent in Self-Defence”, Philosophy and Public
Affairs 23 (1994): 74–94; Jeff McMahan “Self-Defense and the Problem of the Innocent Attacker” Ethics 104
(1994); David Rodin War and Self-Defense (Oxford: Clarendon Press 2002), at pp. 70–99; Jeff McMahan “The
Basis of Moral Liability to defensive Killing” Philosophical Issues 15 (2005), and; Kimberley Kessler Ferzan
“Justifying Self-Defense” Law and Philosophy 24 (2005): 711–749.
2
One might be said to be the cause of someone’s death though there has been no significant causal connection.
For instance, assume that Nice is very pretty and out walking. Stare drives past, and cannot resist turning head.
As a consequence Stare drives off the road and eventually dies in a crash. In some sense it was her beauty that
caused his death. We should not, however, say that Nice contributed to the death of Stare.
1
inducing her to harm the victim as an automaton. In my use of the term here, an innocent
contributor is essentially someone who initiates or sustains a harmful causal sequence but is
not morally at fault for having done so.3
When discussing the permissibility of defensive force against innocent contributors, it is
common to envisage conflict situations involving a contributor and victim, one of whom will
be killed. In such cases it is either permissible to kill innocent contributors in self-defence or
it is not. I think we gain valuable insight into this issue by envisaging situations where the
harm may be divided and shared between the two parties.
Bill becomes a contributor to harm suffered by Alice when Bill initiated or sustained a
causal sequence that led to Alice’s’ harm. Envisage that Bill through no fault of his own is
in such a significant causal relation to harm suffered by Alice. We can imagine different
levels of harm, from loss of a finger, through loss of limbs, to loss of life. Alice is not in a
position to defend herself, but Cathy, who is close by, could intervene in the causal chain to
prevent these harms. Since Cathy cannot save everyone from harm, we need to ascertain
which rule she should act on to divide it.
Now, there are various ways of distinguishing between innocent contributors. We may,
for instance, distinguish between those intending harm because they have good reason to
believe killing a particular person is a necessary corollary of a just cause (innocent
aggressors); those causing harm accidentally by their actions (innocent agents); and those
harming others through no actions of their own at all (innocent threats).
Perhaps the most promising way to introduce a morally relevant distinction among these
innocent contributors would be to distinguish between those who intend harm (innocent
aggressors) and those who do not (innocent agents/threats). Yet, if the idea is to let intention
be a morally decisive factor, the proposal leads to counterintuitive results as soon as we
consider standard cases of self-defence. An innocent aggressor may have set out to harm
someone, but hurt someone else by mistake. It seems irrelevant whether Bill intended to kill
Alice or someone else and mistook Alice for this person. Whatever could permissibly be
done to prevent (innocent) Bill from killing Alice when he intended to kill her, could also
permissibly be done to prevent him from killing her even though he thought he was killing a
3
According to Philippa Foot, the relevant distinction between doing and allowing harm is primarily one
between a) initiating or sustaining a harmful causal sequence, and b) allowing a harmful causal sequence to run
its course. See Philippa Foot “Killing and Letting Die” in B. Steinbock and A. Norcross Killing and Letting Die
(Fordham University Press: 1994): 280–289, at p. 281. Ultimately, of course, what it is to initiate and sustain a
causal sequence needs to be made precise. I shall not do so here.
2
different person altogether.4 Hence, the fact that an aggressor intends to harm a particular
individual does not imply that the implications of being an aggressor apply to him only when
he is about to harm the intended target. At any rate, the implications of an intention to harm
a particular individual carry over with equal force if the aggressor mistakes the identity of
his victim.
We therefore need to consider whether the mere fact of innocently acting on an intention
to harm, though one is about to harm wrong person, would distinguish implication-wise this
aggressor from one who innocently acts in such a way as to cause harm unintentionally
(innocent agent). Both are completely innocent. One person is unlucky in believing that he
needs to harm another; the other is unlucky for causing harm by his actions. While only the
first intends harm, it is not clear why that should matter as long as he does the right thing
given what he knows and forms intentions that we think he ought be disposed to form. He
could, for instance, have good reason to believe that by harming Alice he would save others
from her impermissible actions.5 That intentions are essential in moral appraisals should not
lead us to conclude that they are significant when the agent is not at fault for forming them.
At this point, we need to ask if there are grounds for distinguishing morally between
people whose actions cause harm (innocent agents) and people whose actions do not, but
who still represent a causal threat (innocent threats). The latter could, for instance, be a
person who is blown by the wind or thrown against another by a villain. Again, it seems
difficult to understand the rationale behind the distinction. After all, we are assuming that
there is nothing risky or otherwise peculiar about the actions of the first agent, but that
unexpectedly, and due to bad luck, his behaviour is about to cause harm. Perhaps the power
circuits at his house were rewired, making the flip of a switch result in harm to his
neighbour. It is hard to see why a different level of force should be permissible against such
an agent as compared to against a person who was blown by the wind in such a way that he
harms another.
A different line would be to distinguish between innocents that have engaged in risky
activity and those that have not. According to Jeff McMahan, the driver of a car that goes out
of control through no fault of her own is innocent but not non-responsible. There might
accordingly be reasons for distinguishing morally between two types of innocent threats:
those brought about by responsible actions, such as taking up driving and risking brake
4
This, of course, applies to culpable aggressors as well, but then much more force may be levelled against
them if necessary.
5
In fact, it might on some occasions be grounds for giving more protection to an innocent aggressor than just
any unlucky agent, as only the former has demonstrated a willingness to incur risk to help protect others.
3
failure, and those where the agent has done nothing to become a threat to others. 6
McMahan’s contention that there are distinctions to be made here seems plausible.7 Against
McMahan, however, I would suggest that responsible agents who have become innocent
threats by choosing to participate in risky activities have to bear more of the cost than
ordinary innocent contributors, and not simply more than innocent bystanders. Being
responsible for engaging in risky activities widens the range of actions we can take against
those engaging in them, though the baseline is not what can be done towards innocent
bystanders, but towards innocent contributors.8 I now turn to determine the latter baseline.
The asymmetrical fair share procedure
Suppose again that Bill through no fault of his own is about to harm Alice. Basically, there
are three positions to take regarding whether we should give priority to the innocent
contributor, Bill, or his innocent victim, Alice. First, reasoning that people have a right to
defend themselves against any unjustified aggression or threat, we might elect to help Alice.
When becoming an unjust threat to another person, Bill might be thought to lose his right not
to be significantly harmed. If someone is unlucky and contributed to harm, this unlucky
person has to carry the cost, including the costs of helping to defend the prospective victim
from harm. We need to take proportionality into account, but the fact that Bill is morally
innocent is understood to be insignificant with regard to the question of defensive force, as
he is causally responsible for being a threat to another person. For situations in which killing
Bill would be necessary to thwart the threat, and the harm or risk to Alice was sufficiently
grave, killing Bill would be permissible.9
Alternatively, we could instead give priority to Bill. The central intuition supporting this
position is that Bill has not done anything to lose his right not to be harmed or killed, and
should be treated accordingly. To become a permissible target, one might argue, one must be
See McMahan “The Ethics of Killing in War” (2004): 693–733 and “The Basis of Moral Liability to
defensive Killing” Philosophical Issues 15 (2005): 386–405.
7
I argue along these lines, though for a different purpose, in ‘Contractual Killing’ Ethics 115 (2005): 692–720.
8
It may at this point be worth observing that the tendency to let intentions play a role may be derived from the
idea of being engaged in risky activity. After all, taking up arms and intending to kill someone is a risky
activity in the sense that there might be some chance that the person is not a justified target after all.
9
One position that appears to take this form is Judith Jarvis Thomson’s right-based account of self-defence. It
seems to put innocent aggressors and culpable aggressors in the same boat, as the right to self-defence is based
on right violations which, according to Thomson, can be committed by innocent as well culpable aggressors.
According to her, culpability has nothing to do with the right to self-defence, since an agent who exercises this
right is not an agent of justice. See J. J. Thomson “Self-Defence”, Philosophy and Public Affairs 20 (1991):
283–310. For a good discussion of Thomson’s position see Yitzhak Benbaji “Culpable Bystanders, Innocent
Threats and the Ethics of Self-Defense” Canadian Journal of Philosophy 35 (2005): 585–622
6
4
morally responsible for putting someone in danger. Innocent contributors would accordingly
be morally indistinguishable from innocent bystanders. And, since there are strict constraints
on what we can do to innocent bystanders to save ourselves and others from harm, there will
be strict limits on what we permissibly can do to Bill as well. Call this view the “on-parwith-a-bystander view”.10
The on-par-with-a-bystander view assumes that an innocent contributor has no additional
duty to shoulder the costs of neutralising the threatening consequences of his or her
behaviour. Harming Bill to protect Alice in a situation in which the former is about to harm
the latter through no fault of his own would be regarded on par with harming an innocent
bystander, David, for the same purpose. Most maintain that not much could permissibly be
done to David to save Alice from harm. It could perhaps be contested whether the on-parwith-a-bystander view gives priority to the innocent aggressor. But notice that in typical
cases the decision to mount defensive force is in the hands of the victim. By holding a
prospective victim’s use of force against the innocent aggressor to be on par with force
directed at an innocent bystander, the upshot would be that the aggressor would be morally
permitted to use defensive force against almost any forceful action taken by the victim, just
as bystanders would.
An attractive feature of the on-par-with-a-bystander view is that it points to something
about which it is easy to agree has moral significance, namely moral responsibility.
However, there are good reasons to distinguish between an innocent contributor and an
innocent bystander and let mere causality have moral significance, though not for the
purpose of blame or judging character. The view I shall propose – the third position – might
be seen as occupying the middle ground between the positions just sketched.
A fair share
In the on-par-with-a-bystander view, the assumption is that mere causal connection is
morally insignificant. There is some plausibility to this. If a person is unlucky and causes
harm to another, the second person is unlucky too, as she ends up as the victim. It seems to
make little sense to single out any particular person as being the unlucky one to determine
who should be given priority. Yet although the direction of the causal link does not tell us
where our priorities should lie, it would be a mistake not to recognize that the causal
connection joining them also distinguishes them from others. An initial proposal would
The most explicit supporter of the on-par-with-a-bystander view is Michael Otsuka, see his “Killing the
Innocent in Self-Defence”.
10
5
therefore be that both have been unlucky, and that we therefore should divide the harm
between them, keeping everyone else out. In the example at hand, harm would be divided
between Bill and Alice, and not imposed on David.
Normally, when we claim that fairness requires that something should be divided between
particular individuals, we take it to imply that it should be divided equally, unless there is
reason to the contrary. When two people are involved and dividing the cost is possible, I
shall accordingly divide the cost equally between them. Hence, if Alice was in danger of
losing both arms due to Bill’s innocent behaviour, and Cathy was in a position to intervene
and divide the cost between Bill and Alice equally, each losing one arm, then she should do
so. To take on a fair share in such cases means roughly that the harm is divided in equal
parts amongst the people in the causal chain.
We can indicate why Bill, as an innocent contributor, should bear a larger share than
innocent bystanders by appealing to Bill’s duty to compensate Alice for damages sustained.
If a person’s property is damaged, the person who did so is generally required to compensate
the victim for his loss. And though one could argue that the innocent contributor has no duty
to shoulder the whole cost and make full recompense, the contributor does, we may
plausibly assume, have a duty to shoulder (at least) a fair share, i.e., fifty per cent if the cost
can be divided.
At this point one might wonder why we should restrict our attention to those involved in
the causal chain and no one else. Given the alleged moral insignificance of contribution,
harm to innocent people should be divided equally between all kinds of innocents, including
contributors, victims, and bystanders alike. A fair share procedure should consequently
include bystanders in addition to innocent contributors and their respective victims. That
would have serious counter-intuitive upshots, however. For instance, it would be permissible
first to impose harm on an innocent bystander if the harm inflicted was reduced from three
limbs to two or less, and second to kill one innocent bystander to save two innocent people.
This proposal would obviate reasons for thinking innocent bystanders have claims on some
special kind of protection, and few are ready to accept this conclusion.
Luck reconsidered
We have been considering the proposal that the innocent contributor and the innocent victim
have both been unlucky, and that they ought to divide the burden between them. But though
the equal-bad-luck view might be plausible on the face of it, there is an important difference
between the two parties. It seems wrong to say that the victim has been unlucky if he or she
6
is able to escape the threat even though it leads to the death of the innocent contributor. The
prospective victim’s bad luck is essentially connected to a prospect of being harmed. By
contrast, it makes perfect sense to say that the innocent contributor has been unlucky in
causing the death of the defendant, even though he should be able to survive himself – and
lucky in that respect.
Clearly, being lucky means different things here. In one situation bad luck refers to being
a victim, in another it refers to being a contributor to harm. One might perhaps want to say
that Alice is unlucky as well in that she needs to allow an innocent person to be harmed in
order to save herself. But although the latter is true, I do not think we look upon this kind of
bad luck in any way close to being a victim or a contributor. The bad luck on the part of
Alice in not being able to save Bill is just the same bad luck that David would be subject to
as an innocent bystander in case he were unable to save Bill.
Anyway, I shall at this point propose that the innocent contributor has been exposed to the
initial bad luck and has a duty to shoulder his or her part of the harm. The victim is still to be
regarded as an innocent bystander with no additional duty to shoulder any particular cost to
alleviate the burden on the innocent contributor. To capture this, I propose an asymmetrical
fair share procedure, which requires the innocent contributor to take on a fair share of the
burden at stake. Since the duty to take on a fair share is asymmetrical, it only applies to
innocent contributors. A rationale behind this duty may be stated as follows. If a person has
innocently initiated or sustained a causal sequence the effect of which is to put someone else
under threat of being harmed, then, other things being equal, he has a duty as a matter of
fairness to shoulder his share of the cost.
There are situations that seem to support this proposal. Envisage innocent Bill again. He
is falling at speed towards Alice, endangering both her legs. Alice has three options. She can
remain where she is, lose her limbs but save Bill’s; she can roll to the left, save one of her
own legs and one of Bill’s; or she can roll to the right, saving both of her own legs but
causing Bill to lose his. If we think the fair share procedure should be symmetrical, the
defendant would have a duty to turn to the left, saving one of her own and one of Bill’s legs.
But it sounds rather implausible. It would seem perfectly permissible, on the other hand, if
Alice flipped over to the right, and saved her own legs, but at the cost of those of the
threatener.
Consider next a situation that gives Alice no possibility of escape. This time, innocent
Bill has three options. He can continue the direction of his fall and save his own legs by
using Alice’s as cushion; he can flip towards the left, which would cost them one leg each;
7
or he could flip over to the right, this time sacrificing both of his legs and saving those of the
victim. It seems plausible to assume the innocent contributor would have to accept half of
the burden, and his appeal to innocence would at most justify his reluctance to bear the
whole load. Hence, the innocent contributor would have a duty to turn, and turning left could
be enough to discharge this duty.
One might think the difference between the two cases has to do with Alice simply having
to flip to one side, saving herself at the cost of Bill’s legs in the first example, while Bill
needs to use Alice’s legs to cushion his fall, and thereby save his legs in the second. But that
is not obvious. It seems permissible for Alice to shield herself even if it costs the innocent
aggressor two limbs, or shoot him in the leg if that saves one of hers. By contrast, if Bill sees
Alice’s attempt to roll over, in which case he will lose both of his legs, he may not alter his
trajectory so that he manages to save one of his at the expense of hers, nor may he stop Alice
from escaping to ensure that she cushions his fall.11
Observe that the duty to compensate is asymmetrical as well. Although innocent
contributors have a duty to compensate costs imposed on a victim, the victim has no
corresponding duty to compensate losses incurred by the innocent contributor. If Bill, for
instance, is unlucky enough to damage a particularly fragile item of his against Alice’s
belongings, he alone has to carry the cost of repairing it, provided, of course, that Alice’s
belongings were clearly not in the way.
In accordance with these observations, third parties could permissibly intervene in a
manner that inflicted a fair share on the contributor. If it is impossible to divide the cost third
parties might nevertheless have reason to impose the whole cost on the innocent contributor
rather than letting it fall on the victim. Although a contributor only has a duty to take on a
fair share, the asymmetry in duty between contributors and victims seem to tip the scale
towards the former when the cost cannot be divided. This being so, it would be permissible
to kill an innocent contributor to save a victim.
Against the on-par-with-a-bystander view
Let me indicate other reasons for granting moral significance to mere contribution. If the onpar-with-a-bystander view were true, we would have to allow an innocent aggressor to kill
11
Well, why not one might ask? After all, when Bill was falling towards Alice he would have a duty to take on
fifty percent, but Alice was at liberty to escape. This leaves fifty percent unaccounted for. One option would be
to say that Bill has to take on fifty percent, and then they are at liberty to fight over the rest. I shall have to
leave that open here. In the last section I indicate contractual reasons why it would be wrong to force Alice to
accept more cost than what is required from innocent bystanders.
8
more than one person, as we ought not to kill one innocent bystander to defend a small
number of people. Hence, if a person was on a killing spree because he had been drugged or
manipulated we could not kill this person unless the number of his potential victims hit a
threshold number allowing us to kill a bystander for the same end. That seems implausible.
According to the asymmetrical fair share procedure, we would be permitted to kill the lesser
number, that is, the innocent aggressor. For the imposition of harm, on the on-par-with-abystander view, we would have to allow an innocent contributor to inflict quite serious harm
on a victim – removing two limbs for instance – and we could still not sever one of the
contributors’ limbs to prevent it, because it is impermissible to subject innocent bystanders
to such treatment. According to the view proposed here, we could require the innocent
contributor to bear a fair share of the cost, preventing him or her even at the cost of a limb.
Friends of the on-par-with-a-bystander view could retort that the proposal is only meant
to cover situations involving killing, and that situations involving harming should be dealt
with according to the asymmetrical fair share procedure. While this would save the view
from some counterintuitive upshots, it would leave some and create others. For instance,
while we permissibly could prevent an innocent contributor severing both of a third party’s
arms by removing one of his, we would have no right to inflict more harm on him than on
innocent bystanders if he were about to kill the victim. To this proponents of the on-parwith-a-bystander view could reply that it covers situations only of killing on both sides.
True, but then it seems odd to say that we might prevent a few people suffering serious harm
by inflicting serious harm on the innocent contributor, but not to prevent the killing of a few
by killing the innocent contributor. Moreover, it seems difficult to explain why we should
treat killing and inflicting (serious) harm differently. Supporters of this view would have to
explain why causal connection makes a difference in situations of harm yet lacks it in
conflict between lives.
Proponents of the on-par-with-a-bystander view seem to believe that a slight increase in
moral responsibility transforms the contributor into a permissible target for lethal defensive
force – as they seem to find it permissible to kill negligent contributors while denying the
permissibility of killing innocent contributors.12 It seems puzzling that such a slight increase
12
One might question whether the difference between innocence and negligence is slight. But observe that we
do not tend to blame negligence very much, perhaps because most of us are guilty of it from time to time. It is
important not to confuse it with recklessness. As George Fletcher points out, according to the Model Penal
Code “the difference between reckless and negligent risk-taking arises at the level of the actor’s awareness of
the impermissible risk. In cases of recklessness, the actor disregards the risk; in cases of negligence, he fails to
perceive it.” George Fletcher Rethinking Criminal Law (Oxford: Oxford University Press 2000), p. 261). In line
9
in moral fault can have such a huge impact. If one denies the significance of mere
contribution it is difficult to explain why contributing to harm with only a very minor degree
in culpability (negligence) is so much more significant than innocent contribution to harm.
After all, the innocent contributor remains on par with an innocent bystander while the
negligent contributor suddenly finds him or herself as a legitimate target of lethal defensive
force when that is necessary to save the victim. Yet how can such a large leap be justified by
a slight increase in moral culpability?
There is no such unaccounted-for gap if one accepts the moral significance of being an
innocent contributor. According to the asymmetrical fair share procedure the first
incremental step comes as a consequence of merely being a contributor, with liability
increasing stepwise as a function of the contributor’s culpability.
Perhaps even more puzzling for proponents of the on-par-with-a-bystander view is the
wide discrepancy between negligent contributors and negligent bystanders that their view
would seem to entail. Almost everybody accepts that negligent contributors are liable to
lethal defensive force, while hardly anyone seems to think the same about negligent
bystanders. But if one denies the significance of mere contribution, it becomes difficult to
explain why negligence in causing harm is so much more significant than negligence in
letting harm happen. Why is it worse – implying that significantly more might be done
towards such a person – to initiate or sustain a harmful causal sequence rather than to let it
happen if mere contribution lacks moral significance?
The slight increase is in moral culpability, one might retort, and that is significant. But
observe that the effect of increased culpability is not supposed to be equal across all
situations. Culpability is supposed to be subject-dependent in the sense that culpability
arising from a failure to save is taken to be very different from a failure in not harming. It is
only culpability in contribution that is supposed to have such significant implications. But
then it seems as if one backslides into the question of contribution again. And at this point,
one cannot say it is significant because one person is contributing to harm, while the other is
not, as one has already denied its moral significance. Why moral responsibility for causing
harm should be regarded as particularly important if one denies that mere causality is
morally significant seems to hang in mid-air.13
with this emphasis on awareness, we can say that while a negligent person would take the fact that he is
exposing others to undue risk as a reason against so acting, a reckless person would not.
13
One might observe that a slight increase in causality has such a huge impact. But then, contribution to harm
is typically something that comes in discrete jumps. If what is needed from you in one situation is to bend a
10
Contractual considerations
I have previously argued on contractual grounds that in conflicts between innocent
aggressors and their innocent victims we ought to give preference to the party who did not
initiate the lethal conflict.14 I did not then consider the option of dividing the harm. Nor did I
focus on causal connections since I was interested in situations of what I called “apparent
aggressors”. The latter implied invoking the notion of reasonableness. An apparent aggressor
is a person that it is reasonable to interpret as threatening, but who in fact is not a danger to
anyone. There might be situations in which a person constitutes no causal threat to others but
is morally responsible for behaving in such a way that it is reasonable to interpret him or her
as threatening. In such cases it is uncontroversial to assume that when deciding where to
allocate the cost, the apparent aggressor would be the one. But there could be situations in
which an agent is not to blame for engaging in a certain activity but appears threatening all
the same. In those cases, I have argued, if it is a reasonable interpretation of the agent’s
behaviour, he or she still has to carry the cost. In this paper I put apparent aggressors to one
side and focus on standard cases involving real aggressors or threats.
I invoke hypothetical contractual reasoning to illuminate how harm should be divided
among innocent people, and will consider the asymmetrical fair share procedure from the
perspective of a hypothetical contract situation. The first thing to decide is who is to be a
part in the contract. It seems plausible to assume agreement about the permissibility of
giving priority to the innocent at the cost of the culpable, at least when the difference in
culpability is above a certain threshold. We can accordingly restrict participation in the
contract to innocent people or to people whose culpability is below this threshold.
By appealing to hypothetical consent, the point is to identify who would have reason to
give their consent in the contractual situation. In general, people make contracts because
they have reason for doing so. Good reasons for entering into a contract are prospects of a
good outcome for the contracted parties. In a hypothetical contract we let certain information
be concealed from the participants, and the veil of ignorance I envisage could be called a
“conflict veil”. As the name suggests, it is only a question of revealing the identity of those
involved in a particular conflict where harm has to be distributed among innocent people. All
finger, while in another you are required to push a handle as hard as possible with the force of your whole
body, no one would take that difference to be significant.
14
See my ‘Self-Defence among Innocent People’ Journal of Moral Philosophy 2.2 (2005): 127–146.
11
other information remains available to the parties. In particular, they have access to the best
available knowledge about their respective probabilities to end up in such conflicts.
As prospective interveners, our central concern is what we know about the participants in
virtue of which they would have reason to consent to rules dictating how third parties should
interfere in the conflict. Clearly, their reasons would depend on their interests. One
important interest, however, would be to reduce their risk of being harmed and of dying. I
leave out other reasons they might have for selecting procedure to keep things simple.
Reasons for accepting rules for regulating actions of self- and other-defence would
accordingly be to avoid harm to the innocent while expending those of the culpable.
Assumptions about the implicated parties’ preferences justify certain actions because by
appealing to these preferences, we can make informed predictions about which rules they
would have consented to. If these assumptions prove wrong, we can no longer plausibly
claim that they would have consented to the rule by which our actions are guided.
Hypothetical (actual preference) contractualism does not exhaust morality. Against a
background of deontological restrictions, its aim is to explain when it might be permissible
to violate specific restrictions. The basic idea is that such violations might be permissible
provided they are in the interest of the implicated parties. As for deontological restrictions, I
take them to be best understood as being in the interests of the individuals they protect. In
particular, deontological restrictions are meant to prevent the sacrifice of individuals for the
benefit of others. Given such restrictions, hypothetical contractualism aims to demonstrate
that it may be occasionally in the interest of the implicated parties to sidestep certain
restrictions. And when that occurs it would not count as unfair treatment if a particular
person ended up as victim of the sidestepping. In the case at hand, the restriction to be
circumvented is the alleged impermissibility of harming or killing an innocent person for a
comparable benefit.
Epistemic uncertainty
We can first observe that being an initiating causal factor will often serve as a proxy for
culpability. In a conflict between two people, the contributor will tend to bear more
responsibility for the situation than a prospective victim. Even if a contributor is not morally
culpable to the fullest extent, they may be guilty of some negligence or recklessness. And,
since being an initiating causal factor is a discernable external feature, as opposed to moral
culpability, sacrificing the initiator is likely to save more innocent people in the long run as
we are less likely to get it wrong. Epistemic uncertainty therefore elicits a presumption in
12
favour of having a rule that gives priority to the defending party, i.e. the victim, because it
will give (innocent) contracting parties a higher chance of survival and of avoiding harm. As
contributors tend to be morally culpable more often than bystanders, participants in the
contract have reason to consent to a rule that supports such priorities.
It is difficult for people under attack to know whether or not aggressors are innocent. A
rule that gives priority to those who respond to a causal threat will therefore enable people to
avoid giving up their life on wrong occasions. Wrong occasions are instances in which the
defender erroneously believes the aggressors to be innocent and therefore fails to defend
herself. If we were to choose a rule that permits self-defence only in cases in which the
contributors were known to be culpable, there would be instances in which innocent people
were killed while the culpable got off scot-free. To avoid this, we should adopt a rule
permitting prospective victims the use of defensive force against initiators of harmful causal
sequences. Participants in the hypothetical contract therefore have good reason to allow a
rule permitting the use of defensive force against initiators inasmuch as compliance to the
rule would not cause innocent deaths while saving the culpable. On the contrary, it would
sometimes save the innocent at the cost of the culpable.
Moreover, it would help third parties coordinate intervention. Third parties might lack
time and ability to coordinate their intervention, and would appreciate a simple rule helping
them decide on which side to intervene. Being a contributor to prospective harm is a
discernible external feature of conduct that can be used for this purpose.
We might observe an incentive-based argument as well. Adopting a rule according to
which the contributor has to carry the cost would minimize the likelihood of such situations
arising in the first place, since people would take pains to avoid becoming contributors. This
goal could perhaps be achieved best by saying that those who initiate a threat are not fully
innocent unless they have taken every reasonable precaution to avoid becoming contributors
and that only fully innocent contributors are without liability. This would give people an
incentive to take all reasonable precautions against becoming contributors. Only people who
had taken such precautions would be treated as innocent parties. A problem with this is that
moral responsibility is not an easily observable feature. It will often be difficult to establish
whether the agent has taken the required precautions. A rule giving priority to the defending
party might serve the contracting parties better. Of course, any known moral responsibility
on part of the contributor would consolidate priorities further and do so according to the
degree of moral culpability.
13
Epistemic uncertainty tips the balance in favour of the defending party. How far will
depend on estimates of likely contributor’s culpability. How often is a contributor about
whom we have no particular knowledge in fact an innocent contributor? Without attempting
precise estimates, it seems uncontroversial to assume that contributors are more likely to at
least have been guilty of negligence or recklessness than the victim who is at the receiving
end of their harmful conduct. Of course, sometimes the rule to prefer the defending party
will be wrong, and what looks like a victim may in fact be the aggressor, or that harming this
person may have been justified for some other reason. But although this is true, the
frequency of such mistakes will hardly be sufficient to give the contracting parties a reason
against favouring prospective victims.
Now, the preference with which prospective victims should be treated will depend on the
epistemic conditions. It is easy to judge wrongly with regard to innocence. If determining
priority on the basis of slightly greater numbers would not save more innocent people than
would be saved by giving priority to the defending party – because, perhaps, some of the
aggressors were culpable – the contracting parties would have reason to support a rule that
gave priority to the defending party even when there are minor asymmetries in numbers.15
Hence, the contractual argument also recommends giving priority to the defending party in
situations with more than one contributor, but at some point, the asymmetry in numbers will
suggest giving priority to the assumed innocent contributors.
Epistemic transparency
I have indicated reasons for selecting a rule that gives priority to the defending party. These
arguments traded on epistemic uncertainty. But although contribution is often correlated
with culpability, there are instances in which we know this is not the case.
Division
From a contractual perspective, the fact that an individual has bad luck and becomes a
contributor has no bearing on choice of distributive pattern. The contracting participants
want to reduce the risk of being harmed. Assuming transparency and full knowledge of the
parties’ innocence, there might be no grounds to decide either way.
An equal split would possibly be in their interest. But why should the contractors be
interested in having harm divided? Why locate a particular duty on innocent contributors?
Why not just leave them on par with bystanders? Assuming knowledge, the uncertainty
15
See my ‘Killing Soldiers’ Ethics & International Affairs 20:4 (2006): 455–475.
14
argument does not apply, although it might be worth mentioning that we seldom have full
certainty in real life anyway.
The incentive-based argument still carries some weight though. By knowing one will be
held accountable as an innocent contributor, one has reason to make serious efforts to avoid
becoming one.16 Again, the expansive responsibility argument would probably provide much
of the same incentive. But besides the limits alluded to above, a further problem with this
argument is that it gives people an incentive to feign innocence since so much is at stake.
This incentive is reduced if one requires innocent contributors to take on a fair share. A
slight increase in culpability would only add a minor cost-related duty, as one would already
be required to shoulder a fair share by merely being an innocent contributor. The latter
would, of course, have an incentive to pretend not being a contributor at all. But that is a
more difficult enterprise.
Another important reason to adopt a fair share procedure, rather than letting the victim
bear the whole cost, comes from the observation that it is better to take a fifty percent loss
than a fifty percent risk of sustaining the whole cost. For instance, it is better to lose one arm
than have a fifty percent risk of losing both. There is likely to be some disagreement here,
but I would guess most people would generally prefer a defined loss to a fifty percent risk of
the double loss. Think of any dividable entity of value in your possession, like money,
property, life span, health, or body parts, and consider whether you would prefer losing fifty
percent of any one of them or a chance of keeping the whole of it at the cost of fifty percent
risk of loosing the whole entity.
It may pay to observe that there is no reason to give priority to the innocent contributor
from the perspective of the contractual position. An alternative to fair share, namely letting
harm fall where fate dictates – on the victim normally – may nevertheless merit
consideration. According to this proposal, when both parties are innocent neither of them
would have any particular duty to shoulder cost, and both would be permitted to use
defensive force on the other. Third parties would have no reason to intervene. But letting the
involved parties fight themselves out of the problem mostly exacerbates the harm, giving the
contractual parties reason to reject this option. Allowing third parties to prevent escalations
could partly mitigate the problem, but it would be unreasonable to expect complete success.
Moreover, this kind of procedure would favour certain groups and types, of people, giving
those who are member of neither reason to reject it.
16
We see this at play in civil law, as strict liability, where innocent contributors are held accountable for
damages that they cannot be held morally responsible for.
15
Asymmetry
Why give moral significance to causal connections at all? If one thinks luck-factors like
being an innocent contributor to harm are irrelevant, one might think the burden at stake
should be divided with no regard to causal connection. The contracting parties could
likewise let all innocents be on par in order to reduce harm, and include victims,
contributors, and bystanders alike. On the face of it, it would seem to benefit all participants
by reducing risk of being harmed and killed.
I previously appealed to the counterintuitive upshots of such views, which would leave
the innocent with no special protection. That will not do contractually. Explaining
restrictions contract-wise entails asking whether the parties in a hypothetical contract
situation have non-moral reasons for accepting them. We cannot assume, for example, that
harming innocent bystanders is morally worse than harming innocent contributors. On the
contrary, the argument should explain the impermissibility of widening the net with
reference to the situation we would end up with if it were permissible to use force against
victims and bystanders. When parties evaluate reasons for accepting rules, they are
interested in long-term consequences across a broad range of circumstances, not how they
work in special or restricted cases. It may be reasonable to discount future consequences, but
not to consider them at all seems irrational.
The argument from epistemic uncertainty distinguished between contributors and victims.
While contribution is often correlated with culpability, victim-hood is normally associated
with innocence. The latter is true of bystanders as well.17 These observations singled out
innocent contributors. But when there is no uncertainty, the contracting participants may
have no reason to discriminate between innocent contributors, victims, and bystanders.
However, a version of what I elsewhere have called “the undermining argument”18 explains
why it is wise to draw a distinction between contributors on the one hand, and victims and
bystanders on the other. The general point is that if there are no constraints, unfortunate
long-term consequences will undermine the antecedent rationality of including everyone in
the general lot. The argument has two steps, of which the first concerns the danger of
intervening with force, as recipients are unlikely to accept it without defence. Fear of
resistance will reduce the number of interventions. Whether anything will be gained by
17
Of course, they may have failed to assist the victim at moderate cost, and in that sense be morally
responsible. I do not discuss such failures here.
18
See my ‘Contractual Killing’.
16
permitting force against bystanders is therefore a moot question, as people would be wary of
involving themselves.
The second step trades on the previous reduction and says that successful interventions by
force might cost more than the achieved level of harm reduction because people will want to
avoid being used for harm reduction purposes. If people are aware of the danger of being
used in a bid to temper overall harm, they will take steps to avoid getting involved in the first
place.19 Contracting participants have reason therefore to grant some people protection
against use of force.
The degree of protection depends on how calculations pan out for innocent bystanders,
victims, and contributors. Remember the typical situation we are discussing. Bill innocently
harms Alice, who is not in a position to defend herself. Cathy is close by and able to
intervene in the causal chain to avert the threatened harm. A fourth person, David, who is an
innocent bystander, is also present.
Cathy and David are bystanders. There are good reasons why Cathy should be allowed to
prevent Bill from harming Alice, but that she should not interfere with David. If Cathy were
permitted to use force against David in order to reduce overall harm, David would be
permitted to intervene against her. They are unlikely to accept this kind of risk in order to
help protect Alice, who we assume is a stranger to them. Generally, this norm would provide
bystanders an incentive to move away rather than helping those already involved. These
negative consequences will have to be measured against the likely benefits of permitting the
use of force in the first place. Some benefits are to be expected. A certain number of people,
who would otherwise not be saved from harm would now be saved. However, it is unlikely
that number thus saved would exceed the number not saved because bystanders will look to
avoid accidents for fear of being used to reduce overall harm. The contracting parties would
therefore have reason to prohibit use of force against bystanders.20
The same reasoning does not apply to innocent contributors. Permitting use of force
against them has, on average, more positive than negative effects. In some situations it will
lead to fewer people being harmed or reduce the severity of the harm; in others it will secure
a fair distribution of the harm. There are, of course, unwelcome consequences of using force
against innocent contributors as well. It will invite defensive force, giving prospective
19
One could perhaps have a rule allowing only use of force against those who do not intend to assist? But that
doesn’t sound very operative.
20
That is, above what one can expect bystanders to be willing to shoulder in the first place, i.e. moderate costs.
This is a good reason why the principle of assistance only tells us to assist others from severe harm if we can
do so at moderate cost.
17
interveners reason for not getting involved (as we see it in real life). Bill, however, is already
posing a risk, nothing new about that. And to disqualify forceful intervention altogether
would leave harming and killing too often at the mercy of innocent contributors.
Prospective victims are in some sense bystanders as well, but permitting force against
them may nevertheless be a case apart. Granted it is permissible to impose a fair share on an
innocent contributor, the idea to be considered is whether one should allow intervention to
prevent a victim imposing more than a fair share on the innocent contributor. Alice may, for
instance, take a shot at Bill, or flip over to the right, refusing to cushion his fall. Would the
contracting parties have reason to permit forcing Alice to accept a fair share?
As already noted, permitting bystanders to force Alice one way or another could
encourage Alice to reply with force, giving bystanders an incentive to move away.
Moreover, identifying the prospective victim will often be difficult, expanding the range of
this implication. We are therefore compelled to ask what would be gained by such
interventions in the first place.
Not that much. After all, what we are trying to avoid here is the prospective victim
subjecting the innocent contributor to excessive force. In most cases we will not know for
certain whether both parties are innocent. It will therefore be used wrongly at times, and
given the paucity of knowledge and the victim’s inability to use force there will be few
instances of correct use. Of course, not sanctioning actions against the victim means
implicitly allowing her to impose more than a fair share on the innocent contributor from
time to time. This may generate reason to permit forceful intervention whenever a victim
tries to impose (unconscionably) disproportionate harm on the contributor compared to what
she is about to suffer.
This may seem intuitively plausible, but our intuitions are likely to vary. Some may think
that victims have no particular duty to protect the person who poses a danger to them; others
that although they have no duty to take on a fair share, they are (slightly) more duty bound
than innocent bystanders who are not under threat. For instance, they may think distance is
morally significant and people close by have a (slightly) stronger duty to address a particular
need than people further away. Prospective victims are pretty close. They could accordingly
have a (slightly) higher duty to bear more of the burden of protecting innocent contributors
from harm. Another reason for thinking more could be demanded from prospective victims,
is that nature, destiny, or whatever has selected them to be victims. Any redistribution would
require human intervention. Anyway, I shall assume that considerations like these, if they
have any weight, will not bring the victim on par with a contributor, but may slightly set him
18
or her apart from other bystanders. We may, for instance, think you ought not duck harm to
save a finger if what is at stake for the other is his or her life, though we might be less
inclined to tell you to intervene if doing so would cost you a finger.21
Concluding remarks
In this paper I have argued that merely being a contributor to harm is morally significant. I
first indicated the intuitive plausibility of the asymmetrical fair share procedure, according to
which innocent contributors have a duty to take on a fair share of the harm in question when
the harm can be divided and reallocated, and a fair share of the risk of being harmed or killed
when redistribution is impossible. The asymmetrical fair share procedure attempts to
accommodate the view that an innocent contributor ought to be distinguished from innocent
bystanders while taking into account that he or she is not at fault and should not be required
to shoulder the whole load.
I then developed a contractual account to explain why causal connection is morally
significant. Epistemic uncertainty made contribution to harm a proxy for culpability, giving
innocent contracting participants reason to prioritize defending parties. As the uncertainty
was reduced, the participants had reason to favour something close to the asymmetrical fair
share procedure. Generally, permitting forceful intervention has a tendency to scare people
off, and must to be traded against the likely benefit of allowing force in the first place. The
contracting participants had reason to permit the imposition of a fair share of the harm on
innocent contributors; as we observed, the negative effects are few in number, the benefits
many. For bystanders, by contrast, the negative effects would predominate, providing reason
to prohibit force against them. For victims, we observed very few benefits and some
negative effects. I tentatively suggested that unless the victim was engaged in very
disproportionate use of force against the contributor, use of force against him or her would
be impermissible.
We can now make sense of the previously observed intuitive difference in luck between
an innocent contributor and a prospective victim. In that situation Bill was unlucky not
merely because he contributed to harm, but because he had become permissible target of
force and incurred a duty to shoulder a fair share of the harm. By contrast, Alice’s bad luck
On the issue of ducking harm, see Christopher Boorse and Roy A. Sorensen “Ducking Harm” The Journal of
Philosophy 85 (1988): 115-134.
21
19
was essentially linked to the prospect of being harmed simply because she did not incur any
additional duty to bear cost by having become a prospective victim.
20