Download Understanding Causation in Law: The Debate over Counterfactuals and Overdetermination and more Slides Law in PDF only on Docsity!
Testing for Causation in Tort Law
DAVID A COADY
The traditional, intuitively appealing, test for causation in tort law, known as 'the but-for test' has been subjected to what are widely believed to he devastating criticisms by Tony Honore, and Richard Wright, amongst others. I argue that the but-for test can withstand these criticisms. Contrary to what is now widely believed, there is no inconsistency between the but- for test and ordinary language, commonsense, or sound legal principle.
Introduction:
Overdetermination and the but-for test
There is a widespread intuition that to say that one thing causes another is to make a certain counterfactual claim; roughly, the claim that if the cause had not occurred, neither would the effect.' In philosophy this intuition has motivated a variety of counterfactual analyses of causation.' In legal literature, especially that focusing on tort law, the same intuition has given rise to the 'hut-for' test. Tony Honore explains the legal significance of this test in the following passage: Tort lawyers have traditionally held the view that, whatever the meaning of causal connection, the way to test whether it exists in a given case is to ask whether in the circumstances the hannful result would have occurred in the absence of the wrongful act. This is the widely adopted 'but-for' test ... J
This is TOugh because there are some things which seem to stand in causal relations (eg unchanging states and facts) despite being poor English to describe them as either 'occurring' or 'not occurring'. David Hume's claim that one object causes a second when the counterfactual "if the first object had not been, the second never had existed" is true, has inspired these attempts. see An Enquiry Concerning Human Understanding. Section VII. David Lewis is largely responsible for the recent enthusiasm for the idea, see 'Causation' in his Philosophical Papers Vol 2 (Oxford, 1986) 159-213. Hereinafter, 'Causation'. Tony Honore, 'Necessary and Sufficient Conditions in Tort Law' in David GOwen (ed), Philosophical Foundations of Tort Law (Clarendon Press: Oxford, 1995) 363-386 at 383. Because of their more immediately practical concerns, legal scholars have been inclined to use counterfaetuals to test for the presence of causation, rather than to explicate the meaning of
84 (2002) 27 Australian Journal of Legal Philosophy
While conceding that the but-for test works well in most cases, Honore, and other contemporary legal scholars, have argued that there are cases in which it will find an act not to be a cause, even though it clearly is.' [ will argue that, properly understood, none of these cases provide a good reason for rejecting or modirying the but-for test. The alleged counter-examples to the but-for test are usually called
cases of overdetermination. In tort law a case of overdetermination is a
situation in which two wrongful acts are followed by a harm; and if either of the wrongful acts had occurred without the other, the harm would still have occurred; but if neither of the wrongful acts had occurred, the harm would not have occurred.' The but-for test does seem to lead to counter- intuitive results in some cases of overdetermination. I believe, however, that this appearance is deceptive and can be explained away. Previous attempts to defend the but-for test have tried to do so by arguing, in effect, that there is no such thing as overdetermination. I will argue that these attempts to 'get rid of overdetermination are misguided. The but-for test is quite compatible with the existence of genuine cases of overdetermination.
Trying to get rid of overdetermination
Rollin Perkins, when considering a hypothetical in which someone is struck simultaneously by two bullets, each of which would have been instantly fatal by itself, claims that the but-for test will accurately find that both shooters cause the victim's death: Whenever that would not have happened when and as it did happen, had it not been for this, this is an actual cause of that 6
6
'causation '. Otherwise the philosophical and legal debates have been remarkably similar. Honore, ibid 383-84; Richard W Wright, 'Causation in Tort Law' (1985) 73 California Law Review 1735-1828 at 1775-76; Richard W Wright, 'Causation, Responsibility, Risk, Probability, Naked Statistics and Proof: Pruning thc Bramblebush by Clarifying the Concepts' (1988) 73 Iowa Law Review 1023-28. The same tenninology has entered the philosophical debate about causation through David Lewis, see 'Causation: Postscript E' 199. This debate is closely analogous to that in tort law, although naturally the philosophical debate is not restricted to causation between wrongful acts and harms. Although it is possible that more than two wrongful acts could ovcrdetermine a harm, I think we can assume that such cases will be quite rarc. Furthennore it is easy to extend what I say to them. Consequently I will restrict my comments to cases in which there are only two wrongful acts. Rollin Perkins, Criminal Law (2nd ed, Foundation Press, New York, 1969) 689 (emphasis in original).
86 (2002) 27 Australian Journal of Legal Philosophy
Even if you disagree with Lewis's position that a philosophical analysis of
causation can be satisfied with getting intuitively correct answers only in
'real world' cases, it seems hard to fault a practical legal test for restricting
its ambition in this way.
Nonetheless I do not endorse this strategy. The real problem with it is
that it is inconsistent with many intuitively appealing negative causal
judgements, because it counts anything that influences the time and/or
manner of a harm as a cause of it. This problem is evident in a hypothetical
discussed by Becht and Miller in which an inattentive driver hits a
pedestrian who runs into the path of the driver's car: if the driver had been
attentive, he could have swerved a little, but not enough to avoid causing
the pedestrian an equally serious injury. If we adopted the Perkins, Becht
and Miller strategy of ascribing very stringent conditions of occurrence to
harms, the but-for test would find the driver's inattention to be a cause of
the harm. This follows from the fact that they would have to count the
injury caused by the actual inattentive driving and the counterfactual injury
caused by attentive driving as different harms. Nonetheless, Becht and
Miller concede that the intuition of most laymen and lawyers is that the
inattentive driver in this example causes no harm: While insisting that this
intuition "is actually not true", they are understandably sceptical about the
prospects of their position being widely accepted lO^ Finally they decide to
speak with the vulgar after all; saying that the driver's inattention "was not
a cause" after all, and calling the process by which they arrived at this
conclusion "equating the injuries"." Richard W. Wright has objected,
surely correctly, that this "introduces an inconsistency into their theory that
undermines their use of the minute-detail approach to support a finding of
causation in the merged-fires case.,,
Becht and Miller might reply that the different treatment of the
inattentive-driver case on the one hand, and the merged-fires and
simultaneous-bullets cases on the other, is justified by the distinction
between causation by an omission in the former case and causation by a
positive act in the latter two. But there is no textual support for this
suggestion and it seems to lack any independent motivation. I conclude that
the need to preserve a distinction between merely affecting how or when a
harm takes place, on the one hand, and causing it, on the other, implies that
9 10
" 12
Becht and Miller, above n 7, 29. Ibid. Ibid. Wright (1985) above n 4, 1775. Wright describes the Perkins, Becht and Miller approach as a "modification" of the but-for test. I think it is better to see it as combination of that test with a particular view about the identity conditions of hanns; namely that an actual hann could not have occurred at a different time nor in a different manner.
Tesling for Causation in Tort law 87
we should not always take hanns to have extremely stringent identity
conditions. Nor should we modify the but-for test to make the issue whether
the hann would have occurred at the time and in the manner it did in the
absence of the wrongdoing, rather than whether the hann would have
occurred at all in those counterfactual circumstances.ll
The two cases of overdetennination that have been considered so far
have both been instances of what Wright calls duplicative causation 14 In
such cases the causal status of the wrongful acts are symmetrical with
respect to the hann they overdetennine; that is, they each have an equal
claim to being causes of it. Intuitions tend to differ in such cases about
whether we should say that both wrongful acts cause the hann, or whether
we should say that neither does, but at least it is clear that there is no reason
to say that one does, whereas the other does not. Because it is unclear what
to say about such cases they are poor guides to the adequacy of any
proposed test for causation. Some laymen and lawyers, for example, will
follow Becht and Miller in thinking that both fires in their merged-fires
example are causes, some will follow Wright in denying this.I' Intuitions
seem equally unclear in the two-bullet case, despite the following argument
by Perkins that we must accept that both shooters cause the death:
In the two-bullet case posed, if either shooter can claim conectly
that his shot was not in fact a cause of death, so may the other. The
unavoidable conclusion would be that the deceased did not in fact
die as a result of being shot - which is absurd. 16
But this conclusion is avoidable. It does not follow from the premise that
the victim did not die as a result of being shot by either shooter that he did
not die as a result of being shot by the combination of them; a combination
which one can think of in either set-theoretical or mereological tenns.
Unless there was a conspiracy or other incitement, there seems to be
nothing counter-intuitive about the conclusion that neither shooter caused
the death. That is not to say that there is anything particularly intuitively
appealing about this conclusion either. Intuitions about cases of duplicative
causation just seem to be too indecisive to bear the weight of theory.
Consequently, the but-for test is compatible with the existence of genuine
cases of duplicative causation.
13
14
I' 16
These two tests would amount to the same thing for all practical purposes;
differing only over the metaphysical issue of the identity conditions of
harms.
Wright, (1985) above n 4, 1775. In the philosophical literature these would
be called 'symmetrical overdetermination' or 'symmetrical redundancy'.
Ibid 1779.
Perkins, above n 6, 689.
mortally burns the victim but before the victim dies of the burns someone
else kills him with a blow to the head;20 the defendant pushes the victim
from a tall building but on the way down the victim is shot and killed
instantly by another. 21 In response to such examples 1 will return to the idea
of ascribing stringent, though not this time too stringent, identity conditions
to the harm. In such cases there is no need to appeal to a detailed
description of the manner in which it occurred; an idea which has already
been undermined by drawing attention to the distinction between causing an
event and merely affecting how it happens. Instead we can restrict ourselves
to a detailed description of the time at which the harm occurs, since it
would have been different, but for the preempting cause.
However, we must be careful. We do not want to say, for example,
that a counterfactual death that occurs at any time other than an actual one
is ipso/acto a different death. That would entail that saving a person's life
was causing that person's eventual death; since that death would not have
occurred but for the life-saving action. Just as there is a distinction between
affecting the manner of a death and causing it, there is a distinction between
affecting the time of a death and causing it.
Arguably this distinction is only legitimate in one temporal direction.
Although we typically do not want to say that delaying death is causing it,
we typically do want to say that hastening death is causing it. Someone who
brings it about that instead of dying now you die later is usually a life-saver,
rather than a killer, and someone who brings it about that you die an
"untimely" death is usually a killer, even though you would have died later
anyway. We can accommodate this asymmetry by distinguishing actual
deaths from any counterfactual deaths which would have occurred later than
them, while identifying actual deaths with counterfactual deaths (of the
same person) which would have occurred earlier than them. This will mean
that the but-for test will find the preempting causes (which have been
considered so far) to be genuine causes, without the undesirable side-effect
of finding life-saving actions to be causes of the deaths they delay.
In the following passage Tony Honore makes it clear that he would
reject this suggestion:
What has to be shown in a tort action is that the defendant's
wrongful act caused the hann, in this case the victim's death. We
know from the way in which the law structures actions for wrongful
20 21
State v Scates, 50 N.C. 409 (N.C. 1858).
Jerome Hall, General Principles of Criminal Law (1st ed, Babbs-Merrill.
Indianapolis, 1947) 262.
90 (2002) 27 Australian Journal of Legal Philosophy
death that what is legally relevant is death, not death at this or that
time or plaee or by this or that process. 22
This seems to assume incorrectly that we can individuate harms
independently of when, where or how they occur. It is particularly clear in
cases in which the harm is death that we cannot draw a clear-cut distinction
between causing it on the one hand, and causing it to occur at a certain time
and place or by a certain process on the other. This is why a lawyer cannot
legitimately argue that his client's so-called causing of death was instead a
hastening of death; that he is guilty merely of causing death at a certain time
and place and by a certain process, rather than many years later in bed and
of old age.
We ordinarily think that the earlier death occurs, all else being equal,
the more of a harm it is. Furthermore, we ordinarily think of causing death
to occur at an earlier time than it otherwise would have as causing death
simpliciter. It is true that some lawyers and lay people may be reluctant to
describe a person who hastens death by a matter of minutes or hours as a
killer, especially if he or she does so with a benevolent motive. This
reluctance may be increased, if the hastening of death is the result of an
omission rather than a positive act. Does a nurse kill a patient by taking him
off life-support at his request when it is clear that he is going to die soon
anyway? Does a doctor kill a patient when she slightly hastens that patient's
death by giving him a dose of morphine with the sole intention of relieving
his pain? Of course people opposed to such practices will say 'Yes!'. I
submit that those who are in favour of them should overcome their
reluctance and agree. This shared use of terminology makes a meaningful
debate about whether or when mercy killing can be justified possible.
Sometimes we are justifiably reluctant to say something, because it is false.
At other times, and I think such cases illustrate this, we may be reluctant to
say something, because it suggests a falsehood. We may be reluctant to
describe some death-hastener's act or omission as killing, because that
would imply that he or she did something wrong, or failed to do something
right, because death is usually a very significant harm.
My suggestion that a counterfactual death which occurs later than an
actual one should always count as a different death is not an ad hoc
stipulation designed to protect the but-for test against troublesome cases of
22 Honore, above n 3, 378. Honore cites Wright (1985) above n 4, 1777-8 and
Wright (1988) above n 4, 1025-6, as authority for this claim about the way
in which the law is struetured. I will leave it to the reader to determine
whether this is a reasonable interpretation of Wright's position in those
passages. I do not think it is; though it is easy to see how they eould be
interpreted that way.
92 (2002) 27 Australian Journal of Legal Philosophy
question was not a great harm, since the victim would have died shortly afterwards anyway.
The McLaughlin Hypothetical
Not all cases of preemption in the literature are amenable to this treatment. In the well known McLaughlin hypothetical: person A seeks to kill person C by poisoning water needed by C to cross a desert, but, before C has occasion to have a drink, person B drains the poisoned water from the keg and C dies of thirst. 27 The standard view of this case is that B, and not A, causes C's death; that is, that B's action is the preempting cause and A's action is the preempted alternative. It is not essential to this hypothetical, however, that B's action hastens death. In fact it will make the case more interesting if we assume that the poison was sufficiently fast-acting that B's action delayed C's death. 28
Not everyone, however, shares the standard view of this case. Hart and Honore have long held that neither A nor B cause death. Honore has recently recanted and joined the standard view. His reason for changing his mind is) however, not convincing: My current reasoning is that 8's conduct introduces a condition, lack of water, that in the circumstanees. including the absence of an alternative water supply, is sufficient to bring about and does bring about C's death from dehydration."
But it appears that A's conduct also introduces a condition that in the circumstances is sufficient to bring about C's death, although not his death from dehydration. 30
This illustrates the fact that our considered judgements about the causes of what seems pre-theoretically to be a single event (or state, or
27
28
(1925-6) 39 Harvard Law Review 149, ISS fn 25. In McLaugWin's original example B empties the water keg and fills it with salt. Hart and Honore's Causation in the Law are responsible for the story as I am presenting it. This is the fann in which it is now usually discussed. The possibility of the preempting cause being a delayer rather than a hastener is characteristic of what the parallel philosophieal literature has called early preemption, in which the alternative (ie, preempted) process is cut off as a result of a side-efTect of the main (ie, preempting) process. This contrasts with the previous examples of late preemption, in which the alternative is cut ofT by the premature occurrence of the effect itself. See Lewis 'Causation: Postscript E' above n 5. Honore, above n 3, 378. I leave it to the reader to decide how Or whether Honore's appeal to the fact that death was by dehydration can be reconciled with the previous quotation, drawn from the same page, in which he says that "what is legally relevant is death, not death at this or that time or place or by this or that process."
Testing for Causation in Tort Law 93
omission) can depend on how it is described. For example, you close the
door while in a bad mood; as a result of your bad mood you slam the door. It seems your bad mood caused the slamming, but not the closing, even though it also seems that there is a sense in which the closing and the slamming are one and the same event.'1 Similarly, we can and should distinguish between the causes of C's death and the causes of C's death by dehydration, and between both of these and the causes of C's death in the desert. It is plausible to suppose that neither A nor B causes C to die in the desert (though the combination of their actions does), whereas B alone causes C to die of dehydration." I do not think there are any legitimate intuitions or legal principles that can decisively detennine whether B causes
C's death simpliciter. Our criteria for distinguishing between causing an
event and merely influencing how and when it occurs lead in different
directions when we consider C's death qua death. Ordinarily we do not
think that delaying death, unlike hastening it, is causing death. This suggests that neither B nor A are causes. On the other hand, we do ordinarily think that having a significant enough influence on the manner of a death constitutes killing. This suggests that B alone causes C's death.
Consequently conceptual clarification alone seems unable to detennine whether B is guilty of murder or only guilty of attempted murder. It is not surprising then that legal scholars and philosophers disagree about the issue. It seems plausible that the matter can only be resolved by considerations of policy, rather than metaphysics." That would require a
31
12
Jl
The example is from Jaegwon Kim, 'Causes and Counterfacluals' (1973) 70 Journal of Philosophy 570-572. This phenomenon has given rise to a philosophical debate. See Alvin I Goldman, A Theory of Human Action, (Prentice-Hall, New Jersey, 1970) ch I, would claim that it means that C's death and C's death by dehydration are in fact different events. Others would claim that, since C's death and C's death by dehydration are obviously the same event, the most fundamental kind of causal relation must be between something other than events. Thus Jonathan Bennett has claimed that it is a relation between facts, see Events and Their Names (Hackett, 1988), and Christopher Hitchcock has claimed that it is a relation between events-in-contrast-to-altematives, see <The Role of Contrast in Causal and Explanatory Claims' (1990) 85 Synthese 395-419. I would argue that we can (and should) accept that there is a sense in which C's death and C's death by dehydration are different events, while also doing to justice to the intuition that they are the same event, see my 'Preempting Preemption' above n 17. Some sense of how this is possible can be gained by comparing it to the 'issue' of whether London and Greater London are different cities. The last-wrongdoer rule, for example, which was explicitly justified entirely in terms of policy. states that the wrongdoer closest in time to the effect was alone responsible for it. This implies that 8 is guilty of murder. The last-wrongdoer rule is discussed in Laurence Eldredge, 'Culpable
Testing for Causation in Tort Law 95
This approach may not be applicable to the McLaughlin
Hypothetical. But if the details are filled out in such a way that the victim's
death would have been just as significant a harm were it not for B's action, I
submit that we should say that B did not cause the victim any harm. This
can be hard to see, because there is considerable (though not, I think,
decisive) intuitive appeal to the idea that B alone causes C's death. Hence
we may be tempted to argue that since C's death was a harm, B caused that
harm. The persuasive power of the argument will be undermined, however,
if we remember that legitimate intuitions about an event's causes can
depend on how that event is described. We should be ready to distinguish
between the factors that caused C to die, and those that caused C to be
harmed.
Wright has called appeals to the details of the harm in cases of
overdetermination nothing more than "proof by tautology"; suggesting that
prior to deciding which details are relevant and which are not one must
already have made a decision about the issue the test is supposed to
determine, that is, the causal status of the wrongs. 35 have tried to show that
Wright is mistaken. Instead we should decide which details of the harm are
relevant, by considering its identity conditions qua harm. I submit that once
we do so any appearance of conflict between our best causal judgements
and the determinations of the but-for test will disappear. In the landmark
case of March v Slramere the Australian High Court held that the but-for
test was not conclusive. Instead it was decided that causation should be
determined by ordinary notions of language and common sense." If I am
right, there is no conflict between the but-for test and ordinary language or
common sense.
"J6 Wright, (1985) above n 4, 1777-78, and Wright, (1988) above n 4, 1025. (1991) 171 CLR 506.
