Understanding Causation and Attribution of Responsibility

Presented at the Commercial Conference of the Supreme Court of Victoria/University of Melbourne, Banco Court

Introduction and summary

The underlying theme for today’s conference is causation. When Justice Digby kindly invited me to speak on causation I had just concluded an article, which was published earlier this year, entitled "Unnecessary causation" (2015) 89 Australian Law Journal 1. My presentation today draws heavily from that article, although some arguments are refined. My central thesis is that the metaphysical concept of causation (the core causation enquiry is metaphysical, not factual) should be understood only in one sense. That sense is necessity. An event will only ever be a cause of an outcome if the event is necessary for the outcome. That is, causation requires that the outcome would not have occurred "but for" the event. Or, to put the proposition negatively, the event is not a cause of an outcome if the outcome would have happened anyway.

There are two broad points that I will make in this paper.

First, I will explain why I believe that the only meaning of causation is "necessity" or, in the common parlance, a test of "but for" causation. I will also explain reasons why judges have been reluctant to embrace this meaning.

Secondly, I will explain that although this test of causation is simple and elegant it does not replace the difficult normative questions that must be asked. Instead, it makes those questions more transparent. They are as follows:

(1) Since causation is concerned with a relationship between "event" and "response", how do we characterise the relevant "event" and the relevant "response"?

(2) If causation is found to exist, what principles should be applied to determine whether responsibility should be imposed?

(3) If causation is not found to exist, should responsibility be imposed in any event?

 

(1) The meaning and test for causation

The "common sense" approach in Australia after 1991

I start with the leading causation decision of the High Court of Australia in relation to the law of torts. This decision posed a test for causation which I respectfully submit may be in decline. This is the "common sense" test of causation. Although its genesis is much earlier, the "common sense" approach to causation has been well known in Australia since March v Stramare.

In March v Stramare, an intoxicated and speeding driver collided with a truck which was parked at night, with hazard lights, in the centre lane of a six-lane road. The High Court unanimously held that the truck driver and his employer were liable.

On an application of the "but for" test, the answer to the causal inquiry was simple. The truck driver’s carelessness was necessary for the speeding driver’s injury, and but for the truck driver’s negligence the speeding driver would not have suffered the losses that stemmed from his injury. As McHugh J explained:^[1]

when the damage suffered by a plaintiff would not have occurred but for negligence on the part of both the plaintiff and the defendant, a conclusion that the defendant’s negligence was not a cause of the damage cannot be based on logic or be the product of the application of a scientific or philosophical theory of causation. It has to be based upon a rule that enables the tribunal of fact to make a value judgment that in the circumstances legal responsibility did not attach to the defendant even though his or her act or omission was a necessary precondition of the occurrence of the damage.

This approach to causation accords with linguistic use. In particular, it is unlikely that anyone would claim that A caused B if A made no difference to whether or not B occurred.

However, this approach by McHugh J did not command the support of the other members of the High Court. The leading decision was given by Mason CJ, with whom Toohey and Gaudron JJ agreed. Although their Honours all agreed with McHugh J that the truck driver was liable, Mason CJ preferred a "common sense" approach to the issue of causation in preference to the "but for" test.^[2]

The same "common sense" approach is taken in criminal law. In Royall v The Queen,^[3] a majority of the High Court considered the meaning of causation in the context of s 18(1)(a) of the Crimes Act 1900 (NSW). That statute described an act of an accused person 'causing the death charged' committed in particular circumstances. A majority of the court, adopting the approach from Burt CJ in Western Australia,^[4] held that it was sufficient if a jury were told that the question of causation was not a philosophical or scientific question, but that it was 'a question to be determined by them applying their common sense to the facts as they find them, they appreciating that the purpose of the enquiry is to attribute legal responsibility in a criminal matter'.^[5]

Problems for common sense causation

There are several problems with the "common sense" test for causation.

First, as Dixon J of the Victorian Supreme Court recently observed with great cogency, the 'common sense' approach is not a legal test.^[6] Instead, the common sense approach encourages a pure form of top down reasoning. Judge Posner famously illustrated this idea by reference to concepts of top down and bottom up reasoning. Top down reasoning describes the process by which the legal scholar or judge develops a theory and then uses it to organise, criticise, accept or reject decided cases. In contrast, a scholar or judge using 'bottom up' reasoning starts with the cases and moves from there (usually not very far).^[7] In the jurisprudence of constitutional law, 'top down reasoning' has become a term of derision.^[8] A "common sense" approach appeals to intuition. It suggests that the judge ought to reason downwards from the intuitive sense of a conclusion. As Gummow J extra-judicially expressed the point, '[t]o proceed on the footing that the law ought to be X, and that the law is therefore X, and any decision of an ultimate appellate court to the contrary is therefore in error, and to teach students accordingly, is unsatisfactory'.^[9]

I should emphasise that, unlike some theorists, I do not say that top down reasoning is always illegitimate. Indeed, the development of the common law always requires some departure from pure 'bottom up' reasoning. This is not to endorse reasoning to a result by reference to some preferred social policy. Instead, it is to accept, as Posner explains, that the difficulty with pure bottom up reasoning is that it begs the question of how a legal scholar is able to reason from one case to another without some conception of theory, system, or principle independent of the particular cases.^[10]

Indeed, the "common sense" approach is not actually "common" sense. In March v Stramare itself, the sense of the result was not common between the High Court and the Full Court of the Supreme Court of South Australia. Indeed, almost all of the difficult cases of causation which reach ultimate appellate courts do so because the "sense" of the result is not "common".

Secondly, the common sense approach is, in part, based upon a linguistic error. The concept of 'common sense' causation arguably would not have survived without the powerful support of Professors Hart and Honoré. In a widely read work, they argued that a common sense approach to causation could be deconstructed, although conceding that there would be a penumbra of uncertainty.

Professors Hart and Honoré asserted that 'cause' in everyday speech means more than a 'but for' or necessary condition.^[11] I doubt whether this is correct. An example they gave is where a fire has broken out. They say that the lawyer, the historian, and the 'plain man would refuse to say that the cause of the fire was the presence of oxygen'.^[12] But it is misleading to speak of the cause of the fire. There were multiple (necessary) causes. Each of the lawyer, the historian, and the 'plain man', aiming for some precision, would surely have no difficult in saying that the causes of the fire were holding a lit match to paper in the presence of oxygen. Alternatively, as John Stuart Mill put it, the 'whole cause' includes all necessary conditions.^[13]

Thirdly, the reasons why the common sense test was adopted in March did not require that test. In March, Mason CJ gave a number of examples of situations in which he considered that causal questions were affected by factors other than the 'but for' test:^[14]

(i)  Where a factor which secures the presence of the plaintiff at the place where and at the time when he or she is injured but the risk of the accident occurring at that time was no greater.^[15] An example of this is a taxi driver who is dangerously speeding in breach of conditions of contract with the customer and, had he not been speeding, the taxi would not have been in the position where it was hit by a falling tree.^[16]

(ii)  Where a superseding cause, sometimes described as a novus actus interveniens, is said to 'break the chain of causation' which would otherwise have resulted from an earlier wrongful act.^[17] So, for instance, in M'Kew v Holland^[18] a defendant's negligence injured the plaintiff's leg but the plaintiff's subsequent action in attempting to descend a steep staircase without assistance or a handrail was held to 'break the chain of causation'. Professors Hart and Honoré also argued that novus actus interveniens is an example where a necessary event is not a cause. They give an example of a person who provides arsenic to another who uses it to poison a victim. They suggested that 'the causal explanation of the particular occurrence is brought to a stop when the death has been explained by the deliberate act'.^[19]

(iii)  Where there are two or more acts or events each of which would be sufficient to bring about the plaintiff's injury.^[20]

There is a simple and clear answer for why Mason CJ's points (i) and (ii) do not present problems for the 'but for' test. As Professor Stapleton has argued, the law must distinguish between questions that are concerned with causation and questions that are concerned with the scope of liability for consequences.^[21] Professor Stapleton explains, footnoting March, that courts unfortunately conflate questions that are concerned with the scope of liability for consequences with questions of causation.^[22]

Professor Stapleton considers that point (iii) is an example of causation, although one which does not require necessity.^[23] This is an awkward approach. It amounts to saying that 'causation' embodies two fundamentally different concepts. In many cases causation means that an event must be necessary for an outcome. But then the same concept of causation permits an outcome to be treats as caused by an event even if the relevant outcome would have been exactly the same without the event. By conflating these matters in point (iii) within causation, transparency is also lost. There is no longer any need to explain why a person should be responsible for an event that would have happened anyway despite his involvement.

A better approach would be for point (iii) also to be treated as a legal rule arising independently of the metaphysics of causation. It is an example where causation is unnecessary. Like many other examples considered below, it requires justification for why causation is either replaced by a different rule, or disregarded.

There are significant signs that the law is moving towards an acceptance of a necessity test for causation, that is a "but for" test. In particular, the Civil Liability Acts in Australia provide for the causation element for liability, when required, to be one of 'necessity' or 'but for' causation. Although the legislation also includes 'scope of liability for consequences' under the rubric of causation, it is clear that this is a separate enquiry from the necessity enquiry. As the High Court said of the New South Wales Civil Liability legislation in Wallace v Kam,^[24] it ‘involves nothing more or less than the application of a "but for" test of causation’.

The various Civil Liability legislation also recognises that there can be possible exceptions to causation. As I will explain, this is a very desirable approach. By identifying the single but-for causal rule, courts are forced to confront the reasons for imposition of liability for an event even if the event was not necessary for the outcome. As it turns out, there are numerous such instances in the law. All of them need to be justified.

 

(2) Making transparent the normative questions

At the start of this paper I mentioned that causation

(1) Since causation is concerned with a relationship between "event" and "response", how do we characterise the relevant "event" and the relevant "response"?

(2) If causation is found to exist, what principles should be applied to determine whether responsibility should be imposed?

(3) If causation is not found to exist, should responsibility be imposed in any event?

(1) The normative question of characterisation of event and outcome

One difficult issue for causation is the characterisation of the event and the outcome that must be causally linked. That characterisation process involves normative questions.

Take an example derived from the facts in the United States Supreme Court decision in Burrage v United States.^[25] That case concerned a statute which contained the phrase 'death …resulted from the use of th[at] substance'. In that case, Mr Burrage provided the heroin used by Mr Banka, a long time drug user. Mr Banka died from a drug overdose after an extended drug binge including the heroin. The expert evidence was that Mr Banka may have died even if he had not taken the heroin. Giving the opinion of the court, Scalia J explained that the expression 'results from' should bear the ordinary causal meaning of 'but for' causation. He assumed that the reference to "that substance" was a reference to the heroin only. On that approach, Mr Banka's death had not been caused by the use of the heroin. The relevant event was "use of a substance, namely heroin only" and the relevant outcome was "death". But if the event were changed to be "the use of any prohibited substances" then the heroin and other prohibited substances did cause the death.

Another example is Performance Cars Ltd v Abraham.^[26]In that case, Mr Abraham was found to have carelessly driven into the Rolls Royce owned by Performance Cars, he infringed the rights of Performance Cars. Mr Abraham was lucky. The same panel of the Rolls Royce had been previously damaged by another wrongdoer who was liable to pay for the repairs. The Court of Appeal rightly said that Mr Abraham was a wrongdoer. If the relevant question for causation was whether Mr Abraham’s conduct had caused denting of the Rolls Royce panel then the answer is "yes". However, if the question is whether Mr Abraham’s conduct had caused financial loss to Performance Cars then the answer is "no". Loss is essential to a claim for negligence so Mr Abraham was not liable to pay damages for a car that had previously been damaged.

On the other hand, outside the law of negligence it has sometimes been possible to characterise the relevant outcome as the injury rather than the loss that has been suffered and to find that a substantial award is required to vindicate the plaintiff’s rights even if no loss has been suffered. For some time, these damages were described as "vindicatory damages".^[27] They attracted ferocious academic defence.^[28] But the contrary result was reached by the United Kingdom Supreme Court in Lumba v Secretary of State for the Home Department.^[29]In that case, the appellants were unlawfully detained pending deportation because their detention was under an unlawful blanket policy. Six justices held that if the policy had been lawfully applied then the appellants would have been detained in any event and therefore they suffered no loss and there was no justification for an award of "vindicatory damages". These damages, to the extent to which they continue to exist, were confined to "constitutional rights."^[30] That decision has been criticised by one academic who argues that it confused "the nature of the wrong, effectively treating the illegality of the detention as the wrong" rather than the violation of a right to liberty.^[31] In Australia, a case raising similar issues was last month given special leave to appeal to the High Court of Australia.^[32]

(2) The normative question of whether liability should be denied despite causation of injury

I will skip over this point briefly because it is uncontroversial. Although different concepts can apply in different cases to deny liability there are a number of circumstances in which liability is denied even though causation of loss exists. For instance, liability might be denied because there is no duty. Or liability might be denied because the injury, or the loss, about which complaint is made was not within the scope of the duty owed. Or liability might be denied for the extent of the loss claimed because that extent of loss is too remote, or involved a novus actus interveniens or was otherwise not within the scope of liability for the consequences claimed.

(3) The normative question of whether liability should be imposed despite the absence of causation of injury or causation of loss

There are a large number of instances where liability is imposed despite the absence of causation. In some cases, liability is imposed despite the absence of causation of loss. Obvious examples are instances where a defendant owes a debt to a plaintiff. It is irrelevant whether the defendant would have squandered the money if it had been paid, or if the money would have been stolen or lost. Causation of loss is not required because loss is not required. Slightly more controversial is the application of the same approach to cases involving the accountability of a trustee or company director as a custodian of assets.

Other well-known examples where liability for loss is imposed even if the defendant was not necessary for the loss (and, in that sense, a cause) include instances of multiple tortfeasors and cases of deceit.

An example of multiple tortfeasors is Kuwait Airways Corporation.^[33] Iraqi Airways committed the tort of conversion by taking possession of planes belonging to Kuwait Airways. The Kuwaiti planes had been brought to Iraq by Iraqi armed forces after the 1990 invasion of Kuwait. When they were in Iraq they were held by Iraqi Airways. The Kuwaiti planes were later destroyed by the coalition bombing of Mosul. Kuwait Airways sued Iraqi Airways for damages for conversion. Iraqi Airways argued that the planes would have been lost to Kuwaiti Airways even if they had not been converted by Iraqi Airways. 'But for' the wrongdoing of Iraqi Airways, the loss of the planes would still have occurred as a result of the prior wrongful act of conversion by the State of Iraq.^[34] Hence, it was argued, Iraqi Airways should not be liable to pay damages. The argument failed. The House of Lords held that Iraqi Airways was liable to pay damages. Lord Hoffmann, who sat on the court, later described the decision of the House of Lords as being based on the notion that 'it was not necessary that the conversion should have caused the loss. If you convert someone's property you have to pay for it or give it back'.^[35]

The difficult question then is why causation of loss is unnecessary for intentional wrongdoing that deprives a person of possession. One possible answer, although not without difficulty, is provided by Dr Douglas.^[36] He argues that by abandoning the requirement of causation (but for) in cases of strict liability torts prevents strict liability from becoming meaningless. Otherwise, Douglas suggests, the focus would shift from the intentional nature of the conduct, however honest and reasonable, to questions of blameworthiness.

Another example is the tort of deceit. A wrong has occurred but it is not necessary for the plaintiff to prove that the misrepresentation caused the loss that was suffered. The classic statement of this position in relation to deceit is Edgington v Fitzmaurice.^[37]In that case, the plaintiff lent money to a company due to his mistaken belief that the loan was secured by a charge. He also relied on statements in a prospectus that were fraudulently made by the directors. The Court of Appeal was not concerned with whether the plaintiff would nevertheless have lent the money but for the deceit. As Bowen LJ explained, '[t]he real question is, what was the state of the [p]laintiff's mind, and if his mind was disturbed by the misstatement of the [d]efendants, and such disturbance was in part the cause of what he did'.^[38] This approach has been applied on many occasions. For instance, in Gould v Vaggelas,^[39] Brennan CJ spoke of the need for a misrepresentation to be 'one of the real inducements to the plaintiff to do whatever caused his loss'.

In Standard Chartered Bank v Pakistan Shipping Corporation (Nos 2 & 4),^[40] Lord Hoffmann said that the law 'takes no account' of reasons that influence a person to act other than the material misrepresentation because it 'would not seem just that a fraudulent defendant's liability should be reduced on the grounds that, for whatever [other] reason, the victim should not have made the payment which the defendant successfully induced him to make'.^[41] A broad appeal to 'sound policy' and 'justice' is not an explanation for the absence of a causal rule. A better answer might be to say that the law's concern with individual autonomy is such that a fraudster will be liable for losses that he or she has caused, or losses to which he or she has contributed. The need to protect autonomy must be the factor that justifies the latter extension. The earliest cases that justified the absence of a causal rule did so on the basis that it was impossible to enquire into contributions to a person's mind: '[w]ho can say that the untrue statement may not have been precisely that which turned the scale in the mind of the party to whom it was addressed?';^[42] ''How is it possible to say in what manner the disclosure would have operated on Kay's mind';^[43] 'You cannot weigh the elements by ounces'.^[44] It may be that this rule is now too well established to be disturbed. But the premise might be questioned. Are people always incapable of weighing relative contributions to their decisions? Suppose the plaintiff in Edgington had given evidence that although the fraudulent statements by the defendants were a part of his decision making process, he would have lent the money in any event because of his belief that it was secured by a charge'. If a person is capable of giving that evidence, and making that assessment, then some other rationale might need to be found for the replacement of causation in this context with a rule of material contribution.

Each of the examples I have given so far involves departure from a necessity test of causation for reasons which have been well accepted in the law even if those reasons might be debatable in theory. In the case of a debt, no causation of loss is required. In the case of multiple tortfeasors and deceit, the test of causation is replaced by a test of contribution. In the language used by the High Court of Australia, the test is one of causation or material contribution. As Gummow, Hayne and Crennan JJ said in Amaca Pty Ltd v Booth^[45]

The 'but for' criterion of causation proved to be troublesome in various situations in which multiple acts or events led to the plaintiff's injury,^[46] for example, where the development of a particular medical condition was the result of multiple conjunctive causal factors. In such cases what may be unclear is the extent to which one of these conjunctive causal factors contributed to that state of affairs. These situations have been addressed by the proposition stated by Lord Watson in Wakelin v London & South Western Railway Co^[47] that it is sufficient that the plaintiff prove that the negligence of the defendant 'caused or materially contributed to the injury'.

There are, however, cases at the margins where liability is imposed despite the usual requirement for a causation test, and despite the absence even of any proof of material contribution. One such case came before the House of Lords which involved a situation where multiple employers had exposed an employee to asbestos.^[48] No employment could be proved to have been necessary for the employee's subsequent mesothelioma. The House of Lords was asked if any employer 'caused' the mesothelioma. When the appeal books were received, Lord Hoffmann went in to Lord Rodger's chambers to speak with him about the Roman debate on this question. The brilliant Alan Rodger instantly recalled Digest 9.2.11.2 where Ulpian, quoting Julian, recounts the solution to such a scenario under chapter 1 of the Lex Aquilia: if several people strike a slave and one cannot tell whose blow killed him, all are liable.^[49]

When the House of Lords heard the case in 2002, it was generally (perhaps incorrectly) assumed that all the employers had committed a wrong, much like all the persons who struck the slave. The focus of the enquiry was on whether the employers should all be liable for the full loss caused by the mesothelioma where the evidence accepted was that the mesothelioma had been caused by a single 'guilty' fibre. The House of Lords reached the same conclusion as the Romans and held all employers fully liable in solidum.

Lord Hoffmann, later said that the decision he, and the others, had reached failed the test for acceptable law: a rational and justifiable basis to depart from normally applied principles of law.^[50] I do not venture a conclusion here to the difficult question of causation that arose because that question has not yet been finally resolved in Australia. In 2012, I was listed to sit on an appeal where this question had been raised. The appeal settled almost on the eve of the hearing. Rather than attempt to offer an answer to the question in Fairchild, I make two observations.

The first observation is that the analogy with D 9.2.11.2 was apt but Julian was not necessarily asking the same questions as the House of Lords in Fairchild. One difference is that under Roman law, the striking of the slave infringed the rights of the slave owner or, as the Romans would have expressed it, gave rise to an action. But it is not immediately obvious that a wrong was committed in Fairchild. The mere exposure of an employee to the possibility of harm (such as from inhaled asbestosis fibres) might not be wrongful. Suppose that one of the employee plaintiffs in Fairchild had not yet contracted mesothelioma. Could he still have sued his employer for exposing him to the possibility of mesothelioma? If not, then Fairchild was more like the problem of the two hunters in Cook v Lewis and less like the case of multiple people striking the slave. Another difference between D 9.2.11.2 and Fairchild is that in Fairchild the House of Lords was asked whether each defendant was liable for all losses arising from mesothelioma. In D 9.2.11.2, Julian asked only if the person striking the slave was liable. The discussion of the quantum of liability was different.^[51] Subsequent to Fairchild, the question of liability was put differently before the House of Lords: was the employer liable for increasing the chance that the employee would suffer loss.^[52]

The second observation is to reiterate Lord Hoffmann's most powerful point: if a common law claim is brought for loss suffered that was caused by wrongdoing, then before a court departs from the requirement that the wrongdoing was necessary for the loss (and hence abolishes or replaces the rules of causation) there should be a rational and justifiable basis in principle for doing so. A re-orientation of causation requiring focus only upon necessity would permit these questions of principle to be exposed, analysed, and, if possible, justified rather than concealed within counter-intuitive assertions of a multifarious notion of 'causation' or within the broad rhetoric of 'common sense'.

 

Conclusion

There are two short points of this paper. The first is to suggest that causation has only one meaning. That meaning is necessity which is applied by a test, as lawyers commonly call it, of "but for". The second point is to emphasise that this apparently simple test is not a simple solvent for the question of whether liability should be imposed. But it does make the liability questions more transparent.

---