Should Tort Law Demand the Impossible?

Should Tort Law Demand the Impossible? I. Introduction It is widely assumed by philosophers that ‘Ought’ Implies ‘Can’ (hereafter OIC). OIC holds, roughly, that if one ought to x then one has the ability to x; or, formulated negatively, that the inability to x implies that it is not that case that one ought to x. OIC is usually discussed in the context of morality, but is also relevant to legal theory. If OIC constrains moral duties, we might think, presumably it constrains legal duties too. Admittedly, there is no straightforward connection between moral and legal duties: not all moral duties are legally enforced and not all legal duties enforce moral duties. Nevertheless, since both law and morality are in the business of imposing duties tout court, shouldn’t fundamental constraints like OIC apply to both? Tort theorists should be particularly interested in this question, since two features of tort law – strict liability and the objective standard of care – are prima facie in conflict with OIC. In fact, some theorists explicitly reject the maxim after reflecting on the normative character of tort law. Gardner 2001, 119. To frame the discussion ahead, let’s state two arguments that purport to show that tort law justifiably violates OIC: The Argument from the Objective Standard of Care: (1) Negligence imposes duties to adhere to the objective standard of care. (2) The duties specified in (1) sometimes require agents to do what they cannot. (3) The objective standard of care sometimes violates OIC. (4) The objective standard of care is justified. (5) The objective standard of care is justified in sometimes violating OIC. The Argument from Strict Liability: (1) There are reasons to x, regardless of one’s ability to x. (2) The reasons specified in (1) sometimes create duties to x, regardless of one’s ability to x. (3) The duties specified in (2) sometimes violate OIC. (4) Strict liability imposes the duties specified in (2). (5) Strict liability sometimes justifiably violates OIC. I will try to show that neither of these arguments succeeds. In doing so, I hope to defend one important connection between law and morality that I call the Functional Convergence Thesis. I will elaborate on this claim later but, in brief, it is that both legal and moral duties should be constrained by OIC because of their integral function for prospective practical reasoning. Here is our route that that conclusion. In section II we clarify OIC and consider underlying justifications for it. In section III we discuss how strict liability and the objective standard of care can be interpreted as violating OIC. In section IV we reject the Argument from the Objective Standard of Care. In section V we reject the Argument from Strict Liability. Finally, in section VI we defend the Functional Convergence Thesis. II. Ought Implies Can Clarifications There are several ways that OIC can be further clarified. I will focus on clarifying the ‘ought’ and ‘can’ terms in OIC. I will leave aside the interpretation of the ‘implies’ term. For more on this, see Sinnott-Armstrong 1984, 93. The ‘ought’ term, as I will understand it, ranges over practical reasons and duties. It is therefore broad enough to constrain both prudential and moral reasons. One cannot have a self-interested reason to perform an impossible act any more than one can have a moral reason to do so. However, this interpretation is narrow enough to exclude purely evaluative ‘ought’ statements, such as ‘it is bad when good people suffer’. On the distinction between prescriptive and evaluative statements, see Barcan Marcus 1966, 580-2; Humberstone 1971, 8-11; and Harman 1977. Whilst this claim may imply a reason in the right circumstances, by itself it merely draws attention to a bad feature of the world without stating that an agent has a reason or a duty to do anything. The second important question regarding the formulation of OIC is how we interpret ‘can’. Interpreted too broadly, OIC fails meaningfully to constrain the reasons and duties that apply to us. But interpreted too narrowly it rules out too much. If the requirement is that an alternative action is possible for the agent, or that her action is caused by a decision that is not subject to causal determination, then OIC is true only if determinism is false. The truth of determinism may threaten OIC in this way, but the possibility of wider definitions of ‘can’ should encourage us to be cautious. Defending a compatibilist conception of OIC is beyond the scope of this chapter. For those who think that OIC is inconsistent with determinism and determinism is true, the arguments from strict liability and the objective standard do not pose a special challenge. As an example of the perils of narrow definitions of ‘can’, consider objections to OIC based on excused wrongdoing. Critics argue that psychological conditions like kleptomania demonstrate that OIC is false, since the compulsive thief acts wrongly, albeit non-culpably, even though she cannot avoid stealing. But these are dubious counterexamples as it is unclear whether the kleptomaniac lacks the ability to refrain from stealing. Compulsions typically impair rather than destroy our abilities and compulsive people often regulate their behaviour in some situations. The success of the counterexample depends on the precise definition of ability and the empirical assumptions we make about kleptomania, which can distract us from the core considerations that determine the validity of OIC. If we want to know whether OIC is true, it is better to challenge it with less controversial examples of inability. In keeping with this broad understanding, I will not provide a specific analysis of ‘can’. As far as possible I will use uncontroversial examples and acknowledge when competing conceptions of ability yield different implications. Justifications Despite the widespread acceptance of OIC, there is no consensus about its justification. A number of arguments have been offered in its defence. One is that OIC must be true because it is the best explanation of other intuitive judgements. For an elaboration of this argument, see Howard-Snyder 2006. Suppose that a friend is trapped under a boulder. You recognise that it would be best to lift the boulder and save your friend, but you lack the strength to do this. Instead, you run to fetch help. If you had the strength, you would be obliged to lift the boulder. Given that you cannot do this, you ought to run and get help. OIC offers the best explanation for the existence of this second-best obligation: it arises because a duty to perform the best action is ruled out by OIC. This argument supports OIC by identifying implications that we independently accept. One limitation of this approach is that, although plausible implications provide evidential support for the principle, they do not reveal its underlying rationale. This is problematic, at least for our purposes, as the rationale for OIC might bear on the question of whether it should apply in other normative contexts such as tort law. It is therefore helpful briefly to distinguish different justifications. One view is that OIC is justified because it is unfair to require people to do what they cannot. This defence is intuitive but is flawed as a general justification. Morality and law place unequal burdens on agents in other areas (we will return to this idea later). Morality distributes costs unevenly through the operation of two forms of luck. See Nagel 1979. First, circumstantial luck determines the bundle of rights and duties one possesses at any time. Some individuals may find themselves burdened with more obligations than others due to circumstances beyond their control. Second, constitutive luck refers to the influence of luck on our dispositions, attitudes and personality traits. To some extent, these constitutive features of our character are beyond our control and they affect the costs that different agents bear to fulfil the same duties. The saint may fulfil her duties gladly, whilst the sinner must battle powerful cravings, but morality does not impose lesser duties on the sinner for that reason. A different problem is that OIC constrains reasons as well as duties. But it is not unfair to posit a reason to do the impossible if this reason is not decisive, since reasons do not bind. Even decisive reasons do not necessarily bind in the moral sense. See Tadros 2016, 15-16. A related point is that OIC also constrains prudential reasons: there is no reason to turn one’s hat into money and buy a car, just as there is no reason to turn one’s hat into money and donate to charity, yet fairness cannot explain why prudential reasons are also ruled out by the impossibility of acting on them. These observations suggest that fairness is, at best, a partial justification. Some defend the maxim by observing that duties to do the impossible are pointless. For this view, see Streumer 2007, 365. However, it should be noted that Streumer does not see his argument in purely instrumentalist terms. One way to understand this claim is that pointless duties have undesirable consequences. Either they will not be fulfilled or, worse, agents will waste time with failed attempts to fulfil them. As we will see later, critics of this view have a response: positing a duty does not necessarily imply an attempt to fulfil it. But there are other reasons to doubt this view. Rejecting OIC might have better consequences in certain circumstances. Imposing duties that people are sometimes unable to fulfil might induce them to exercise more productive care than a duty that imposes lower expectations. But, intuitively, OIC is a general constraint and does not vary according to these contextual factors. We should not look to pragmatism to explain the deep significance of OIC for our practical reasoning. The pointlessness objection should be understood as highlighting a more substantial defence of OIC. There is a deeper way in which OIC allows reasons and duties to be action-guiding. What motivates this argument is not the worry that people will attempt the impossible, but that the potential for guiding action is a crucial feature of reasons. From a deliberative perspective, an agent cannot consult her reasons to decide what to do if they require the impossible. The constraint placed on what constitutes a practical reason reflects the potential of all reasons to be instantiated in action. During the deliberative process, an agent assesses various reasons for action. In order for reasons to be practical and function as reasons to act (as opposed to reasons to believe, feel and so on) they must have the potential to be instantiated in action. If they do not, the function of practical deliberation as a process geared towards deciding how to act would be undermined. This potential has nothing to do with probability: we can have a reason for an action even if it is unlikely to be acted upon due to cost, difficulty or weakness of will. One the other hand, if an agent has no prospect of performing some action, there cannot be a reason to do it if her deliberation is to maintain its function as a process geared towards deciding what to do. Unlike the fairness argument, this is not a normative defence of OIC. A contrast is sometimes drawn between normative and semantic interpretations. Carter 2001, 79-80. On a semantic interpretation, ‘ought’ presupposes ‘can’, in the sense that ‘ought’ questions do not arise regarding impossible actions. The functional argument does not insist on the validity of OIC for purely semantic reasons, however. It is not based on any claim about the meaning of the word ‘ought’. Instead, reasons that violate OIC lack a feature that is fundamental to their role in practical deliberation. This could be taken as evidence that ‘ought’ semantically implies ‘can’, since it would be curious if the meaning of the term did not reflect this function, but the argument does not rely on this semantic claim. Reasons to do the impossible are not necessarily unintelligible on this view. They are ruled out because they cannot perform their central function in practical deliberation. A virtue of this justification is that it answers the argument sometimes made by critics of OIC that rejecting the maxim does not require giving up the idea that obligations are action-guiding. Talbot 2016 and Gardner 2013. According to these critics, unfulfillable obligations can guide us by requiring a host of ex post actions, such as apology or repair. The problem with this view is that these obligations cannot offer ex ante action-guidance. A separate but important point is that unfulfillable obligations do not plausibly offer ex post action-guidance either, since secondary obligations need not arise breach of primary obligations. For further details, see the discussion of Gregory Keating’s account of strict liability duties in section III. And ex ante action-guidance is integral to the way reasons and duties function in the deliberative process. When an agent is deliberating about what to do ex ante, it is not enough for her to know that her unfulfillable obligation may guide her action retrospectively. On the functional account, duties must be prospectively action-guiding to fulfil one of their central roles in practical deliberation. III. Does Tort Law demand the Impossible? Both the Argument from the Objective Standard of Care and the Argument from Strict Liability have interpretative as well as normative elements. Steps (1) and (2) in the former make claims about the correct interpretation of negligence duties. Step (4) in the latter makes a claim about the correct interpretation of strict liability. Our question is whether tort law should demand the impossible, not whether it does. The latter is an interpretative question that requires much greater attention than I can dedicate to it here. The answer also depends on the specific contours of the duties in question, which may vary by jurisdiction. Nevertheless, it will be instructive to make some general points to understand the prima facie tension between OIC and relevant aspects of tort law. Let’s begin by considering the objective standard of care. The standard is objective in the sense that it depends on what a hypothetical reasonable person would do or refrain from doing in the relevant context. The standard expected of a defendant is not typically reduced by her personal lack of prudence, skill or knowledge. As a result, an individual who is unable to meet this standard due to limitations for which she is not responsible may still be liable for negligence. We must be careful here. The objectivity of the standard does not necessarily demand the impossible. It certainly allows that a person need not be morally blameworthy for her failure to possess the appropriate abilities. But it is consistent with this standard that such a person could have avoided breaching her duty. Similarly, we should not confuse impossibility with low probability. An amateur footballer is less likely than a professional to score a penalty kick, but even if the amateur scores 3 in 10 whilst the professional scores 9 in 10, both have the ability to score in the sense that failure was not unavoidable for either of them, even though it was more likely for the amateur. Most formulations of the objective standard do not distinguish between actions a person cannot perform and those she can perform with greater cost or difficulty, which makes it difficult to determine whether it conflicts with OIC. Consider these legal truisms: (1): A person is negligent if they do something a reasonable person would not do or fail to do something a reasonable person would do. (2): The standard of care does not vary according to the limitations and idiosyncrasies of the defendant. Neither formulation directly shows that OIC and the objective standard are in conflict. Reference to ‘meeting a standard’ is also confusing because it does not describe a single action, but rather the performance of a range of actions. Whether a person can meet a standard depends on how their abilities affect their actions across a range of cases. To illustrate, suppose that a junior doctor knows how to perform a procedure but is inexperienced and does it incorrectly 7 times out of 10. Suppose that a second junior doctor lacks the required knowledge to perform the procedure and does it incorrectly 10 times out of 10. Finally, assume that a reasonable doctor could perform the procedure correctly 9 times out of 10. In that case, both junior doctors fail to reach the reasonable standard of care if we judge their actions across a range of cases. Nevertheless, the first doctor sometimes performs the procedure correctly and thus imposing a duty on her to do so would not demand the impossible. The general claim that a person is unable to meet a standard is misleading if we mean only that it is difficult, costly or unlikely that she will act reasonably. Such a standard may offend against principles of fairness, but it does not violate OIC. What about the second junior doctor? Since she performs the procedure incorrectly 10 times out of 10 (because she lacks necessary knowledge), wouldn’t imposing a duty on her to meet a reasonable standard violate OIC? There are two further reasons to be cautious. First, the definition of ‘can’ in OIC may be controversial here. The doctor has the physical and psychological capacities necessary to perform the procedure but lacks the relevant knowledge. It is arguable that the action should still be considered impossible for the purpose of OIC. Possessing the physical and psychological ability is redundant if one lacks the opportunity to act. See Vranas 2007, 169. But since we have not defended a definition of ‘can’, we cannot rule out the possibility that it should be understood in a fact-relative rather than evidence relative sense, which allows that the doctor can perform the procedure. Second, whether OIC is violated in this case depends on how we specify the doctor’s duty. If the duty is to perform the procedure correctly, perhaps it violates OIC. But if the duty is to avoid engaging in an activity that she cannot perform to the appropriate standard, then this duty need not violate OIC since the doctor could have refrained from engaging in the activity at a prior time. But this is not a plausible way to understand the duty. It is obviously permissible to become a doctor, even though this involves knowingly engaging in risky activities that one may lack the knowledge to perform to a reasonable standard. We may say that one’s prior opportunity to avoid placing oneself in this position justifies the imposition of the objective standard, but this does not change the fact that the duty violates OIC. Cases in which the defendant’s inability to meet the standard stems from a physical impairment rather than lack of knowledge are also difficult to assess. For example, in the English case of Mansfield v Weetabix Ltd, [1998] 1 WLR 1263, 1268 (CA). the defendant crashed a lorry into a shop owned by the plaintiffs, causing extensive damage. Although he was unaware of it at the time, the defendant suffered from malignant insulinoma, which caused him to enter a hypoglycaemic state before the crash. The Court of Appeal found the defendant not liable because he had no prior warning of his condition. Some consider it obvious that the law does not impose duties that defendants are physically incapable of fulfilling, Wright 1995, 258. but others such as Allan Beever argue that the court misapplied the objective standard in Mansfield as it allowed the subjective limitations of the defendant to determine the standard of care. Beever 2009, 80 – 81. According to Beever, the objective standard, properly understood, would have imposed a duty on the defendant even though the injurious act was committed in a hypoglycaemic state. So far we have seen that lack of knowledge and physical impairment are possible examples in which the objective standard, properly understood, violates OIC. Both examples are nonetheless controversial. Another strategy is to address justifications of the objective standard directly to ascertain whether they distinguish internally between duties that are difficult or costly and those that are beyond a defendant’s abilities. As we will see in the next section, the arguments that seek to justify the objective standard naturally cover cases of unavoidable injury, regardless of where we stand on the interpretation of the law. So even if we have some reservations about the truth of step (2) in the Argument from the Objective Standard of Care – whether negligence really does impose duties on agents to do what they cannot – we must consider normative arguments for the objective standard to be satisfied that the argument fails overall. A similar point can be made about the Argument from Strict Liability. Step (4), the interpretative claim, states that strict liability imposes straightforward duties to avoid injuring. This view is defended by John Gardner, See Gardner 2001. to whom we will return later, but others understand strict liability differently. For example, Gregory Keating distinguishes between harm-based and rights-based strict liability and argues that the former arises from harm caused in the course of justified conduct. The relevant duty is not to avoid harming tout court, but rather to avoid harming-without-repairing. This duty is then justified on grounds other than the wrongness of the injurious conduct. Keating 2004. I will not attempt to settle the debate between these two views, at least as a matter of legal interpretation. The point is to emphasise that The Argument from Strict Liability relies on the interpretative claim that strict liability imposes non-conditional duties to avoid injury, which is controversial. Still, if we are to reject the possibility of a system of tort law that violates OIC on principled grounds, we should accept Gardner’s interpretative claim arguendo and consider whether it has a sound normative basis. IV. The Objective Standard of Care We are now in a position to assess the normative elements of both arguments. Let’s begin with the objective standard of care. One argument sometimes adduced in its defence is that each person has a right to expect an equal level of care. For example, see Stevens 2007, 109. Assertions about rights distributions are normatively shallow: we need to explain why this right has priority over other putative rights that are inconsistent with it, such as a defendant’s right that she is not held to an unattainable standard. One response to this challenge invokes a principle of fairness. This unfairness may take two forms. First, it is unfair that similarly situated victims enjoy a higher degree of protection than others. If the objective standard is discarded, one injured party may receive compensation whilst another is left to burden the injury because of the characteristics of the defendant – facts for which she is not responsible nor able to control. This is not an adequate defence of the objective standard, however, since we tolerate this form of unfairness elsewhere in the tort system. A common critique of tort liability is that it excludes most harms: victims of natural diseases and disasters; those harmed by non-negligent accidents who are not covered by strict liability regimes; and those who are unable to satisfy all elements of negligence. Radical differences in how well these victims fare are often dictated by facts beyond their control. To take the fairness argument seriously, we need to know why it should operate locally to support the objective standard but ignore the many other instances in which victims of accidents fare differently for reasons beyond their control. In addition to that, reducing this unfairness between claimants creates similar unfairness between defendants. The objective standard allows defendants to fare differently depending on whether they have the ability or the opportunity to avoid being negligent, which is also sometimes outside the scope of their control. A different fairness argument holds that the objective standard is a fair compromise between the interests of defendants and claimants. It is argued that a subjective standard is best for the defendant whilst strict liability is best for the claimant. The reasonable standard of care, by contrast, represents a fair balance between these competing interests. Beever 2009, 81. One objection to this is that it is impossible to strike a fair balance between two broad groups – claimants and defendants – given relevant variations in individual cases. What counts as a fair balance between two parties to a suit depends, in part, on the features of the case and the characteristics of those parties. In cases where a person is unable to meet an objective standard, holding her to such as standard is akin to a particularly harsh form of strict liability. A reply to this problem is offered by Allan Beever: the objective standard does not take into account the relationship between the specific parties because the characteristics of each party are irrelevant if the other does not know about them in advance. Beever 2009, 86. Tort law typically deals with interactions between parties who have no prior relationship, so it is unfair for one party to appeal to her own limitations, of which the other party is unaware, to escape liability. Admittedly, knowledge of a person’s abilities and limitations changes the character of the interaction in ways that legitimately alter the standard of care. If a person represents herself as a professional, for example, she cannot later rely on the fact that she is an amateur to evade liability. But it is difficult to generalise the reasoning in this case as the defendant is responsible for creating a reasonable expectation of competence, which is not true in the other examples we have considered. In cases where the defendant does nothing to mislead others, why are the defendant’s limitations irrelevant? One possibility is that claimants are disadvantaged because they cannot plan around defendants’ limitations. But this returns us to an argument we have already disputed, since we frequently expect claimants to tolerate this form of bad luck and uncertainty. If we expect victims to let the burden rest where it falls because they are not harmed by negligence, for example, why not apply the same argument in the present case? Similar worries to those sketched above apply to all attempts to justify the objective standard by appeal to a claimant’s entitlement to expect consistency in her interactions with others. For example, Richard Wright argues that claimants have an interest in sufficiently secure expectations. Wright 1995, 258-9. No doubt claimants have such an interest, but why it should take precedence over the defendant’s interest in being held to a standard she is capable of attaining? And why be so concerned about unequally secure expectations amongst claimants with respect to the standard of care when claimants to do not fare equally in so many other ways? Appeal to fairness has not decisively favoured the Argument from the Objective Standard of Care. An alternative defence is offered by Avihay Dorfman. Dorfman’s argument is novel because it seeks to justify the objective standard with reference to the interests of defendants as well as claimants. It is an equality-based argument consisting of three theses, or facets of equality. The first facet of equality is summarised as follows: ‘Treating someone who does not meet the objective standard of care as having acted wrongfully accords with our moral intuitions concerning the respect we owe to our fellow creatures as persons with freestanding claims over their own practical lives, claims for constituting equal worth.’ Dorfman 2012, 377. Arthur Ripstein also offers an equality-based argument for the objective standard, though his approach is different from Dorfman’s. See Ripstein 1999, 84-87. Thus, the objective standard is supported by the respect we owe to defendants. The second facet of equality shows that a subjective standard of care, ‘rather than expressing respect… communicates the unappealing notions of inferiority and, more generally, hierarchy.’ Dorfman 2012, 391. By contrast, the objective standard coheres with tortfeasors’ self-conception as able and competent, and avoids patronising and insulting assumptions about their inadequacies. Of course, this creates the possibility that the disadvantaged will be further burdened with a standard of care that they are not well-equipped to meet. This motivates the third facet of equality, which requires a social effort to neutralise the inegalitarian effects of the objective standard of care. For example, accessible public transport should be provided to alleviate the harsh consequences on those motorists, such as the sight-impaired, who are not able to reach the standard of a reasonable driver. There is much to recommend Dorfman’s argument. He recognises the desirability of removing social impediments to meeting the standard of care rather than creating exceptions that risk marginalising disadvantaged groups. Dorfman also claims that the objective standard does not demand the impossible but only that those who face disabilities or limitations use whatever means are within their power to meet the standard. I am less optimistic that the objective standard can be made entirely consistent with OIC through alterations in the social environment, positive though these changes are. It is also not clear whether Dorfman’s claim that the objective standard need not demand the impossible is consistent with the rest of his argument. He emphasises that ‘the fact that people suffer from undeserved deficiency in caring skill does not yield (normative conclusions) concerning how tort law ought to respond to this state of affairs.’ Dorfman 2912, 398. There is no principled distinction in Dorfman’s view between possible and impossible demands placed on agents. The cost or difficulty of duties can be mitigated by changing the social environment, but there is no ultimate normative barrier to requiring actions of which some agents are incapable, so the tension between tort law and OIC remains. This brings us to the central question: does Dorfman’s equality-based argument offer a principled defence of the objective standard of care? The argument has some affinity with the social egalitarian critique of luck egalitarianism. See Anderson 1999 and Wolff 1998. One strand of this critique is that compensating individuals for their undeserved deficiency is disrespectful as it labels them inferior. Similarly, Dorfman argues that allowing exceptions to the objective standard because of defendants’ deficiencies fails to recognise them as equals. Conversely, holding them to the same standard respects their self-conception as persons with freestanding claims over their practical lives and avoids subjective judgements. Dorfman sometimes suggests that a subjective standard would set the terms of respect according to one’s own view of what respect requires. Dorfman 2012, 376. However, it is important to recognise that a subjective standard of care is objective in the sense that it is not fixed by each individual’s own conception of what they owe to others, but rather by an objective appraisal of their capacities. Dorfman’s argument really hinges on whether a subjective standard violates equality-based duties of respect. For this purpose, the claim that a subjective standard can be disrespectful is insufficient. It must be shown that that the right that correlates to the duty of respect is not waivable. If the right is waivable, defendants could opt for any relevant deficiencies to be considered in determining the content of their duty of care. Surely a rule that gives a defendant the option of adducing special evidence to demonstrate why she should be held to a lower standard, or else default to the objective standard, shows greater sensitivity to her self-conception and therefore more respect. Those who find exemptions disrespectful could be measured against a default objective standard simply by taking no action. But for others, offering such evidence may not be demeaning. Some defendants who are not capable of meeting the objective standard are not members of otherwise marginalised groups, such as junior professionals. Others who are members of such groups may nevertheless not find beneficial exemptions disrespectful. On the contrary, they might think that the objective standard is exclusionary as it imposes a greater threat of legal responsibility relative to others. Of course, individuals will surely vary, and some may indeed conform to Dorfman’s picture, but it is unwise to second-guess a person’s self-conception. Insofar as the law should be sensitive to it, the argument supports a waivable right rather than an objective standard. Dorfman’s claim that we have equality-based duties towards defendants with impairments or inexperience to hold them to an objective standard are questionable. When we clarify that these duties must exist regardless of whether an individual defendant’s conception of respect demands it, the claim is much less convincing. It is open to Dorfman to defend a paternalistic account of respect-based duties that is insensitive to the self-conception of defendants, but this seems contrary to the spirit of his argument. V. Strict Liability and Duties to Succeed John Gardner argues that when the law holds A strictly liability for injuring B, it asserts that A breached a straightforward obligation not to injure B. A could have avoided breaching this obligation only by not harming B. Even if A was reckless, she could have fulfilled her obligation by fortuitously not injuring B. Similarly, no degree of care on A’s part prevents her violating her duty if she does, in fact, injure B. The duties imposed by strict liability do not require agents to take extreme care or indeed the highest level of productive care. Strict liability duties are simply duties not to harm others. Gardner defends these obligations by distinguishing between reasons to try and reasons to succeed. He situates the distinction within an elaboration of Tony Honoré’s defence of outcome responsibility, which he calls the Biography Argument. Gardner 2018, ch. 2. For Honoré’s discussion, see Honoré 1988, 530. The Biography Argument starts with the observation that some outcomes are constitutive of actions. The death of a living thing, for example, is constitutive of the act of killing. If one fails to bring it about, one fails to kill. Compare this to the act of shooting: one commits this act by pulling the trigger of a gun irrespective of whether this results in a death. Once the nature of these actions is properly understood, we see that outcomes are an important part of what we do as agents. To deny this is to deny that we do almost anything and therefore to efface our agency. Gardner 2018, 65. From here, the argument moves to the view that we are responsible for the outcomes we bring about. At this point, the argument reaches a crucial juncture. Two claims that are consistent with OIC (and with which I agree) are that outcomes can be constitutive of actions and that we can have reasons to perform actions, even when success is beyond our control. For example, I may have a reason to score a goal even though the constitutive outcome for this action (the ball crossing the goal-line) is not fully within my control. The claim that I want to contest is the existence of reasons to succeed in performing some action, when the success of that action is wholly outside one’s control. This brings us to Gardner’s defence of reasons to try and reasons to succeed. Gardner explains that a reason to try to x is not a reason to x but a reason to act because the act will hopefully bring about x. It is a mistake to assume that reasons to try and reasons to succeed are identical. To support the distinction, he first argues that, if acting with a view to x-ing would not contribute to one’s x-ing, one may still have a reason to x without a reason to try to x. Consider: Drowning Person: A stands on a clifftop looking on helplessly as B drowns below. A flails and waves and sinks to her knees, desperate to find a way to rescue B. Eventually A realises that nothing she can do will have any chance of saving B. Since no amount of trying will succeed in saving B, A has no reason to try, but she nevertheless has a reason to succeed. If she had no such reason, she could walk away without compunction. The situation leaves a ‘trace’ on A’s life as an agent, and this is explained by her reason to succeed in rescuing B. Gardner 2001, 137. Conversely, one may have reasons to try that do not correspond with reasons to succeed. If A’s daughter is drowning, for example, A may have a reason to try and save her as an expression of love, even if such an attempt is doomed. Gardner 2001, 137 – 8. Finally, Gardner argues that reasons to succeed are primary in the sense that the intelligibility of reasons to try depends on the existence of reasons to succeed. The act of trying logically requires something that one is trying to achieve. If one has no reason to x, one cannot intelligibly try to x, as one cannot act for the reason that what one does will bring about x. Thus, if there are no non-derivative reasons to succeed then there can be no derivative reasons to try. We need not quarrel with Gardner’s claims about reasons to try, since such reasons are plausible and do not violate OIC. However, his defence of reasons to succeed poses a threat via the Argument from Strict Liability: (1) There are reasons to x, regardless of one’s ability to x. (2) The reasons specified in (1) sometimes create duties to x, regardless of one’s ability to x. (3) The duties specified in (2) sometimes violate OIC. (4) Strict liability imposes the duties specified in (2). (5) Strict liability sometimes justifiably violates OIC. As we noted before, there are other ways of understanding strict liability, on which (4) is false. If strict liability imposes conditional obligations, these are not duties to succeed in not injuring others, but rather to pay compensation if one causes injury. However, since it is not my aim to dispute the interpretation of strict liability, let us grant (4) for the sake of argument. Equally, it should be clear that duties to succeed violate OIC. Even on the most permissive interpretation of ‘can’, it will sometimes be impossible to secure an outcome, regardless of the care one exercises. Since Gardner does not argue explicitly from (1) to (2), let us also grant arguendo that, if there are reasons to succeed, some of these create duties to succeed. This leaves (1). The structure of Gardner’s argument for (1) is an inference to the best explanation. He claims that recognising reasons to succeed provides the best explanation for important moral phenomena. He offers two types of examples: moral tragedies and cases involving derivative reasons. An example of the first type was already introduced: in Drowning Person A looks helplessly as B drowns below her, knowing that she cannot hope to save him. A’s experience is horrifying, tragic and leaves a ‘trace’ on her as an agent. These claims capture something intuitive about the phenomenology of the case, but it is not clear what a ‘trace’ really is. Presumably it resembles the concept of ‘agent regret’, which refers to the special type of regret experienced by an agent who has harmed someone, even if she could not have avoided it. The two are not identical, since agent regret typically involves causing harm and A fails to prevent harm in Drowning Person. Nevertheless, we can recognise A’s experience as tragic and acknowledge the pain and frustration that attends the realisation that she is unable to save B. The second example is that of derivative reasons. Consider: Accessibility: A is wheelchair user. She is unable to ride the buses under current circumstances as they are not wheelchair accessible. Gardner claims that A has a reason to ride the bus, even though it is not wheelchair accessible. Of course, A does not have a reason fruitlessly to attempt to ride the bus (this would be a reason to try): she just has a reason to ride the bus. This reason explains a host of other reasons derived from it, such as the reason to lobby for improved accessibility on public transport. Ulrike Heuer offers a similar argument in defence of reasons to do the impossible. See Heuer 2010, 235-246. Is Gardner’s argument correct? Only if positing reasons to succeed best explains the relevant moral phenomena. Before we reach this conclusion, we should consider competing explanations. In particular, we can revisit the distinction between prescriptive and evaluative normative claims and consider the potential of purely evaluative facts to explain the relevant phenomena. In Drowning Person, an alternative explanation of the tragedy faced by A is the evaluative fact that something of great value will be lost if B dies. John Oberdiek appeals to value pluralism to explain the phenomenon of ‘moral residue’ in moral dilemmas. According to value pluralism, there are irreducibly distinct values that are sometimes incompatible. Oberdiek 2004, 332-333. When one value takes priority over another, something valuable is inevitably lost, and this explains why choices between incompatible values have a ‘tragic’ quality. There are two important components of Oberdiek’s explanation: choice and loss of value. In Drowning Person, the effect of the situation on A cannot be attributed to her choice, since she has none, but perhaps the great loss of value is enough to explain her response. Unfortunately, there is a rejoinder to this suggestion. Similar losses of life and value are a ubiquitous feature of human life. Although we are aware of this fact, our response to these losses rarely mirrors A’s response to B’s death in Drowning Person. The fact that a valuable life is lost seems insufficient, by itself, to explain A’s reaction. We may rely on the purely psychological fact that people are more affected by suffering when it is visible or close. For a well-known version of this argument, see Singer 1972. This possibility cannot be ruled out, but we should not be too quick to give up the idea that there is a normative as well as psychological explanation for the phenomenology of Drowning Person. At this point, Gardner may reassert that A has this reaction because she has a reason to succeed in saving B, which gives her a special relation to B’s plight that she does not have with other people. However, there are other explanations of the phenomenology that do not rely on loss of value or reasons to succeed. A’s situation is tragic not just because she recognises a loss of value, but because she experiences a moral emotion that I will call agent anguish. Agent anguish is the affective response to the recognition that, in close possible worlds, one could have prevented a great loss (and therefore would have a reason or duty to do so). It is not an emotional response to the loss of life tout court. Rather, it represents the particularly intense or desperate reaction to being close to averting tragedy. If only the world had been slightly different, A might lament, she could have prevented disaster. Agent anguish is not only a familiar emotion that explains A’s reaction in Drowning Person. It also explains how the moral phenomenology of A’s experience progresses. A’s initial reaction is to cast around desperately for ways to save B and only when she realises that this is impossible does she flail and wave and sink to her knees. There is a recognisable change in her behaviour: she goes from frantic deliberation to tragic despair. This marks the point of realisation that she cannot save B: the point when agent anguish takes hold. Agent anguish explains the special proximity that A has to B’s death, but it does not rely on the postulation of any reason to save B. This discussion does not settle the phenomenology of Drowning Person, but it shows that the first plank of Gardner’s argument is not dispositive. Next consider Gardner’s second argument based on derivative reasons. In Accessibility, is A’s reason to lobby for accessible public transport best explained by her reason to ride the bus? Talk of a reason ‘to ride the bus’ is confusing since the action favoured by the reason is not properly specified. A reason to ride the bus here and now, even though such a thing is impossible, clearly violates OIC, but a reason to ride the bus at some point in the future when circumstances may have changed need not. So, let’s assume that the reason in question is a reason to ride the bus in an instance in which doing so is impossible. The ‘derivative’ reasons in this case can be explained with reference to purely evaluative claims. The fact that it is a violation of A’s fundamental interest in accessing public services generates reasons both for her and others to take steps towards implementing wheelchair-accessible transport. In fact, far from demonstrating the existence of reasons to do the impossible, the best explanation for this case relies on OIC. Though Gardner holds that A has a reason to ride the bus, he presumably accepts that A has conclusive reason to do something else, such as lobby for improved accessibility, which is a next-best option from her perspective. Why would she have conclusive reason to take the next-best option? A natural explanation is that A has no reason to take the best option since this is impossible, For a defence of OIC along these lines, see Howard-Snyder 2006. but how does Gardner explain this? One could argue that A has conclusive reason to take the next-best-but-possible option whilst also having a reason to take the impossible-but-best option, but then if OIC does not preclude a reason in favour of the best option, why not a conclusive reason? This alternative explanation is given support by a more offensive argument against reasons to succeed. Gardner faces the problem of explaining how reasons are appropriately limited. If OIC is false, they cannot be limited by the fact that the agent is unable to achieve the relevant result. Do we have reasons to save members of past generations? Do we have reasons to rescue people who are drowning on the other side of the world? Do we have reasons to rescue people through impossible means, such as wishing them saved? Most of those who deny OIC do not think that we have such ‘crazy reasons’. The term is Bart Streumer’s, who develops this objection in response to Heuer. See Streumer 2010. But this creates a challenge. Those who accept OIC have a ready explanation why such reasons do not exist. Those who reject it are left with the task of explaining why we have reasons to perform some impossible acts but not others. To meet the problem of ‘crazy reasons’, Gardner proposes a baseline to determine which impossible actions we can have reason to perform. He argues that, if at least one conceivable human being has the capacity to x and A is a human being, then the impossibility of A x-ing does not preclude the possibility that A has a reason to x. In other words, ‘ought implies that a conceivable human can’. See Gardner 2013. What should we make of this fix? Even if this solution rules out ‘crazy reasons’, it does not prevent an implausible proliferation of reasons. On this view, it seems that one has a reason to run a hundred metres in 9.5 seconds, since such a feat would make one famous and can be achieved by a conceivable human. Similarly, one has reasons to jump 2.5 metres high and 9 metres long; to reach something on a shelf that only the tallest person in the world could reach; or escape an attacker through a gap through which only a toddler could crawl. The force of the ‘crazy reasons’ problem is not just that we don’t have bizarre reasons, but also that our practical reason is not swamped with limitless reasons to do things that we, as individuals, lack the ability to do. Leaving aside the difficulty of defining the terms ‘human’ and ‘conceivable’, the proposal is also vulnerable to an arbitrariness objection. Why should conceivable human achievement be the appropriate baseline? To illustrate the objection, consider the following cases: Truck 1: A is trapped under a truck. B cannot lift the truck but C, the strongest conceivable human, could lift it. Truck 2: The same as 1 except the truck under which A is trapped is marginally too heavy for even the strongest conceivable human. Gardner’s version of OIC implies that B has a reason and perhaps a duty to lift the truck and in Truck 1 but not in Truck 2. This is hard to believe. Lifting the truck is equally impossible for B in 1 and 2, so why should the conceivability of a human who has this ability make any difference to what B has reason to do? Of course, one could adopt a wider definition of conceivability and argue that a human with superhuman strength is conceivable, but to stretch the definition of conceivability in this way would be to re-introduce the ‘crazy reasons’ problem that this move was meant to prevent. VI. The Functional Convergence Thesis So far, we have seen that the Argument From the Objective Standard of Care and the Argument From Strict Liability do not succeed, primarily on normative grounds. A final attempt to resurrect these arguments might concede the justification of OIC that we offered at the outset but deny that the law should be constrained by it. It would be simplistic to think of the law’s sole function as implementing morality for many reasons: some moral duties ought not to be enforced; many elements of the law are not justified by reference to underlying moral principles; pragmatics has a part to play in doctrine and practice; and so on. If morality and the law need not be completely intertwined, why should we worry that duties in tort law violate OIC? There is, of course, much to say on the matter. Here I will defend only one claim, but one that explains why OIC should constrain the law, even if morality and law can diverge in other areas. The claim appeals to the justification of OIC that we defended earlier. According to what I will call the Functional Convergence Thesis, morality and the law both embody functional roles in our lives as agents engaged in practical reasoning. To perform these roles, it is not necessary that the law enforce all our moral duties or treat morality as an absolute constraint on legal duties, but it is necessary that legal duties, like moral duties, can fulfil their function in our practical reasoning. Duties that violate OIC cannot do this. If OIC was based on fairness, any objection to legal duties that violated it would be much weaker, since fairness is only one consideration that the law can legitimately consider. On the functional view, OIC ensures that reasons and duties have the potential to guide action prospectively, allowing them to fulfil their deliberative function fully, and so the objection to legal duties that violate OIC is much stronger. Without adhering to OIC, legal duties are robbed of their potential to feature in practical deliberation the way moral duties do. This divergence would be concerning because it threatens the idea that both morality and law address us deliberating agents. To see how the functional convergence thesis distinguishes between acceptable and unacceptable divergences, consider the following example. Suppose that those who could not avoid negligently harming others were subject to conditional duties. On this view, negligence is divided in two distinct types. Some negligent actors, who could have avoided their conduct, harm others as a result of their wrongdoing and are liable on this basis. Others, who could not have avoided their conduct, are liable despite having acted permissibly. The basis of this liability obviously cannot be wrongful conduct, but may be based on other factors, both principled and pragmatic. This preserves the current distribution of liability and satisfies the pragmatic considerations that are usually thought to favour the objective standard of care. The difference between this sketch and current regimes may seem trivial, since the same people pay the same costs. However, it is not trivial because only this hypothetical division of negligence liability preserves the potential for legal duties to play a prospectively action-guiding role by adhering to OIC. Conversely, a person who is unable to avoid negligence but is subject to a legal duty to do so cannot appeal to this duty in her deliberation in the normal way. It is not just that the duty does not guide action, but rather that it is ruled out at the start because the agent cannot fulfil it. This is how the functional convergence thesis distinguishes between impossibility and weakness of will. The duty imposed on the weak-willed agent may not be acted upon, but it can still fulfil its function at the deliberative stage. The agent can reason that, considering all relevant factors, this is the action she ought to perform. She can complete this process and decide to act, even if she does not follow this through because she finds herself lacking motivation. The duty that demands the impossible, on the other hand, cannot feature in the agent’s deliberation in this way. If it is not constrained by OIC, agents cannot always consult their legal duties in deciding what they ought to do. Perhaps these instances are sufficiently rare to pose no practical threat to the law’s functioning. But this should not encourage us to lose sight of how radically a legal duty that violates OIC alters the law’s understanding of us as deliberating agents. Bibliography Anderson, E (1999) ‘What Is the Point of Equality?’ Ethics 109:2 (1999) 287-337 Barcan Marcus, R (1966) ‘Iterated Deontic Modalities’ Mind 75 (1966), 580-2 Beever, A (2009) Rediscovering the Law of Negligence (Oxford, Hart Publishing, 2009) Carter, I (2001) ‘“Ought” Implies “Practical Possibility”’ in Carter, I and Ricciardi, M (eds) Freedom, Power and Political Morality (London: Palgrave Macmillan, 2001) Dorfman, A (2012) ‘Reasonable Care: Equality as Objectivity’ Law and Philosophy 31: 369 – 407 (2012) Gardner, J (2018) From Personal Life to Private Law (Oxford: OUP, 2018) Gardner, J (2013) ‘Reasons and Abilities: Some Preliminaries’ The American Journal of Jurisprudence 58:1 (2013), 63 – 74 Gardner, J (2001) ‘Obligations and Outcomes in the Law of Torts’ in Gardner, J and Cane, P (eds) Relating to Responsibility: Essays in Honour of Tony Honoré on his 80th Birthday (Oxford: Hart, 2001) Harman, G (1977) The Nature of Morality (Princeton: PUP, 1977) Heuer, U (2010) ‘Reasons and Impossibility’ in Philosophical Studies, 147:2 (2010) 235-246 Honoré, T (1988) ‘Responsibility and Luck: The Moral Basis of Strict Liability’ Law Quarterly Review 104 (1988), 530 Howard-Snyder, F (2006) ""Cannot" Implies "Not Ought"." Philosophical Studies 130:2 (2006), 233-46. Humberstone, L (1971) ‘Two Sorts of “Oughts”’ Analysis 32 (1971), 8-11 Keating, G (2004) ‘Strict Liability Wrongs’ in Philosophical Foundations of the Law of Torts Oberdiek, J (ed) (Oxford: OUP, 2014) Nagel, T (1979) Mortal Questions (New York: CUP, 1979) Oberdiek, J (2004) ‘Lost in Moral Space: On the Infringing/Violating Distinction and its Place in the Theory of Rights’ Law and Philosophy 32 (2004) 325-346 Ripstein, A (1999) Equality, Responsibility, and the Law (Cambridge: CUP, 1999) Singer, P (1972) "Famine, Affluence, and Morality" Philosophy and Public Affairs 1:3 (1972) 229–243 Sinnot-Armstrong, W (1984) ‘“Ought” Conversationally Implies “Can”’ Philosophical Review 93:2 (1984) 249-261 Stevens, R (2007) Tort and Rights (Oxford: OUP, 2007) Streumer, B (2010) ‘Reasons, Impossibility and Efficient Steps: Reply to Heuer’ Philosophical Studies 151 (2010), 79-86 Streumer, B (2007) ‘Reasons and Impossibility’ Philosophical Studies 136 (2007), 351 – 384 Tadros, V (2016) Wrongs and Crimes (Oxford: OUP, 2016) Talbot, B (2016) ‘The Best Argument for “Ought Implies Can” is a Better Argument Against “Ought Implies Can”’ Ergo 3:14 (2016) Vranas, P (2007) ‘I Ought, Therefore I Can’ Philosophical Studies 136: 167-216 (2007) Wolff, J (1998) ‘Fairness, Respect, and the Egalitarian Ethos’ Philosophy & Public Affairs 27:2 (1998), 97-122 Wright, R (1995) ‘The Standards of Care in Negligence Law’ in Philosophical Foundations of Tort Law Owen, D (ed) (Oxford: OUP, 1995) Cases Mansfield v Wheetabix [1998] 1 WLR 1263, 1268 (CA)