Abstract and Keywords
This article discusses different uses of ‘responsibility’ in our everyday social, moral, and legal discourse. Moral responsibility serves as a template for more institutionalized forms of responsibility, thus the capacities it presupposes and its criteria of liability can illuminate other forms. The article further discusses criminal responsibility, notably the problems of finding an adequate theory of criminal legislation, appropriate response, and criteria of responsibility. It shows that the periodic oscillations between managerial and retributive approaches to punishment reflect a deeper debate about the nature of state–social relations. The article focuses on the debate in tort theory between instrumentalist and corrective justice views. The aim of the article is to show how a relational understanding of responsibility clarifies the debates within tort theory.
Claims of responsibility are notoriously multifarious. H. L. A. Hart's tale of the drunken captain, here adapted slightly, still shows this best:
(1) As captain of the ship, Smith was responsible for the safety of his passengers and crew. (2) But he drank himself into a stupor on his last voyage and was responsible for the loss of the ship and many of its passengers. (3) The doctors initially thought his drinking might have been the product of a paralytic depression, but later concluded that he had, in fact, been fully responsible at the time he became drunk. Smith initially maintained that the exceptional winter storms were responsible for the loss of the ship, but at trial, (4) after he was found criminally responsible for his negligent conduct and sentenced to ten years imprisonment, (5) he declared that no legal penalty could alleviate his guilt, for which he sought to atone. (6) Some of the survivors of the wreck, however, declared that they wished to put their nightmare behind them, and forgave Smith. (7, 8) Meanwhile, the president of the cruise line issued the following statement: ‘Although the company must accept its legal responsibility for the loss of life and property, we bear no culpability for the disaster, since Smith fraudulently concealed from us his earlier employment problems, and our alcohol screens turned up no evidence of his drinking.’1
(p. 549) The story rehearses different uses of ‘responsibility’ in our everyday social, moral, and legal discourse; the numbers distinguish either different senses of responsibility, or different exemplary contexts in which someone takes responsibility or is held responsible. Here is Hart's catalogue: First is a claim of role responsibility: Smith, in virtue of his position as captain, had specific obligations to safeguard his ship and his passengers. A claim of role responsibility states the expectations of an agent's conduct towards some charge. Second is a claim of causal responsibility: the captain's insobriety is cited as the cause of the vessel's loss. Causal responsibility might be better thought of as a species of explanatory responsibility, causation being typically the best explanation of an event.2 Third is a claim of capacity responsibility: the captain's decision to drink was not the product of a pathology, or some other non-deliberative causal process, but rather reflected his exercise of a power of rational self-determination. Being responsible, in this sense, simply is a matter of having the competency of self-government. Four, five, six, and seven relate to claims of different kinds of individual liability responsibility, respectively accountability to the demands of the criminal law, tort law, and morality. Finally, eight involves a claim of collective responsibility, a claim whose distinguishing feature is that the responsible subject involves a plurality of individuals.
Much of the modern literature involves attempts to refine, reduce, and compare elements of Hart's taxonomy—to show, for example, why moral and criminal liability share a common foundation, why role responsibility is the foundation for liability responsibility, or why collective responsibility cannot be reconciled with individual responsibility. R. A. Duff, for example, distinguishes causal, prospective, and retrospective responsibility, making claims of causal responsibility factual and the other two normative. Prospective responsibility (what I have called role responsibility) is defined by norms governing conduct, and retrospective responsibility is accountability for failure to meet those norms. Capacity responsibility is then defined derivatively, in terms of whether an individual is an appropriate candidate for prospective or retrospective responsibility; only responsible agents can be held responsible.3 T. M. Scanlon similarly distinguishes between judgments of substantive and attributive responsibility. Judgments of substantive responsibility involve claims about what people are required to do for one another, and judgments of attributive responsibility are judgments that some act or event is a proper basis of moral appraisal.4 Finally, Stephen Perry, following Tony Honoré, makes a distinction within the field of attributive responsibility, between act and outcome responsibility.5
However we slice the idea of responsibility, it is apparent that we need considerable information to deploy the term. First, we need to know the object of the agent's (p. 550) putative responsibility. Is it a task, a status, someone's well-being, conduct, or an event? Secondly, we need the ground for demanding this responsibility—that he performed some act, caused the event, was invested with duties of a specific sort. His having or lacking a capacity for responsibility enters as a precondition of his responsibility, but being responsible includes the nature of his relation to the act, state, or outcome, whether he did it, caused it, manifested it, and so on.
In other words, we need to know whether the agent accepted the role, performed the act, caused the harm; and whether the agent did so consensually or involuntarily, intentionally or accidentally, sanely or madly. The idea that attributions of responsibility rest solely on facts about agents and their relations to certain harmful (or favourable) events or states is familiar and attractive. To give this idea a name, let us call it retributivism. The fundamental idea of retributivism is that responsibility is a moral property of agents that consists in or supervenes upon underlying facts of agency and upon agents' connections to the world. Such facts uniquely determine the moral desert of the agent; it is then a primary job of our moral and legal institutions to mete out to agents the response they deserve. On the simple retributivist picture, responsibility is a moral fact, pertaining to a relation between an agent and an object of assessment.
The retributive conception of responsibility is not wrong. It is radically incomplete. For claims of responsibility are more elliptical than I have so far indicated, in two ways. First, beyond the facts of agency, capacity, and causation, we need to know the response demanded of and to the agent, and conditions of warrant for that response: is it (among the range of possibilities) contrition, or civil liability, or criminal punishment, and what are the criteria for appropriate application of each? The truth of a claim of responsibility depends on the mode of demanded response. Smith, for example, may be justly liable in tort but not in criminal law. If his fault is minimal or non-existent, then resentment by his victims may be unwarranted, even though he himself must properly regard the accident with great regret.
Attributions of responsibility occur not in a juridical vacuum, but in specific interpersonal and circumstantial contexts. Such attributions are fundamentally relational: they depend upon the character of moral, legal, and social relations among the actor, the victim, and the evaluator. Consequently, we need to know what I will call the position, or identity, of the respondent to the agent, as well as the relation between them. The justification for demanding a given response depends on the position and the relation of the respondent to the agent. Is the respondent a victim, a court, the agent himself or herself, a bystander? Smith, arguably, owes his victims but not the state an apology; and the state, but not his victims, has a right to punishment after a fair trial. At that, Smith may only be justly punished by the state with the appropriate jurisdictional relationship to him—even a scrupulous adjudication of the merits of his conduct by an alien court would be irrelevant to the justice of his punishment.
Relational and positional dependence reflect a number of deep facts about responsibility claims. First, the complex set of practices involved in taking responsibility, (p. 551) projecting responsibilities, and finding and holding persons and collectives responsible can only be made sense of against the background set of social, political, and legal relationships and their constitutive norms. Secondly, the contextual and relation-dependent nature of responsibility claims means that, fundamentally, responsibility is a social practice and not the neutral registration of independent moral facts. Claims of responsibility are things we do, revelations of our agency. Thus, thirdly, making responsibility claims, of ourselves or others both constitutes and transforms our agency and our relations to one another. Consequently, judgments of responsibility must be understood in terms of the ideals of agency and community that they reflect and effect.
The fundamentally relational character of responsibility is reflected in a recent efflorescence of philosophical work on moral and legal responsibility, including the works mentioned above. This literature has demonstrated, sometimes merely implicitly and sometimes despite itself, that the traditional chestnuts of the topic—such as the problem of psychological determinism, the legitimacy of strict liability, or the distinction between tort and criminal responsibility—will be cracked not with a priori arguments but instead with examinations of the relationships and expectations that give point and structure to our responsibility practices. Indeed, shifts and divisions within recent philosophical literature make sense only with the realization that different theories reflect different conceptions of the background relations. The shift in contemporary moral and legal thinking, away from the systemic view of consequentialism and towards a deontology focusing on individual responsibility, manifests the central social and political dilemma of late modernity: reconciling individual meaning and autonomy within the increasingly consolidated social world.
The subject of responsibility could clearly consume much of the subject-matter of law, including many matters treated elsewhere in this volume. This chapter will not pretend to be a complete treatment of the idea, but rather is a sketch of some important sub-themes within the topic. Section 2 discusses moral responsibility. Moral responsibility serves as a template for more institutionalized forms of responsibility, thus the capacities it presupposes and its criteria of liability can illuminate other forms. Section 3 takes up criminal responsibility, notably the problems of finding an adequate theory of criminal legislation, appropriate response, and criteria of responsibility. I will try to show that the periodic oscillations between managerial and retributive approaches to punishment reflect a deeper debate about the nature of state-social relations. In Section 4 I treat the exemplary case of tort liability, focusing on the debate in tort theory between instrumentalist and corrective justice views. Again, my aim is to show how a relational understanding of responsibility clarifies the debates within tort theory.
Thus, I leave several topics for further exploration. In particular, I do not discuss the question of what Ronald Dworkin has called ‘political responsibility’, that is, the responsibility on the part of the state generally or state officials particularly to justify their conduct—a responsibility particularly at issue in jurisprudence and (p. 552) administrative law.6 Although political responsibility might be thought of as a version of role responsibility (at least in many instances), I have omitted its discussion because it belongs properly to jurisprudence, discussed extensively in this volume, and to political theory more broadly. Engaging those questions would force too great a digression from the central focus of this chapter on individual responsibility. For similar reasons I will not treat, except in passing, the subject of ‘social responsibility’, as it relates to the social welfare obligations of individuals or other entities. Finally, my discussion of legal responsibility in private law is limited to tort law, although the subject of responsibility in contract law also raises interesting philosophical questions, for example, about the relation between promise and contract, and the justification of promissory estoppel and unconscionability principles.7 For the most part, however, the relevant issues of agency and repair, and the social ideals they presuppose, are aired in the tort law discussion.
2 Moral Responsibility
Moral responsibility names a set of practices, and our conduct, consequences, and character are the objects of those practices. We hold ourselves and each other morally responsible for how we act, what we bring about, and who we are. This is an entirely unremarkable claim, and would seem a natural starting-point for discussions of moral responsibility. But philosophical discussions of moral responsibility have instead often been waylaid by the challenge of reconciling a conception of responsibility with a naturalistic understanding of human deliberation and action. The philosophical problem of responsibility arises from two powerful ideas. On the one hand, it seems that we do not morally praise or blame others for acts not somehow the products of their choices; and even when the acts are the products of their choices, we will withdraw blame if we discover that, for some reason, the person could not have chosen other than as he or she did. Had the ship been sunk by an unnoticeable iceberg, Captain Smith would have been off the hook, as he would also have been had his drinking been the product of a disease (though he might have been responded to in some other way, e.g. therapeutically). This observation is then transformed into the metaphysically more ambitious claim that moral responsibility for an act (or the consequences of an act) requires both that the agent could have done otherwise, and (p. 553) that the agent is responsible because his or her choice was the cause of the act. On the other hand, any plausible conception of humanity's place in nature must make room for the idea that our choices and actions are as subsumed under natural laws as all other phenomena.
The conjunction of a naturalistic understanding of human action, thus subsumed, and the conception of responsibility in terms of a capacity to choose freely among a range of options generates the metaphysical free will problem. For a naturalistic understanding of human action suggests either that choices are the determinate products of antecedent events, or that they are the products of pure indeterminacy. Either way, the conception of responsibility as capacity to do otherwise is undermined.8 Moral responsibility, extrapolated from this argument, is a fundamental aspect of our social lives, and yet it seems to require a kind of freedom unavailable in the world we inhabit.
There are two things to notice about the genesis of this problem, both related to the concept of responsibility from which it arises. First, the problem arises from an underspecified understanding of responsibility: the putative requirement of free choice is taken roughly as an intuitive axiom. The result has been to interpret the notion of free choice in terms of counterfactual possibilities, and then to compare that interpretation against claims about physical necessity. Secondly, the understanding of responsibility is basically solipsistic, in that only facts about the agent, his choice, and his acts, are relevant to the ascription of responsibility; relations to other agents are irrelevant. The result is that the metaphysical notion of being responsible is taken as primary, and the notions of holding or taking responsibility are derivative. One is morally responsible in general if one possesses the relevant capacity for free choice, and morally responsible for a particular act or event if that act or event resulted from such a choice.
The metaphysical problem of free will has inspired much difficult and interesting work.9 But the mere capacity conception of responsibility, coupled with the disregard for the social relations in which ascriptions of responsibility are embedded, has meant that the metaphysical debates tended to reveal little about the underlying notion. The social and psychological meaning of responsibility was neglected. This all changed with Peter Strawson's seminal ‘Freedom and Resentment’, an article that (p. 554) aimed to reverse the traditional direction of explanation. Rather than explain the notion of holding someone responsible in terms of a capacity for responsibility, Strawson suggested taking the idea of holding responsible as primary, and then understanding the capacity sense of responsibility in terms of the liability sense of responsibility. The result is to ground the abstract notion of moral responsibility in a set of social practices of holding ourselves and one another responsible, not in a metaphysical conception of free choice. Obversely, Strawson extracts the incapacity for responsibility from our social practices of excuse. The hope, then, is that the metaphysical free will problem can be disarmed by showing that the social practices of excuse do not generalize under the threat of causal determinism but are, rather, context-specific.10
What Strawson noticed was that ascriptions of responsibility have a crucial affective dimension. Our practices of accountability are made up of natural patterns of emotional reaction, or ‘reactive attitudes’, to the welcome and unwelcome attitudes of others manifested in their conduct towards us.11 When I blame you for slapping me on the back of the neck, I am venting my resentment at the hostility implicit in your act; and when I am grateful to you for courteously holding the door for me, I am expressing my delight at the goodwill you demonstrate. My responses to your actions flow principally from my assumptions about the sentiments expressed by your conduct, not the consequences produced by it. Thus, when I discover that the attitude to which I am reacting is absent or different than I had supposed, my reaction naturally transforms. If I discover that you slapped my neck in order to swat away a bee, then I will no longer resent the action as an attack upon me. Or, if I discover that you have been merely careless in swinging your hand around, I may revise my resentment to focus upon your disregard rather than your hostility. My reactions similarly shift when the attitude is present, but has a suspect aetiology—perhaps an effect of your paranoid delusions. Now I do not resent your hostility, but try to understand it, because it no longer expresses your considered sentiments, but only the state of your mental health.
There are two points to notice here. First is that the capacities and incapacities presupposed by our reactive attitudes are straightforwardly psychological, not metaphysical. ‘Will’ thus names an item accessible to naturalistic investigation. Since there is no evident reason to think that the truth or falsehood of determinism bears on the nature or exercise of these capacities, the psychological concept of responsibility can be unyoked from the metaphysical concept. Once the two are unyoked, it is difficult to see the motivation for the metaphysical problem our ordinary responses, as distinct from the intrinsic philosophical interest in whether our behaviour has ultimate external causes. The test for this claim is whether, if we really believed determinism (p. 555) was true, our ordinary responses would erode. But that test is pragmatic, not logical, and cannot be resolved by theoretical discussion. Strawson's view thus opens up conceptual space for an independent investigation of the norms internal to the practices of responsibility, norms whose content can be divorced—in great part even if not entirely—from metaphysics.
Secondly, and relatedly, our disinclination to express reactive attitudes to partly or wholly non-responsible agents is explained not merely by the quality of their wills, but by the nature of our relations with them. Thus, the norms governing our practices of responsibility are in part social norms, deriving from and governing our relations with others. Though children and the insane do indeed manifest attitudes of hostility and goodwill, we tend to take what Strawson calls an ‘objective’ rather than a ‘participant's’ view of their attitudes. Instead of attempting to define the quality that responsible agents' wills have and non-responsible agents' wills lack, Strawson emphasizes the way that our awareness of cognitive and affective limitations in non-responsible agents naturally precludes them from participating in the relationships characteristic of adult society.12 We see them not as accountable subjects but as the objects of understanding, treatment, or education—that is, as quasi-participants in therapeutic relationships.
It should be clear from this brief description that a Strawsonian view does not insist that we must have these reactive attitudes in every case in which they might be warranted. To borrow an example from Jay Wallace, a rogue might act in a way that would warrant recrimination, but be so charming that we cannot work up the indignation.13 Moral responsibility is, in any event, a normative rather than descriptive concept: someone's being responsible is a matter of being warranted by the relevant social norms in having certain attitudes towards them. Just as I may form unwarranted attitudes towards someone whom I mistakenly take to be responsible, so I may fail to form warranted attitudes. Studies of the moral emotions by such writers as Patricia Greenspan, Jean Hampton, Michael Moore, Herbert Morris, Jeffrie Murphy, Samuel Scheffler, Gabriele Taylor, Bernard Williams, and Richard Wollheim, have also contributed to a normative understanding of our emotional responses.14 We may also extend the notion of a warranted response from affect alone to acts of contrition, punishment, gratitude, and reward. In its most general sense, to be responsible is for certain responses to be warranted, in virtue of what one has done and why one has done it.
(p. 556) The claim that responses are warranted by governing social norms necessarily implies some social relativism. Relativity to social norms can arise in at least two innocuous ways: social norms define the nature of the act in question and they regulate the appropriate response. A remark that is a mild tease in one society (or social subgroup) can be a grave insult in another; and an insult that demands redress in one place may permit a cheek-turning in another. But anchoring responsibility in local norms may seem to imply as well a less palatable, more thoroughgoing relativism, leaving no room to criticize quaint local traditions such as scapegoating or ritual sacrifice. Moreover, social norms conflict and are frequently indeterminate in their demands even when they do not conflict. Thus, the Strawsonian approach may well lead to questions of responsibility that can receive only partial and limited answers.
There are, however, a couple of responses to these worries. First, social norms are rooted in a collection of human needs, wants, and dispositions that are only semi-plastic—influenced but not fully determined by physical and social environments. The Strawsonian account recognizes local variation, to its considerable advantage; but the degree of that variation should not be overestimated. In every culture where accidents and injuries happen—which is to say in every culture—the responsibility practices that arise will persist only if they cohere with other normative and explanatory concepts. Practices bearing too little relation to such basic considerations as causality and proportionality, for example, are unlikely to flourish over time, for they will fail to cohere with other basic cultural and scientific institutions. While pockets of magical thinking will surely persist in any culture, they are unlikely to remain the bedrock of responsibility practices. There thus will be room in any culture with a notion of causality (which is to say every practically feasible culture) for criticism of pure scapegoating. Moreover, nothing in the Strawsonian account precludes grounding (or criticizing) some set of responsibility practices in terms of some non-relative ethical standards (for example, standards of fairness or equal treatment). Warrant can emerge from local context, or from absolute ethical standards (if they exist), or both; no deep relativism is implied.
The second point concerns indeterminacy. It is a consequence of the Strawsonian view that when responsibility norms conflict (either local or absolute norms), there will be no clear answer what response is warranted, despite disagreement among the participants in the debate. But ‘under-determinacy’ is the better term for this state of affairs, not ‘indeterminacy’, for it is not the case that no response is warranted, but rather that the set of applicable norms is insufficient to warrant any unique response. And it seems a virtue of the Strawsonian account to imply such under-determinacy, for under-determinacy is surely also a feature of the moral (and legal) lives the account aims to reflect. Moreover, under-determinacy, unlike indeterminacy, makes room for argument, as participants contest the relative weight or priority of different potential norms, for example when the spirit of a rule is best honoured by an exception, or when mercy's place must be subordinated to collective security.
So Strawson's suggestion is helpful, not just for the way in which it allows us to avoid metaphysical thickets, but by making room for the relational and positional (p. 557) character of responsibility. It should now be clear that the attitudes and expressions of agents only warrant response given a certain understanding of the nature of the relationship between agent and respondent. In Strawson's very rough division, the relationship must be either participatory or potentially participatory: the agent to whom we respond must be someone with whom we will or could co-operate in social life. Our attitudes and expressions both indicate and constitute the nature of a participatory relationship. In general, we care about our relationships with others in virtue of the ways they can make our lives good (or bad), both in themselves, and as vehicles for promoting our interests. So the responses characteristic of accountability are warranted by the point and demands of the relationship. What we take responsibility and hold each other responsible for are deviations between our actual conduct and the norms constitutive of the relevant relationship.
Acknowledging relationality entails, as Strawson acknowledges, a necessary variability in warranted responses depending upon the nature of the relationship in question: what might constitute callous indifference between friends or lovers is simply good manners between commercial transactors.15 For example, if I carelessly break a neighbour's vase at a party while dancing on his grand piano, my neighbour is warranted in resenting my carelessness and asking for an apology, though not in, say, smashing my glasses. Reciprocally, an apology or restitution is warranted on my part (and perhaps even obligatory). But my responsibility does not end with a simple interaction between my neighbour and myself. There are countless other positions from which other agents may respond to my act. For example, other guests at the party may also feel indignant at having their pleasant evening disrupted by my loutish behaviour, and they may expect a public display of contrition for their sake, though they could not appropriately feel personally aggrieved in the same way as my neighbour. Perhaps some of the guests are relatives of my neighbour, however, and they may take the event more personally than friends and acquaintances present. I may also be accountable to my own family for the harm, since they will now be embarrassed before the neighbour, and I may owe them a promise to take more care in the future. Finally, to and from the public at large only very constrained responses are warranted. While anyone who heard about my accident could consider me a fool, and say so, a more direct response to me personally would be thought self-righteous and nosy; and it would be self-abasing of me to confess my shame to a random person met in the street.
This essential and obvious fact of responsibility, its relational and positional dependence, is unexplained on the retributive, desert-based model. The retributivists' exclusive focus upon an agent's intentional state and actions dictates that all warranted responses flow from a single constant value: what the agent deserves. The response warranted by desert is thus univocal, dependent upon facts about the agent rather than the agent's relations to others. One could object that the variability of warranted responses can be made consistent with the retributive model: an agent (p. 558) ‘deserves’ multiple and varied responses from different people. On this interpretation, ‘desert’ just means that some response (or set of responses) is warranted on some ground. While there is nothing objectionable about this use, it falls well short of the traditional ambitions of the desert model, namely itself to provide a justification for hard treatment and prescribe the upper and lower limits of that treatment.
Strawson's own account works best where the form of background participatory relationship that grounds and warrants response is most conspicuous, that is, in the domain that I have called social accountability. His account is less helpful in explaining the special character of our moral responses to agents with whom we share no particular set of relationships—for example, my reaction upon reading in the paper that an employer has exploited its workers. Here I am outraged, towards the employer and on behalf of the workers, though I cannot in any deep sense identify myself with either of their positions. Strawson says that the relationship among moral respondents in such cases is simply a ‘generalized’ form of the claim to goodwill made by members of participatory social relationships. He does so in order to explain what he calls the ‘vicarious’ nature of moral reactions: responses like moral indignation are ‘essentially capable’ of being directed at others' attitudes towards others as well as at attitudes directed towards ourselves.16 Strawson says these vicarious reactions are ‘humanly connected’ with participant reactions, though he does not explain the nature of this connection.17 Strawson is surely right to suggest that it is a deeply rooted fact that humans—or at least members of minimally cohesive societies—have a propensity to pass judgment generally on others' compliance with social norms.18 Indeed, it is hard to imagine how a society could maintain its normative structure if its members were not disposed to monitor and censure each other for non-compliance. It is in this propensity, layered and modified through cultural forms, that the institutions of judgment, punishment, and repair find their ground.
Sometimes responses to agents are not motivated by the attitudes those agents manifest, nor by their failure to conform their conduct to appropriate norms. Sometimes an agent's mere causal linkage with a harm may warrant a response from others. The responses characteristic of accountability for consequences can also only be understood in terms of the moral and social relationships among the parties, and their different positions with respect to the harm. The striking asymmetry in accountability for consequences between the responses of agents themselves, on the one hand, and victims and onlookers, on the other, has not been fully appreciated. In particular, agents can reproach themselves for faultless conduct that causes a harm, even when their victims, and onlookers, do not reproach them. This asymmetry of responses to (p. 559) consequences reflects the deep role that causal relations have for agents in structuring their understanding of (or relationship to) themselves. Those affected by the agent, in contrast, care less about causal relations in the absence of faulty conduct.
While conduct-based responses are warranted by the way that agents' behaviour manifests attitudes of respect, contempt, or indifference regardless of whether that conduct causes harm, consequence-based responses are warranted by the fact of a harm regardless of whether the conduct was faulty. Causality, in isolation from conduct, indicates nothing about how agents have previously viewed their relations with others. The ready-to-hand example of the significance of causality is Oedipus. Despite their initial strangeness, the characters' reactions in Oedipus Rex can be intelligible to modern readers once the magical elements of fatalism and pollution are stripped away. Reasons of consequence explain these reactions: ‘incest’, after all, describes a situation, not a content of will, or an attitude. Oedipus has, by his own actions, brought on (and engaged in) this situation, and this contingent, causal connection grounds his horror and self-reproach.19
Oedipus' response to the fact of his causal role is what Bernard Williams calls ‘agent-regret’: regret that a state of affairs obtains whose occurrence involved one's own agency.20 Agent-regret rests on no sense of wrongdoing, and is compatible with impeccable conduct, even conduct so recognized by the agent.21 However, it seems a mistake to distinguish agent-regret fully from guilt, for although an awareness of wrongful acting is a typical part of guilt, awareness of having done something awful, even if unwittingly, can suffice.22 Oedipus' response was partly shame at his incestuous disgrace. But his horrible self-mutilation can only be explained by something else, something that we can recognize as a form of guilt: a gesture at repaying a wrong he has done. The causal relation itself need not be entirely direct to trigger guilt. If, while tending a friend's cat, it slips outdoors despite my protections and gets hit by a car, I will feel not merely sorry for my friend but guilty towards her. Although the death is not my fault, I have provided for its occasion, and so my relations to her differ from those of any other sympathetic friend. Indeed, because of the friend's trust in me, I am likely to feel even worse than the driver who, also let us assume faultlessly, actually killed the cat.23
These examples bring out a striking feature of consequential accountability: where conduct is not at issue, there is an especially radical asymmetry in response among (p. 560) the various positions that the harm itself creates. My friend is unlikely to resent me, even though I feel guilty. More precisely, if the accident is not my fault, then my friend would be unwarranted in resenting my role, since I will not have acted badly, while my feelings will be warranted by my causal role and our prior relationship, as well as by the protective role I assumed towards his cat.24 Likewise, Oedipus' compatriots more pity than despise him for his crime. The principal reason for this asymmetry is that agents' causal relations necessarily inform their conceptions of themselves, of who they are. For victims, by contrast, the significance of the harm consists largely in the mere fact of its occurrence, and not its causal link to a particular agent.
The relation between agents and their effects is one of identity, in a certain sense. What an agent has caused is an important part of that agent's history and life, as important as what that agent has intentionally done, believed, and hoped for. The regret signals the fundamental unluckiness of the causal connection between this agent and those consequences. Because regret for faultless accidents maps the agent's actual (as opposed to idealized) course through the world, the general absence of such regret is found primarily among children and extreme Kantians, for whom the fantasy or ideal world is more salient than the real. As H. L. A. Hart and Tony Honoré have suggested, it is through claims of causal authorship that ‘[i]ndividuals come to understand themselves as distinct persons, to whatever extent they do, and to acquire a sense of self-respect …’.25 It is important to note, however, that ‘what I have done’ does not name a naturally limited universe of events: agents are causally related to infinitely many events, under infinitely many descriptions, and only some of those events, under some descriptions, will be salient. The concept of what an agent has done is given itself by our practices of accountability and conception of causation. Beyond bodily movements themselves, the extension of an agent's field of causal influence is given by a complex and deeply rooted normative conception.26
The shape of that conception—what causal relations are picked out as warranting a response—is the subject of an enormous literature. Although some writers have attempted to locate normative considerations in a metaphysical conception of causation, most have instead adopted a non-normative, context-neutral conception, then (p. 561) relied on pragmatics to explain ordinary usage.27 According to generic conceptions, a person's act is typically one item among enormously many causally relevant events and conditions that are jointly sufficient for an event's occurrence. As many philosophers have argued, whether that act is highlighted as noteworthy (‘the cause’), by the agent or another, depends in part upon its relation to stable background conditions, its role in durable structures of events, its susceptibility to intervention or control, and so on.28 The relevance of the agent's intervention in the cat and Oedipal cases is obvious. But I want to suggest that, in more difficult cases, agents' social and moral relations to others are especially important to agents' seeing their acts as causally connected to harms. This is particularly true of omissions, as when my failure to bring a sick child promptly to the doctor results in suffering: the nature of my accountability will depend upon my relation to the child. But my seeing myself as the positive cause of another's misery also depends upon my understanding of the structure of our mutual relations. If we are competitors in business and my low prices unintentionally drive you into bankruptcy, I may see your failure to meet my prices, rather than my own act, as the cause of your demise.29 In contrast, if we are friends and my unintentional act results in your suffering, I am likely to reproach myself for my causal role and do what I can to make amends.
My gesture of repair as an agent is, in these cases, more complicated than just the reaffirming or re-establishing of the character of a relationship between agent and victim. When I see myself as accountable for a harm I merely cause, and when repair of that harm is at least possible in part, my gesture of repair is directed at myself as well as at my victim. It is directed at the victim in so far as it is an attempt to compensate for a burden I have imposed. And it is directed at myself in so far as it provides a way for me to transform my trajectory through the world, eliminating what is unfortunate about what I have done. Here we see a further asymmetry in the responsive positions of agent and victim, particularly in cases of faultless wrongdoing: while my victim may be indifferent to the source of compensation, I may feel that it must, in symbolic part at least, come from me.30 And even if neither I nor my victim feels it (p. 562) necessary that I provide the compensation, an apology or other gesture of repair may also be called for, and that can come only from me.
This account of causation as a source of reasons warranting response may seem circular, for if merely singling out a causally relevant factor as the cause depends upon a prior conception of appropriate relations between the parties, then the relevant notion of causation is doing no independent normative work.31 The notion of cause and warranted response are indeed interdependent and so, in a sense, functionally circular, but the circularity is not vicious. We make our causal contributions in social as well as physical space; the norms and interests that define that social space inevitably play a role in helping to delineate the causal relations we perceive. Once we have identified a given act as the cause of some harm, on the basis of background expectations of appropriate behaviour, then we are led to modify our conception of that background, and so alter our future perceptions of what is a cause and what a mere condition. My friend forgives me this time for letting the cat out; either the driver or the cat itself may be regarded as the cause of its death. But if several more cats die while in my care, my friend's perceptions of my causal role in the harms, and so her responses to me, will undoubtedly change.
As I have said, the position of victims, and the responses warranted by their relations to the harm, differ dramatically from the agents' own responses, particularly in cases of faultless causation; and these responses also depend upon the way victims view their relations to agents and onlookers. For agents, their causal relation to a harm warrants feelings of self-reproach. But because the agents manifested no ill conduct or will, victims' resentment on that basis is unwarranted. No prior moral or social relationship has been devalued by the harm, but only a distribution of goods distorted. As a result, the victim's response is more likely to be a demand for compensation unaccompanied by reproach. Whether this claim for compensation is seen as having normative force, by victim or agent, is itself a product of the relationships among the parties and society at large. ‘It wasn't my fault’, when true, is a perfect excuse from accountability for conduct, but it bears no direct relationship to the moral question of compensation. Given a certain understanding of social and moral relationships, ‘that you caused it’ can sufficiently warrant a claim for compensation. (The embedding of compensatory demand in the relationships constituted by a legal system, in the form of tort law, is the subject of Section 4.)
So far my discussion has concerned the event-paradigm of greatest traditional interest to moral philosophers, when one person injures another directly. But it is worth noting that many of the harms and miseries of modern life fall outside the paradigm of direct action. Think of buying a table made of tropical wood that comes from a defoliated rain forest, or using a CFC-based air-conditioner, along with (p. 563) 10,000,000 others, and so jointly putting a hole in the ozone layer; being a citizen of a nation that bombs another country's factories in a reckless attack on terrorists; or inhabiting a region seized long ago from its aboriginal occupants; helping to design an automobile that the manufacturer knowingly sells with a dangerously defective fuel system, or working in a health-care bureaucracy that carelessly allows the distribution of HIV-contaminated blood. All of these examples are instances of a mediated relation to harm, where injury is brought about through the actions of others. And many of them are cases where what any one individual does makes no difference; only together do individuals cause harm.
These mediated relations to harm are the domain of complicity. Just as purely consequential responsibility tests will-oriented models of responsibility, so complicitous accountability puts pressure on consequence-oriented models. For it is a familiar fact of our moral and legal practices that we blame, punish, and demand compensation from complicitous agents even though what they did made no difference. The bank would have been robbed regardless, the ozone hole formed, the battle fought. The puzzle arises because, if causal contribution is necessary to responsibility, then no one is responsible, for no one makes a difference. And even when an individual difference is made, say when one person acts as lookout during a robbery, our practices of blame and rules of punishment go far beyond the causal contribution. What complicitous responsibility centrally challenges is an appealing, intuitive, principle of responsibility, that someone can only be responsible for events over which he had control. Call this the ‘control principle’.32 An account of responsibility that aims to reveal rather than replace these pervasive practices of responsibility will have to show how responsibility can outrun both causation and control, without becoming simply a free-form virtual guilt shared by all.
In other work I have tried to do this.33 Briefly, I argue that once we have in hand an analytical understanding of co-operation, a normative account of complicity follows suit. Individuals who co-operate share what I call ‘participatory intentions’, that is, intentions to do their parts of some collective act. Participatory intentions ground our basic practice of action- and outcome-ascription in co-operative contexts, so that, for example, when two of us together write an opera, you writing the music and I the book, each of us can truly say, ‘we wrote the opera’. Each of us should be regarded as an author of the opera, albeit an inclusive author, in virtue of our individual collective participation in its creation. Responsibility for it—praise or blame—then tracks the ascription of authorship. This is because the will of each can be deemed manifest in the collective product.
(p. 564) Now, differences in particular causal contributions change the responses warranted to particular individuals; it is reasonable to celebrate Mozart more than his librettist, Da Ponte. And there is a truth in the control principle: individuals who cannot control whether they participate at all (hostages or dupes, for example) cannot be held responsible for the collective harm.34 But in cases of full overdetermination, when no individual really does make a causal difference, blame (or praise) may still fairly lie. Derek Parfit's famous ‘harmless torturers’, each of whom gives a torture victim an individually imperceptible but aggregatively awful electrical jolt, provides a stark example of the problem.35 Parfit himself struggles to accommodate consequentialist ethics to a form of responsibility that seems, on its face, precisely independent of individual consequence. Others have attempted to develop a theory of causation that makes sense of such cases.36 I am sceptical, myself, whether these approaches work, even on their own terms. Whatever the ultimate account of complicitous responsibility, however, it will have to go at least partly by way of the participatory intentions of the agents—their will, independent of its effects, to join in a collective act that does injury. For in the absence of any salient individual causal contribution, surely it is the co-operation itself that explains responsibility. Implication follows participation.
3 Criminal Responsibility
A working theory of criminal responsibility presupposes an answer to one question, and must answer two more questions. It presupposes an answer to the question of what norms should define the domain of criminal law. And it must provide answers to the following questions: first, what counts as a violation of those norms; and secondly, what responses are warranted by their violation? Clearly, these questions must be answered together if they are to be answered intelligibly. If the criminal norms aim primarily at conduct as opposed to consequence, then the criteria of responsibility will emphasize causation over quality of will. If the norms protect very great or vulnerable interests, then more serious responses are likely to be deemed warranted.
(p. 565) And if the responses deemed warranted for violation are very severe, then the criteria of responsibility ought to be narrow—assuming some background political principles against the infliction of suffering or favouring the retention of individual liberty. This is not to say that a theory of one of these subjects determines answers to the other two, but only the more modest point that the criminal norms, criteria of responsibility, and responsive practices must hang together in reflective equilibrium.
Criminal norms have traditionally protected the most important interests in life, security of body, and security of possession. By protecting these interests from malicious incursion, criminal law makes social life possible by making social trust possible. Relying on state power to quell each other's urges to act selfishly or viciously, we can forge the co-operative relations that make our lives good.37 It is true that the reach of criminal norms in modern times has extended beyond these core interests into many regulatory domains. These regulatory domains often use only weak criteria of responsibility, forgoing requirements of knowledge and intent. The extended reach of criminal norms provokes worry even when the actual sanctions are not severe, because the expressive, condemnatory aspect of criminal norms carries over from the core concerns.38 But what determines whether a given interest will be expressed and protected through criminal law is only in part a function of its intrinsic importance. It is also a function of the special responsive position of the state as the expressive and enforcement agency, as well as of the state's relations to other social institutions. Demands for a moralized criminal law—a law punishing private, consensual behaviour on grounds of its immorality—reflect in their proponents not just a concern to maintain a (probably illusory) normative status quo, but also a deep insecurity about the capacity of non-coercive social institutions to govern behaviour. ‘There oughtta be a law!’ is spoken not by the discoverer of a new norm, but by someone unhappy about an old norm's current efficacy. A similar point holds for the criminalization of regulatory matters: the choice to rely upon sanctions, as opposed to tax- or market-based approaches, often reflects both an articulated judgment about the efficacy of different means to the same result, as well as more inchoate beliefs about the need for state authority to supplement private forms of social ordering.
Just as contested ideals of state–civil society relations explain debates about the allocation of authority between criminal and other norms, so they also explain debates about the proper response to violations of those norms and criteria of responsibility.
(p. 566) Discussions of warranted response have typically come in the form of different theories of punishment. Theories of punishment divide into two groups. On the one hand, there are true theories of punishment, which attempt to offer a justification for the intentional and condemnatory hard treatment of violators of criminal norms. In this group there is some discussion of the proper sort of hard treatment, whether it includes physical pain, execution, incarceration, or shaming penalties. But discussion primarily focuses on how to justify a treatment whose unpleasantness is assumed—whether it is to be justified in retributivist or expressive terms.39 The second group consists not in theories of punishment per se, but in theories of the proper treatment of offenders, where the proper treatment may not involve state-inflicted suffering at all. All such theories are, self-evidently, instrumentalist, and utilitarian theories are the most obvious examples. Hard treatment will be justified, if it is, through its role in deterring other crime or in subordinating the offender to social authority. Gentler, rehabilitative and educative theories also fall into this group, as do reparative theories—that is, approaches to offence that attempt to mend the social ties severed by the criminal offence.40 Finally come theories that represent a hybrid of instrumental and intrinsic concerns. Hart's theory of punishment, further developed by Mackie and Scanlon, takes this form: a system of punishment whose infliction is sensitive to offenders' wills is justified both as a means of maintaining civil order against a background of general liberty, and as a system peculiarly appropriate to beings who value the ability to determine by choice whether they will come into conflict with the state.41
Contrast criminal responsibility with moral responsibility. Within morality's broad limits, variety reigns. Friends and family members can reproach each other for minor defects of character as social acquaintances cannot. The fury and rage expressed by lovers at betrayal, well-warranted though it may be, would be wholly out of place even between friends. Likewise, the poignant guilt properly felt at the betrayal of a friend might well be considered self-lacerating if it were directed at all moral transgressions. Social morality is effective precisely because there is room for (p. 567) play in its joints. Though the norms governing warranted response have shifted enormously through time and across cultures, and depended crucially upon the state's eagerness and capacity to keep the civil peace, there have always been limits to appropriate response, even if those responses have greatly transgressed the generally pacific borders of contemporary Western elite social morality. I will stipulate here, however, that absent circumstances of self-defence, the limits of moral response are the limits of language and feeling. Physically violent or coercive responses to individuals are only morally permitted to the state.
Legal systems protect the interests that morality protects, centrally the means and liberties necessary to live well as a rational and reflective, project-centred agent. To the extent that legal systems do anything more than simply express (vehemently) these norms, then it is necessary to conceive law instrumentally to some extent, judging systems better or worse in their capacity to secure these interests. But this need not be a crassly functionalist conception of law any more than of morality, which also performs a function of protecting the interests and relationships that make our lives good. Law is good because the interests it protects are valuable; and legal responses are warranted by the importance of those interests. If liberties and well-being are values within the law, then legal responses that compromise those values are suspect. While the restrictions upon moral wrongdoing and free-riding that legal institutions dictate are not themselves objectionable compromises to agents' interests, the use of threats and application of sanctions to guarantee those restrictions do compromise autonomy.
It follows that if coercive measures by the state are warranted at all, they are warranted because no non-coercive measures are adequate to protect social interests once moral and legal forms of accountability have failed. Unlike social and moral responses, whose verbal or emotional nature is only of concern to those for whom the relationships they protect have value, coercive responses are of concern to any self-interested agent. While legal systems may depend primarily, as Mackie suggests, upon the efficacy of an adverse legal characterization of certain acts, coercive threats play an essentially ancillary role in motivating those unswayed by a desire to maintain morally appropriate relations.42
The interests justifying legal responses themselves limit those responses. If, as under liberal regimes, legal systems aim to protect meaningful forms of individual autonomy and social co-operation in general, then individuals' autonomy interests will be of concern as well in the administration of legal sanctions. As Hart (and Scanlon following him) has argued, this concern for autonomy, rather than a concern for rectifying moral wrongs, can best explain the general restriction of penal sanctions to cases of voluntary conduct.43 By making the infliction of those legal sanctions that severely infringe individual autonomy depend upon the choices (p. 568) individuals make, the state has done what it can to ensure the autonomy of each citizen. Due process considerations also serve to protect individual autonomy from undue state interference. The concern for autonomy also helps to explain the criminal law's ‘act requirement’, that only voluntary attempts and commissions are punishable, and not inchoate plans or involuntary movements.44 Because who an agent is and what an agent causes are far less sensitive to choice, criminal punishment on these bases is far more restricted.45
The debates among theories of punishment have famously tended to stress an ideal of the person, enhanced or compromised by the relevant punitive practice. Immanuel Kant famously denounced the ‘serpent-windings of utilitarianism’ on the grounds that it uses the offender simply as a means of general social control, thus failing to respect him as a rational agent meriting concern for his own ends.46 By contrast, critics of the retributivist ideal preferred by Kant, according to which it is intrinsically good or right to ensure that wrongdoers suffer, have worried that talk of the rightfulness of punishment served mainly to mask the punisher's desire to humiliate, a desire coming from a sense of resentment not justice. As Nietzsche put it (with characteristic exaggeration), Kant's ‘categorical imperative reeks of cruelty’.47
Disputed ideals of the person do drive the debate over punishment, but to focus only on the person punished involves a kind of ethical solipsism. Equally important is an ideal not of the individual but of social and political relations. Different theories of punishment implicate, and are implied by, different conceptions of the proper relation of the individual to the community. What must strike anyone working in the area of punishment theory is the way in which different theories have come to dominate or recede, it seems, as a matter of shifts in broader political views. When Hart began writing on punishment, for example, the philosophical status quo was reformist and rehabilitative, not retributive. Along with other writers, he rejected rehabilitative theories out of a concern about the reduction of the offender to a psychological system to be manipulated by the state.48 Hart rejected retributivist theories of punishment as well, partly on the familiar conceptual ground that they (p. 569) depended upon a ‘mysterious piece of moral alchemy’ that made two ordinarily impermissible acts amount to justice.49 But his conceptual argument (or observation—he hardly took retributivist theories seriously enough to argue against them) can be fairly seen as a product of a general sentiment that retributivism in punishment was faintly barbaric, as compared with enlightened utilitarian social policy. By the 1980s, however, retributivist theories had come to flourish, propelled in significant part by the work of Michael Moore and Andrew von Hirsch.50 It seems hardly coincidental that fashion in philosophical theories of punishment has tracked fashion in political practice (or vice versa), as particularly US penal policy has shifted from rehabilitative to fiercely punitive practice and increasing emphasis on individual rather than social responsibility.51
This sociology of recent theorizing is meant to do more than point out the obvious fact that philosophers too are creatures and creators of the Zeitgeist. It also demonstrates the analytical point that theories of warranted response must be interpreted in terms of the background conception of social relations they presume. I have already mentioned how rejection of rehabilitative theories was early driven by worries about the therapeutic politics they presumed.52 The rejection of utilitarian theories has as much to do with their treatment of individuals as means, as with the more general, managerial conception they hold of the state. In utilitarian political theory of a crude but familiar kind, the state is conceived as an expert at social engineering, attempting to maximize net social satisfaction.53 Doubts about utilitarian political theory, related both to its implicit dependence on expertise and to its failure to see individual members of society as co-operating agents, not just joint consumers, have led to its displacement in the field of distributive justice.54 A theory of punishment resting on a conception of the state as social manager is equally undermined by these doubts. Retributivist theories, with their emphasis on individual dignity, point up the defects of utilitarian views. But retributivists have thus far failed to come up with a conception of the state that makes the infliction of just punitive deserts a legitimate objective (p. 570) of the state.55 It certainly is possible to conceive of the state as the people's agent in delivering deserts, both retributive and distributive; but this conception is hardly uncontroversial, resting as it does on a metaphysically robust and pre-institutional understanding of desert.56 The reparative justice theories now emerging reflect a more communitarian ideal of social relations in their focus on reconciling individuals with their societies. Unless there is reason to think that some particular political conception will come to hold sway—and I see no such reason—the relationality of responsibility means that debates among punishment theories will go unresolved.
The concomitant of the relationality of criminal responsibility is its positionality. It is not merely a legal conceit that while the prosecutor represents ‘the people’, the court represents impartial justice. For the position of justice taken by the law is very special and circumscribed. When legal institutions assume the partisan position of the victim and the posture of resentment, the rights and liberties of defendants are severely compromised, situations for which the sedition trials of the twentieth century are the best exemplars.57 The warranted response of victims to hostile behaviour is resentment; but resentment is wholly inappropriate from the institutions of justice. The position the criminal law represents is not simply an integration over all social and moral positions, and legal responses do not represent whole, overall responses to wrongs. Instead, legal responses are ideally made from a particular position, that of the state, and represent one form of response among many. Regardless of what individuals deserve, the state's responses flow from the relations that tie each individual to one another, agent and victim alike, and are limited by the claims internal to those relationships.
Reminding ourselves of the special position of the state is particularly helpful in getting a handle on the old chestnut of theories of punishment, why unsuccessful attempts should be punished less severely than successfully completed crimes. On one side is the view that the proper basis of punitive, as opposed to compensatory, responsibility is either the social danger of the defendant's conduct, the contempt for legal norms evinced by that conduct, or both, and that these bases are the same for unsuccessful and successful attempters alike.58 Since the basis of responsibility is the same, there is no reason to punish differently. Adjusting punishment to actual harm, on this view, simply confuses punitive and compensatory responses. Proponents of (p. 571) differentiated punishment, by contrast, observe that our moral responses as a matter of fact track the harm we do. We blame ourselves more when misjudgment results in real harm; and we resent more the malevolent acts of others, simply because those acts cause us harm.59 The connection between this bit of moral phenomenology and state punishment is a retributive theory of punishment, according to which the state's role is to administer a (univocal) moral desert.
The relation between luck and responsibility is deeply vexed, and it is unclear, to say the least, whether our practices of responsibility can be fully regimented or rationalized in terms of specific bases of response.60 Reconciling responsibility with luck is a deep problem, perhaps an insoluble one, for moral theory. But it must only be solved for the theory of punishment if state punishment ought genuinely to mimic interpersonal moral response and resentment—whether it really is to be Sidgwickian ‘resentment universalized’.61 If the argument for distinguishing successful from unsuccessful attempts can be given no firmer basis than coherence with interpersonal moral practice, then there is no good case to adding actual harm as a factor in calculating punishment, independent of social danger and antisocial will. To do otherwise is to confuse the particular purposes of the criminal responsibility system with the more general expressive and constitutive functions played by our practices of moral responsibility.
The second main point of intersection between criminal law and philosophical interest concerns the criteria of responsibility. In Anglo-American law, the criteria of criminal responsibility converge with the criteria of moral responsibility: where moral claims are warranted, so generally is legal sanction; and where there is moral excuse or justification, so too there is legal excuse or justification. While the expressive function of criminal law makes overlap between moral blame and criminal guilt likely, the very high degree of convergence in modern doctrine and statute is the product of the concerted effort by a number of criminal scholars, notably Hart, Sanford Kadish, Herbert Wechsler, and Glanville Williams, to limit the encroachment of strict liability doctrines.62 That said, any specifically legal conception of culpability must recognize that legal authority is always exercised in doubt. Criminal law presents the most serious epistemic problems, given its focus upon individual intentions. Intentions are inferred from scatterings of circumstance, causal explanations (p. 572) are shaped by the interests of the contesting parties. Unfortunately, just resolution of cases requires good information; and good information is generally expensive and difficult to obtain. No individual accused of a crime can be expected, practically or normatively, to divulge a culpable state of mind. The distinction between premeditated and spontaneous homicide, for example, can be the difference between execution and incarceration. Premeditated homicide can be proven by evidence of advance planning. However, the most subjectively inclined courts have held that killing can count as premeditated in the absence of planning, so long as the accused has the opportunity to reflect on the decision to kill.63 Since obviously no killer will admit to premeditation, and since there is rarely a surviving witness to the crime, the judge or jury's decision often teeters upon a scaffolding of circumstantial evidence and psychological inference.
Despite pervasive doubt and uncertainty, decisions must be made and distinctions drawn, whether in the name of retributive justice or credible deterrence. It is therefore no wonder that evidentiary matters play a central role in the criminal process. Some of the restrictions upon the evidence that can be procured by the state and brought to bear in the courtroom, such as the requirement of a duly authorized warrant for a comprehensive search, stem from a generalized concern about the limits of police intrusion. But other restrictions reflect fundamentally epistemic concerns, such as the exclusion of evidence of a defendant's prior criminal history, or of hearsay reports of the defendant's statements. Although prior criminal history is clearly relevant to the proof of crime in question, such evidence is rightly excluded in many cases on the grounds that its effect upon juries is more prejudicial than probative.64 Without these protective evidentiary rules, a system of criminal responsibility could not possibly be applied in justice.
Contrast the circumstances of criminal justice with those of moral theory. Although moral philosophers since Kant have warned of the inscrutability of individual intention in the first as well as third-personal cases, most moral theories ignore these epistemic problems, including Kant's own moral theory.65 Deontologists focus on agents' underlying intentions and self-conceptions; and utilitarians resort to the idealized fiction of fully informed, ‘ethical’ preferences in order (p. 573) to justify their criterion of right action.66 Whether agents have acted wrongly and are accountable for so acting thus depends upon deep facts about their deliberative and motivational capacities, fine-grained attributions of intentional content, unequivocal motivation, and empirically adequate predictions of future consequences.
In general, the moral judgments we make and the responses we offer may be wildly out of line with the evidence necessary to support their application. The jerk who cuts me off on the highway may be distracted by great personal loss. But this possibility is unlikely to stop me from thinking him a jerk. Moral theory and practice can live by idealized epistemic standards because the stakes in the moral game are low in any particular case. The relationships that social morality plays the dominant role in protecting can usually be repaired through apology and understanding. I may unfairly resent your failure to meet me, not realizing that you had a sick child to take care of. When you have a chance to explain, or when I otherwise discover the reason for your absence, all is again put right between us.
By contrast, the belated acquittal of someone unjustly convicted puts little right, for nothing can repair the violence done to one's sense of autonomy and worth by unjust punishment. To be imprisoned, publicly despised, and stripped of elementary civil rights is to have one's political, social, and moral identity undermined or lost: it is to become an object of the state's authority, rather than a subject who authorizes the state's exercise of that power.67 Freedom and compensation may be valid claims stemming from unjust process or sentence, but they are not a full means of repair. Given the moral and human costs of wrongful conviction, it surely follows that a necessary condition of a just penal institution is that it make very few mistakes. Legal judgments have little point unless actually applied and enforced; they are worthless merely as indicators of moral norms. But in order to be legitimate, legal judgments must be well-rooted in both fact and political morality. Legal theorists and moral philosophers who distinguish sharply between normative and evidentiary issues run the risk of ignoring the social space, with its costs and limitations, in which legal rules are necessarily embedded.68 The problem is that an awareness of the law's epistemic constraints can quickly become licence for a cavalier cynicism about alibi and excuse. To the extent that exculpatory considerations are narrowed because of difficulties of proof, so broadens the scope of legal intrusion.
(p. 574) Within these epistemic constraints, two different kinds of criteria are generally relevant to criminal responsibility: criteria of capacity, and criteria of intentionality. The capacity demanded for criminal liability is, roughly, that demanded for moral responsibility: a capacity to govern oneself by practical reason, responsive to the moral and factual considerations that obtain. The requisite capacity for practical rationality evidently incorporates a number of different components: a perceptual component, for establishing the nature of one's environment; a conative component, through which one finds some possibilities of action desirable as goals and others undesirable; an evaluative-cognitive component, for weighing the reasons for and against the potential goals; an instrumental-cognitive component, for determining how to realize those goals; and a volitional component, through which one actually acts on the desired goals.69 Note that there is nothing about the having or exercise of this capacity that is incompatible with causal determinism.70 This is not meant to beg the free will problem, but only to point out that the metaphysical capacity to act otherwise demanded by incompatibilists is a further requirement, going beyond the core practical capacity.71 And even if compatibilist understandings of moral responsibility are not, finally, acceptable, it is plausible to argue that here is one point where legal and moral criteria of responsibility may reliably diverge. For the moral notion may well be thought to import a theologically or metaphysically ambitious conception of responsibility, related to divine judgment or existential meaningfulness. The sublunary ambitions of law, meanwhile, might be satisfied with a conception of fair attributability for which the practical reason capacity suffices.72
Metaphysical debates notwithstanding, those persons without the capacity to reason practically are manifestly to be treated or incapacitated, not punished. It remains a vexed question in law and philosophy what sort of rational incapacities fatally undermine culpability.73 The traditional M'Naghten requirement is that the (p. 575) defendant will only be excused from responsibility if he does not ‘know the nature or the quality of the act he was doing; or if he did know it, that he did not know what he was doing was wrong’.74 This is obviously an extremely restrictive definition, according to which a defendant who understood the wrongness of his act, but was compelled to do so by Satanic voices in his head, would not be excused. The purely cognitive definition was therefore expanded by the American Legal Institute into the requirement that the defendant be able to ‘appreciate the criminality of his conduct’ and ‘to conform his conduct to the requirements of law’.75 Other definitions have been put forward as well, notably the short-lived but famous definition offered by the federal court of appeals for the D.C. Circuit, that a defendant be deemed non-responsible if his unlawful act was ‘the product of mental disease or mental defect’.76
None of these definitions fully captures the idea of incapacity that juries almost certainly operate with, but all indicate the general scope of the questions surrounding criteria of responsibility.77 The genuinely difficult questions come at the margins of capacity, for example, with agents who suffer delusions but who know that they do so (as with many schizophrenics), or with agents who recognize the wrongness of their acts but who seem completely to lack ordinary concern for wrongness.78 Given the general lack of knowledge about the nature of mental illness and the inhospitability of legal proceedings to nuanced discussion, epistemic constraints are tightest in this domain. This is mainly a problem for retributivists, who may find that any operationizable criteria of mental capacity will err, either by demanding treatment for those who ought be punished, or by demanding punishment for those demanding treatment. For instrumentalists, the choice between incapacitating and punitive responses is less significant.79
Moral responsibility hinged on conduct, in the sense of intentional activity, and causation. The most serious moral responses, such as blame and recrimination, fall where conduct and causation run together: when the agent causes harm with a will that evinces lack of respect for another's interests. Criminal responsibility follows suit. Liability for most crimes is based upon a combination of criteria regarding the defendant's bodily acts and their consequences, and the intentions, knowledge, or awareness with which those acts were done and their consequences produced. The terms ‘subjective’ and ‘objective’ are used, respectively, for the intentional and the (p. 576) conduct, circumstance, and consequence criteria, often also called the mens rea and actus reus elements of a crime.
Unfortunately, ‘subjective’ and ‘objective’ are also used for a wholly different contrast in criminal law, to distinguish between individualized and normalized standards. In this sense, subjective criteria predicate liability upon the actual capacities and beliefs of the agent, while objective criteria predicate liability upon the capacities and beliefs that could reasonably be expected of a generally competent rational agent.80 At the risk of departing somewhat from standard legal usage, I will use the terms ‘individualized’ and ‘normalized’ for this sense of ‘subjective’ and ‘objective’ criteria. Philosophical questions arise about both sorts of criteria.
Standardly, criminal liability requires that a single agent perform the specified acts or cause the specified harms, with or because of a specified mental state or states. First-degree murder, for example, requires the subjective element of a premeditated intention to kill, as well, of course, as the ‘objective’ result that the agent has caused another's death in acting upon that very intention. Second-degree, or ‘depraved-heart’ homicide, does not require killing as an aim, but does require that the defendant believe killing a likely consequence of his actions. For some crimes, the defendant's mental state must be highly determinate: larceny requires not only the objective taking of another's property, but a subjective intent to deprive the other permanently of that property. And there are many crimes that can be committed with a still culpable but not intentional mental state, such as recklessness or gross negligence. The defendant must be engaged in some activity intentionally (e.g. driving), but need not be driving with an intent to kill to be found guilty of vehicular homicide. It suffices if the defendant's objective conduct consists of driving, that conduct causes a death, and, for recklessness, that he is aware of the risks his driving presents. Crimes committed negligently must be handled differently, for the question is not whether the defendant had any particular mental state, but whether he lacked a state he should have had, namely attention to the relevant risks.81 Clearly, subjective and objective criteria interpenetrate, for the objective conduct component is itself intentional—for example, the taking of property, or the killing of another—and may merely be accompanied rather than caused by the relevant subjective state. The subjective component is not, therefore, generally an explanation of the conduct, but rather a mental state relevant to the assessment of the defendant's moral culpability.82
(p. 577) The subjective and objective criteria of responsibility invoke the traditional analytical philosophical problems of giving an account of intentional action, including the problems of relating intention to bodily movement, individuating acts, and intentional omissions; and writers in criminal legal theory have pursued these philosophical problems.83 But it is unclear that a legal theory needs a deep philosophical account of these problems. Take the problem of act-individuation: a defendant throws down a match, thus setting fire to a house and killing the inhabitants. There is a philosophical dispute between so-called fine-grained individuators, such as Alvin Goldman, who argue that the defendant performs many different acts (throwing down the match, burning down the house, and killing the inhabitants), and coarse-grained individuators, such as Donald Davidson, who argue that the defendant performs but one act, the bodily movement of throwing down the match, which act can be described in many different ways, as a house-burning, inhabitant-killing, and so on.84 The relevant questions of criminal responsibility are, however, neutral between these issues in action theory. These questions include whether, for example, the burning or the killing can be traced causally to the match throwing; whether the defendant intended the burning or the killing, or was reckless towards those consequences; and whether the burning and the killing merit independent, cumulative punishments. They are not purely metaphysical questions or problems of action theory. They are, rather, normative, and will be answered instead by reference to a theory of punishment. The terms of these theories are the folk or commonsensical notions of deliberation, foresight, intention, and action; the normative challenge lies in relating these terms to a scale of culpability.85 Even the difficult questions raised by automatism and mind control must be answered in terms of a normative theory of responsibility and the criteria of self-governance, theories which need only presuppose and not analyse the basic idea of doing something for a reason.
More difficult philosophical questions about responsibility arise regarding the question whether the punishment system should deploy individualized or normalized criteria of responsibility. Claims of specific (not general incapacity) excuse from responsibility or justification are the main place issues of individualized and normalized criteria and arise, typically when the defendant unreasonably believes that justifying or excusing circumstances obtain. A defendant who commits a crime under the unreasonable belief that his life is being threatened might plead duress or self-defence. But these issues also arise with crimes defined partly in terms of results, such as homicide. Return to the arsonist, and suppose that anyone reasonably intelligent (p. 578) would have realized there was a substantial chance people might be sleeping in this house.86 But this defendant was in fact so addled or unintelligent that she was not in fact aware of the risk. Should she be punished for reckless homicide, none the less—that is, causing death not intentionally but with a conscious disregard of the relevant risks—on the grounds that any reasonable person would have been aware of those risks, even if she was not? The case for an individualized standard, which would acquit in this instance, is that however indefensible her conduct, she must evince the particular culpable mental state which the law targets. If the justification for punishing someone with that mental state depends on the wickedness of agents with that state, then punishment is morally unjustified; and if the punishment's justification is deterring agents from acting recklessly, then it also misses its target, since she was not, by her own lights, acting recklessly. Similarly in the excuse and justification contexts, there is a strong moral case for individualized standards: the defendant simply did not have the ill will targeted by the criminal norm. Nor is the failure of deterrence in these cases worrisome since, in the excuse case, the norm is not expected to deter in such circumstances, and in the justification case, the norm should not deter.87
Now, these considerations only reach so far. Even if she was not specially deterred by the punishment, others might generally be, and might also be dissuaded from acting recklessly with the hope of being acquitted on erroneous individualized grounds. Secondly, at most these considerations show that punishing her for reckless killing is unwarranted; punishment for negligent killing might still be warranted; and there is no obvious reason to distinguish sharply between the punishment schedules for each. Thirdly, the epistemic limitations of the criminal process may suggest that a fairer process will be one that deploys normalized standards rather than one that is likely to fail if it attempts to discern individual beliefs. These are largely pragmatic considerations. But some, most recently Arthur Ripstein, have tried to make a positive, principled case for normalized standards.88 Criminal norms are devices for allocating autonomy, where ‘autonomy’ means control over person and property—as I put it before, they define a minimum normative content for social and moral relations. In a liberal state, a legitimate system of criminal norms allocates autonomy (p. 579) equally, giving each citizen the same measure of protection and control. The defendant here failed to take the interests of others into account, not by acting badly in the face of awareness of the relevant risks to potential victims, but by failing to consider the risks at all. If she is acquitted, the victims will, in effect, have been deprived of the measure of autonomy to which state norms entitle them. One need not think of punishment as compensation to the victims to think they have a claim on state punishment here. Given the necessarily expressive dimension of punishment, an acquittal may be thought to signal that the state condones the way in which the defendant failed to give due regard to the victims' interests. As Ripstein puts it, the state would otherwise condone the defendant's substitution of private rationality for public reasonableness. Nor does there appear to be unfairness towards the defendant. Assuming she had the capacity to advert to the risks, the norm under which she is punished is a reasonable constraint on her behaviour, and so she has not received less protection to her own autonomy than to which she is entitled.89 After all, she could have avoided punishment altogether simply by not torching the house.
Of course, this argument, like the argument for subjectivism, might be taken instead to support the more limited point, that there be some state response to the particular flaw in the defendant's conduct, namely, that she caused harm through unreasonably failing to advert to the relevant risks. What the argument shows is that there should be a criminal norm prohibiting negligence. Punishment for negligence still incorporates normalized standards of conduct into the criminal law, but by establishing a separate criminal norm. Declining to integrate normalized standards into particular offences, however, may well serve purposes of analytical clarity, as well as focusing attention on the normative question of what response is appropriate for the particular kind of conduct engaged in by the defendant. The same point holds true for claims of excuse or justification founded in unreasonable beliefs: there is clearly a justifiable (and often taken) middle path of treating these as cases of ‘imperfect defences’, and mitigating but not eliminating punishment.
What the dispute between individualized and normalized standards ultimately reveals is, again, how important it is to see criminal law's criteria of responsibility as constitutive of interpersonal normative relations. The debate cannot be settled without an account of the conduct citizens owe one another, the specific meaning and response demanded by failure to meet that standard of conduct, and the role of the state in creating, expressing and defending that standard. The impulse towards individualized standards comes from a view of the state as principally responsible for denouncing or punishing failures to meet that standard; the impulse towards normalized standards from a view of the state as principally responsible for ensuring a fair allocation of autonomy among citizens. Hart famously argued that the law of excuse should be understood not to conceal a particular moral conception of responsible agency, but rather as a way of maximizing citizens' liberty against the (p. 580) background of a deterrence system, by maximizing citizens' ability to control the incidence of coercive force.90 Hart's view offers a healthy reminder of the importance of understanding the distinctive relation between the state and citizens in a liberal order, by relating it to the political value of autonomy instead of moral values implicated by retribution. But it is too narrow a view, for we also expect the state to express the moral force of the conduct norms we set for ourselves. We must not complicate our understanding of the state's functions, and so complicate our understanding of the criteria of criminal responsibility, when we realize that these criteria define both our relations to one another as well as our relations to the state.
4 Legal Responsibility for Accidentsas
In contemporary legal theory, criminal law concerns responsibility for acts and tort law, responsibility for outcomes. As we have seen, this theoretical contrast can mislead, since one can be criminally responsible for the consequences of one's conduct (e.g. murder), and one can be responsible in tort on the basis of one's conduct (e.g. an intentional injury). What chiefly distinguishes tort from criminal law is the nature of the warranted response: tort law governs the state creation of a compensatory response from the agent towards the victim, while criminal law, at least conventionally, solely involves a response from the state towards the agent.91 Tort and criminal law should thus be understood as complements, not necessarily treating different objects of responsibility, but as involving different responses. The complementary nature of criminal and tort liability is worth bearing in mind even in those instances where only one form could lie, as when someone violates a criminal norm without causing any harm (e.g. a failed attempt), or causes harm without transgressing a criminal norm. The latter is the domain of accident, when ordinarily permissible activities go awry and cause harm.
If tort law is defined as the legal norms governing compensatory responses from injurers to victims, then it is apparent that tort law is only one of many possible systems treating responsibility for accidents. Rather than dictate or enforce responses between injurers and victims, the state could, for example, simply ensure victim (p. 581) compensation through a mandatory insurance fund, as in New Zealand.92 Or the state could make injury an occasion for punishment, leaving victims with only the moral compensation of seeing justice done. But the system on which most of the world has converged takes as a central feature the linking of injurers and victims through the enforcement of private compensatory response. Accordingly, the task modern legal theorists have set for themselves is a defence of the legal practice of accountability for accidents.
The range and depth of modern theories of tort is great, and interested readers should turn to Arthur Ripstein's chapter in this volume for a survey. What I will do instead is indicate some of the general patterns of theorizing and to show how the choice among them mainly turns on the ideal of interpersonal and political relations they presuppose. One major division runs through modern theorizing, between what can be called allocative and attributive theories of responsibility, each side of the division reflecting a different conception of the relation among individuals and between them and the state.93 Allocative theories of responsibility treat accidental harm as an incident of communal life, to be handled collectively in the first instance, with individuals bearing liability only if that serves the collective interest. Attributive theories, by contrast, treat harms as problems for individuals; the task of a legal system is to recognize and enforce the reparative obligations individuals have towards one another.
Consequentialist theories, of which the economic models are the most thoroughly worked out, typify the allocative approach. A normative goal is posited, for example, utility or wealth maximization, and then various principles are defended on the grounds that when accident costs are so allocated, utility will indeed be maximized, through readjustment of incentives, spreading effects, and so forth.94 It becomes an empirical question whether, say, a fault principle best achieves the normative goal. Moreover, pursuit of the consequentialist goal may dictate principles that depart very far from ordinary tort practice, such as simply allocating the costs of the accident among the wealthiest. It will thus be purely contingent whether the legal principles of tort reflect anything like the common-sense moral paradigm of injurer-repair. But non-consequentialist approaches may also have an allocative structure, such as Jules Coleman's earlier ‘annulment theory’ of tort. According to the annulment theory, the purpose of tort law is to ensure the rectification of wrongful losses and wrongful (p. 582) gains, where ‘wrongful’ is determined by reference to the norms governing legitimate transfer of holdings and liberties.95 On this view, the function of tort law is to maintain the allocation of holdings provided by the prevailing scheme of distributive justice. Still other allocative approaches are suggested by egalitarian theories, for example, following the principle that accident costs ought to be distributed in such a way that both preserves an equal initial distribution of resources and demands a display of equal concern by individuals.96
Alternatively, the costs of accidents might be allocated so as to maximize individual autonomy, with autonomy conceived broadly in terms of individuals' capacity to engage in effective planning.97 All these allocative principles might generate the same set of operational principles—fault-based injurer liability, no-fault social insurance, strict liability—but the emergence of those principles would be in each instance grounded in a collective responsibility for the costs of accidents. Allocations of responsibility to individuals are derivative.
Attributive theories of tort law, by contrast, make individual ascriptions of responsibility primary. The theory of individual responsibility may be moral, in the sense of being prior to political institutions, or it may be political; but the task of a system of legal responsibility is to give effect to the underlying claims and duties of individual responsibility. So-called libertarian theories of tort law exemplify the attributive approach.98 On a libertarian view generally, agents are regarded as entitled, as a matter of pre-political, natural right, to the profits of their causal interventions. Costs would then be treated symmetrically, as also the entitlement, albeit unwanted, of productive agency. It is a further consideration on the libertarian view that injury diminishes the legitimate entitlements of the victim, depriving him of (some of) the value of his holdings in a way inconsistent with the norms of legitimate transfer.99 But the central concept is one of responsibility for one's accidents, where responsibility is understood in terms of causation.
(p. 583) Now, theories grounding responsibility on causation suffer from a crippling defect, familiar already from my discussion of moral responsibility: in a metaphysical sense, a broad variety of conditions and events count equally as causes of a given harm; only pragmatic, normative criteria can distinguish them. As Ronald Coase pointed out, most accidental injuries arise from an interaction between plaintiffs and defendants—one walking while the other is driving, one using a product while the other is producing it, and so forth.100 Indeed, in the modern world of mass torts and mass production, causal criteria hardly exclude anyone from liability. Take the Bhopal disaster of 1984, when a pesticide plant leaked poisonous gas, killing thousands of nearby residents. The disaster seems to have been the product of lax supervisory and maintenance standards at the plant, under-trained employees, understaffing as a result of low profits, the absence of effective regulatory authority within the relevant Indian ministries, inadequate monitoring by US headquarters, much less by Union Carbide shareholders; coupled with the decision by residents to move to or remain near an industrial facility whose central product was highly toxic.101 Clearly, different causally implicated parties bear very different levels of responsibility for the tragedy. So causal criteria at best determine a range of liability candidates. Only by reference to further, normative criteria can one party be designated ‘the cause’, or one ‘injurer’ and the other ‘victim’, terms that load a direction of causation and not merely a description of harm.
Others have offered attributive theories of legal responsibility grounded in a richer notion of moral responsibility than mere causation. Ernest Weinrib, for example, treats tort liability as simply the reflection of individuals' moral responsibilities to remedy the rights they infringe. Moral compensatory responsibility rests, in turn, on a basically Kantian understanding of the requirements of practical reason. A rational agent who wills an act must perforce accept responsibility for the consequences of that act; to impose the costs of one's act on others willy-nilly is to fail to respect the demand that one act only in accordance with principles that all might follow.102 Compensatory responsibility is self-attributed, in the sense that it follows from the exercise of practical reason.103 Jules Coleman's intermediate work, Risks and Wrongs, similarly ties legal reparative obligations to moral claims of compensation, claims grounded in a normative conception of individual agency.104 Unlike his earlier, (p. 584) purely allocative, ‘annulment theory’ of tort which focused on the general claim of victims that their wrongful losses be remedied, Coleman's newer agency-centred theory aims to show the special moral obligations agents have ‘to repair the wrongful losses for which they are responsible’.105 This principle of corrective justice, Coleman suggests, is simply immanent in our particular and contingent social practices.106 However, the extension of this moral principle to legal responsibility is indirect. A legal system may implement the corrective justice principle directly, through the sort of enforceable, individualized reparative obligations characteristic of the tort system. Or it may not implement corrective justice, and instead implement some sort of purely allocative scheme in which individual reparative obligations do not directly figure.107
Finally, there are theories combining both allocative and attributive aspects. Stephen Perry, building on Tony Honoré's notion of ‘outcome-responsibility’, offers a theory of tort liability grounded independently in agents' moral responsibility for the outcomes they produce and over which they have control.108 This claim of responsibility, not yet rising to a compensatory duty, flows from the phenomenology of agency, as I discussed above in reference to moral responsibility: our self-understanding as persisting, embedded agents depends, in part, upon our seeing ourselves as marking the world.109 As Perry realizes, the interactive contexts that dogged libertarian theories pose a challenge for him as well, for injuries arising from intersecting activities will typically reflect the agency and control of all parties. (Though you hit me with your car, I might have chosen not to go for a walk, and so I equally controlled the outcome.) Thus Perry supplements the notion of outcome-responsibility, which limits prima-facie candidacy for liability, with an allocative principle according to which accident costs ought to lie with those at fault, or who otherwise impose unusual risks on others. Similarly, Coleman, with Arthur Ripstein, has recently put forth a conception fusing allocative and attributive considerations.110 On their view, corrective justice is still a matter of instantiating the attributive (p. 585) principle that individuals must bear the costs of their own conduct. This principle is immanent in a contingent set of social practices and not, as with Perry, derived from a moral theory of agency. But the question of which costs individuals ‘own’ should not be understood simply as a matter of social convention. The question of cost ownership must, rather, be answered by reference to a political theory concerning the proper allocation of risk and responsibility.111 In short, Coleman and Ripstein make a political, allocative principle primary and then attribute specific reparative duties on the basis of the liability criteria it specifies, while Perry makes a moral, attributive principle primary, and then deploys a political, allocative one.112
The debate among tort theorists partly reflects different descriptive concerns: some theorists, such as Epstein and Calabresi, meant their contributions to be largely revisionary, while others, such as Perry, Coleman, and Posner, have claimed to be providing accounts sensitive to the actual content of doctrine, albeit accounts that aim to justify that doctrine. But, as with debates about criminal responsibility, what is really at stake are the distinctive ideals of social relations the views manifest. Further attention to this point by theorists might obviate the pressure to find a basically a priori argument for a moral or political principle justifying reparative obligations.113 I argued in Section 2 that some notion of responsibility is clearly rooted in the experience of agency itself, as well as demanded by the facts of communal, conflicting life. But the responses specific to that notion, in other words the content of responsibility, will inevitably be a product of specific institutional arrangements and social life. It is, of course, a task for historians to document the emergence and transformation of the principles structuring tort law—as has been done, for instance, for the fault principle in Anglo-American law, showing its subsequent limitation in workplace and product contexts as a response to economic, social, and intellectual pressures.114 The philosophical point is not to reject tort theory in favour of history, (p. 586) but rather to recognize the central place that contingent social norms must play in even a philosophical account.
Return to the central debate between purely allocative, economic theories and purely attributive, corrective justice theories. Even assuming that a purely attributive theory can deal with the problem of interaction, the choice between theories depends primarily on a normative conception of the social arrangements to be regulated under the appropriate regime. With highly regularized domains of activity, such as automobile driving, industrial employment, and perhaps mass production and consumption, the systemic, managerial model of social relations presupposed by an economic approach seems both appropriate and attractive.115 These are, in other words, the domains in which a public regulatory response seems correct: they present a collective problem of managing, spreading, and reducing costs, arising out of a generally valued and common activity, and in which the state can legitimately and effectively exercise authority. Within such a specified domain, the anti-individualism and cross-individual trade-offs that characterize the economic approach can be cabined, unthreatening to more general political ideals of individualism. By contrast, an untrammelled extension of an allocative approach to the general run of activity may indeed threaten those social and moral ideals. But much depends on the degree to which legal forms of responsibility are understood to reflect moral forms; and this too will surely vary with the relevant domain of activity, and with the particular social understanding of the relation between law and social morality. In the absence of a conception of such relations, the idea that an agent has a duty to pay compensation is empty. In some social conditions, ideals of personal responsibility and individual autonomy may indeed be threatened by tort doctrine.116 In other conditions, the subsumption under allocative principles of even quite a broad range of activities may simply reflect an underlying collectivist ethos.
Again, my point is not that philosophical reflection on legal responsibility is beholden to particular cultural practice. An especially valuable form of philosophical activity is to point to alternative, more desirable social and political ideals, whether or not these are actually instantiated in legal practice, or are otherwise internal to the culture. I do not, above all, mean to endorse a blanket relativism towards social and legal practice; they are, of course, open to any manner of rational, critical treatment that one's meta-ethics provide. Another task is to engage in philosophy's traditional task of conceptual clarification, attempting to render perspicuous the principles and ideals animating a given legal culture, as well as showing what those ideals logically entail. What I mean to point out is simply that a relational understanding (p. 587) of tort law brings out the relevant dependence on social and moral ideals, and properly focuses attention on the normative crux of theoretical debates.117
I began with a catalogue of the many uses of responsibility, but this chapter has generally sought a unity within the subject. I have emphasized the way claims of responsibility can only be understood as specific social practices, responsive to a background set of social, moral, and political relations and ideals. This basically Strawsonian path through the thickets of responsibility seems to me independently correct, as a way of illuminating important features of claims and responses of responsibility. But it also casts a useful light on a set of debates within legal theory, between retributivists and utilitarians in criminal law, for example, and between allocationists and attributivists in tort. These debates seem currently at a philosophical standstill, though they shift from one decade to the next. But the general turn in moral and legal theorizing about responsibility, towards a relational conception, gives reason to hope that these debates may begin to move ahead, as their adherents confront and attempt to justify the ideals their accounts presuppose. And reconstructing responsibility has importance beyond what it shows about philosophical debate. For it is in understanding responsibility that we see ourselves as actors, creators, empathizers, and sufferers. It is in understanding responsibility, in short, that we know ourselves as persons.
I am grateful to Michael Green and Meir Dan-Cohen for their comments and criticism, as well as to students and colleagues in the Jurisprudence and Social Policy Program, to whom I have presented many of these ideas in different form. I am also grateful to Cambridge University Press for permission to re-use passages also appearing in my Complicity: Ethics and Law for a Collective Age (New York: Cambridge University Press, 2000).
(1) H. L. A. Hart, ‘Postscript: Responsibility and Retribution’, in Punishment and Responsibility (New York: Oxford University Press, 1968), 210–37, 211.
(2) I ignore the question whether omissions can be, strictly speaking, causes.
(3) R. A. Duff, ‘Responsibility’, in E. J. Craig (ed.), Routledge Encyclopedia of Philosophy (New York: Routledge, 1998), R: 290–4.
(4) T. M. Scanlon, What We Owe to Each Other (Cambridge, Mass.: Harvard University Press, 1998), 248.
(5) Stephen Perry, ‘Responsibility for Outcomes, Risk, and the Law of Torts’, in Gerald Postema, ed., Philosophy and the Law of Torts (Cambridge: Cambridge University Press, 2001), 72–130.
(6) Ronald Dworkin, ‘Hard Cases’, in Taking Rights Seriously (Cambridge, Mass.: Harvard University Press, 1975), 88. See also Thomas Nagel, ‘Ruthlessness in Public Life’, in Mortal Questions (Cambridge: Cambridge University Press, 1979), 75–90.
(8) This discussion is obviously greatly simplified and abbreviated. See John Martin Fischer, ‘Introduction’ to Moral Responsibility, ed. John Martin Fischer (Ithaca: Cornell University Press, 1986); Fischer and Mark Ravizza, Responsibility and Control (Cambridge: Cambridge University Press, 1998); Fischer, ‘Recent Work on Moral Responsibility’, Ethics, 110 (1999), 93–139; Bernard Williams, ‘How Free does the Will Need to Be?’, in Making Sense of Humanity (Cambridge: Cambridge University Press, 1995), 3–21; Gary Watson, ‘Free Action and Free Will’, Mind, 96 (1987), 145–72. Fischer and Watson include extensive bibliographies.
(9) For some especially valuable work, see Peter van Inwagen, An Essay on Free Will (New York: Oxford University Press; the work by Fischer mentioned in n. 8; Hilary Bok, Freedom and Responsibility (Princeton: Princeton University Press, 1999); and Susan Wolf, Freedom within Reason (New York: Oxford University Press, 1990).
(10) See R. Jay Wallace, Responsibility and the Moral Sentiments (Cambridge, Mass.: Harvard University Press, 1994), for a deep exploration of Strawson's argument.
(11) Peter Strawson, ‘Freedom and Resentment’, reprinted in Free Will, ed. Gary Watson (New York: Oxford University Press, 1982), 59–80, 62.
(12) Strawson, ‘Resentment’, 66.
(13) Wallace, Moral Sentiments, 76.
(14) See P. S. Greenspan, Practical Guilt: Moral Dilemmas, Emotions and Social Norms (New York: Oxford University Press, 1995); Michael S. Moore, Placing Blame (New York: Oxford University Press, 1997) (esp. ‘The Moral Worth of Retribution’); Herbert Morris, On Guilt and Innocence (Los Angeles: University of California Press, 1976); Jeffrie G. Murphy and Jean Hampton, Forgiveness and Mercy (Cambridge: Cambridge University Press, 1988); Samuel Scheffler, Human Morality (New York: Oxford University Press, 1993); Gabriele Taylor, Pride, Shame and Guilt (New York: Oxford University Press, 1985); Bernard Williams, Shame and Necessity (Berkeley: University of California Press, 1993); and Richard Wollheim, The Sheep and the Ceremony (Cambridge: Cambridge University Press, 1979).
(15) Strawson, ‘Resentment’, 71.
(16) Strawson, ‘Resentment’, 71.
(18) David Hume makes such a claim in his Enquiry, though he grounds the disposition in a notion of self-interest generously expanded by our capacities of sympathetic identification. David Hume, An Enquiry Concerning the Principles of Morals, ed. J. B. Schneewind (Indianapolis: Hackett Publishing, 1983; 1st pub. 1777), § 5, pt. I.
(19) See Bernard Williams' discussion of the Oedipus example, in Bernard Williams, Shame and Necessity (Berkeley: University of California Press, 1993), 56–60.
(20) Bernard Williams, ‘Moral Luck’, in Moral Luck (Cambridge: Cambridge University Press, 1981), 20–39.
(21) As Williams notes, one of the roles played by other moral agents may be to insist upon the rightness of conduct in order to erase the significance of an agent's connection to the harm. Williams, ‘Moral Luck’, 28.
(22) Taylor, Pride, 91.
(23) Of course, the driver him- or herself also now stands in a special normative relation to my friend, the cat's owner, owing her at least the courtesy of informing her that the cat is dead.
(24) This is not to say that I would be morally wrong not to feel at all guilty, but only lacking a full grasp of the relevant moral norms. And self-laceration is clearly out of place.
(25) H. L. A. Hart and Tony Honoré, Causation in the Law, 2nd edn. (New York: Oxford University Press, 1985), p. lxxx. See also Honoré's discussion of ‘outcome-responsibility’ and its relationship to identity in Tony Honoré, ‘The Morality of Tort Law—Questions and Answers’, in Philosophical Foundations of Tort Law, ed. David G. Owen (New York: Oxford University Press, 1995), 73–95, 81–3.
(26) This is consistent with a certain understanding of Donald Davidson's claim that ‘We never do more than move our bodies; the rest is up to nature’. Donald Davidson, ‘Agency’, in Essays on Actions and Events (New York: Oxford University Press, 1981), 43–62, 59. The ‘accordion effect’ that licences further ascriptions of events to my agency relies on causal relations external to me, but the particular relations singled out are, as Davidson would acknowledge, deeply dependent upon our normative concerns. See also Joel Feinberg, ‘Action and Responsibility’, in Doing and Deserving (Princeton: Princeton University Press, 1970), 119–51.
(27) Hart and Honoré's Causation in the Law is still the leading work on the morally relevant notion of causation. But see also Joel Feinberg, Harm to Others (New York: Oxford University Press, 1984); and Richard Wright, ‘Causation in Tort Law’, California Law Review, 73 (1985), 1775–98. For other important accounts of a generic conception of causation, see also J. L. Mackie, The Cement of the Universe (New York: Oxford University Press, 1974); and David Lewis, ‘Causation & Postscript’, in Philosophical Papers (New York: Oxford University Press, 1984), 159–213. Michael Moore began a project of demonstrating that the ordinary language of degrees of causation and intervening causation reflects the actual metaphysics of causation. See Moore, ‘Causation and Responsibility’, Social Philosophy & Policy, 16 (1999), 1–51; and cf. ‘The Metaphysics of Causal Intervention’, California Law Review, 88 (2000), 827–77, 876–7.
(28) Lewis, ‘Causation’.
(29) That I will see it this way says much about my view of what constitutes appropriate relations among competitors.
(30) Williams makes a similar point, in ‘Moral Luck’, 28–9. I do not mean to claim that agents usually will feel this way, except in the most tragic of circumstances.
(31) This charge of circularity is the standard criticism of tort lawyers' use of the notion of ‘proximate cause’: to say of a party's conduct that it was the proximate cause of the harm is virtually to foreclose the question of liability. Much of Causation in the Law can be seen as an attempt to give independent content to the notion of proximate cause. Section 4 takes up the issue of causation in tort law.
(32) The ‘control principle’ is endorsed, in varying forms, by Douglas Husak, Philosophy of Criminal Law (Totowa, NJ: Rowman & Littlefield, 1987), 98; and by Stephen Perry, ‘Responsibility for Outcomes, 82.
(33) I treat this subject at length in Christopher Kutz, Complicity: Ethics and Law for a Collective Age (Cambridge: Cambridge University Press, 2000).
(34) More difficult questions arise concerning those who control their initial participation, but not its ultimate extent—those who get taken on a ride, so to speak. Presumably they are accountable by reference to the risk they run in participating in a potentially injurious enterprise.
(35) Derek Parfit, Reasons and Persons (New York: Oxford University Press, 1984), 80. If the example is too surreal, consider the huge bombing fleet that burned Dresden in World War II. No one plane made a difference, but each contributed to a collective horror. I discuss the Dresden case extensively in Complicity, ch. 4.
(36) See Alvin I. Goldman, ‘Why Citizens Should Vote: A Causal Responsibility Approach’, Social Philosophy and Policy, 16 (1999), 201–17.
(37) Tort and contract law have similar functions, also stabilizing co-operation. Richard Posner's economic analysis of criminal law, according to which the purpose of criminal norms is to prevent intentional bypass of market transactions, can be expressed less crassly in terms of making consensual, co-operative relations possible.
(38) See, famously, United States v. Dotterweich, 320 US 277 (1943) (president of pharmaceutical company convicted of interstate shipping of misbranded and adulterated drugs, despite lack of knowledge or intent).
(39) See Moore, ‘Moral Worth of Retribution’; Morris, ‘Persons and Punishment’, in On Guilt and Innocence (Berkeley: University of California Press, 1976); Joel Feinberg, ‘The Expressive Theory of Punishment’, in Doing and Deserving (Princeton: Princeton University Press, 1970).
(40) See e.g. Barbara Wootton, Social Science and Social Pathology (London: Allen & Unwin, 1959); Jean Hampton, ‘The Moral Education Theory of Punishment’, Philosophy and Public Affairs, 13 (1984), 208–38; Geoffrey Sayre-McCord, ‘Criminal Justice and Legal Reparations as an Alternative to Punishment’, forthcoming in Philosophical Topics, ed. Ernest Sosa and Enrique Villanueva (New York: Blackwells); Steven Garvey, ‘Punishment as Atonement’, University of California Law Review, 46 (1999), 1801–58.
(41) See H. L. A. Hart, ‘Legal Responsibility and the Excuses’, in Punishment and Responsibility, 28–53, J. L. Mackie, ‘The Grounds of Responsibility’, in Law, Morality, and Society, ed. P. M. S. Hacker and Joseph Raz (Oxford: Oxford University Press, 1977); T. M. Scanlon, ‘The Significance of Choice’, reprinted in Equal Freedom (Ann Arbor: University of Michigan Press, 1995), 39–104. See also Alan Brudner, ‘Agency and Welfare in the Penal Law’, in Stephen Shute, John Gardner, and Jeremy Horder (eds.), Action and Value in Criminal Law (New York: Oxford University Press, 1993), 21–53.
(42) Mackie, ‘Grounds of Responsibility’, 187–8.
(43) Hart, ‘Legal Responsibility and the Excuses’; Scanlon, ‘Significance of Choice.’
(44) With the important and largely deplorable exception of Anglo-American conspiracy law, according to which liability can be incurred at a very early stage of planning.
(45) It is worth noting that criminal sentencing is heavily characterological, especially in death penalty proceedings. Punishment is aggravated or mitigated in proportion to the moral worth of the convicted.
(46) Immanuel Kant, ‘Doctrine of Right’, in The Metaphysics of Morals, ed. and trans. Mary Gregor (Cambridge: Cambridge University Press, 1991), 331–3.
(47) Friedrich Nietzsche, On the Genealogy of Morals, ed. and trans. Walter Kaufmann (New York: Vintage, 1967; 1st pub. 1887), book II, § 6 (my translation). Nietzsche's own preferred alternative to punishment, at least in the moral realm, was a kind of confident forgetting exemplified by Mirabeau. Genealogy, book I, ❡ 10, p. 39.
(48) The most famous statement rejecting the rehabilitative ideal is in C. S. Lewis, ‘The Humanitarian Theory of Punishment’, Res Judicatae, 6 (1953), 224–30; see also the essays collected in Sanford Kadish, Blame and Punishment (New York: MacMillan, 1987). Isaiah Berlin's objections to a politics grounded in a notion of ‘positive liberty’ also echo concerns about the manipulative state. See his ‘Two Concepts of Liberty’, in Four Essays on Liberty (New York: Oxford University Press, 1970), 118–72.
(49) Hart, ‘Postscript’, 234.
(50) See Moore, Placing Blame; and Andrew von Hirsch, Doing Justice (New York: Hill and Wang, 1976).
(51) The public shift is easily seen in California's politics, usually a harbinger of national trends, a politics that resulted in such successful ballot initiatives as, in 1994, a ‘Three Strikes’ law that incarcerates a wide range of recidivists for life (codified as Cal. Penal Code § 1170.12); and, in February 2000, Proposition 21, a ballot measure that greatly expanded the reach of prosecutorial authority over the young. It passed overwhelmingly. For a general discussion of changes in criminological practice, see David Garland, Punishment and Modern Society: A Study in Social Theory (Chicago: University of Chicago Press, 1993).
(52) Doubts—possibly mistaken—about the effectiveness of rehabilitative measures probably contributed more to their loss of public support.
(53) This is, roughly, Henry Sidgwick's view, see Sidgwick, Methods of Ethics (Indianapolis: Hackett Publishing, 1981; 1st pub. 1874). It is this view of utilitarianism that John Rawls criticizes in A Theory of Justice (Cambridge, Mass.: Harvard University Press, rev. edn. 1999), § 5.
(54) Although co-operative models of utilitarianism are available; see Donald Regan, Utilitarianism and Co-operation (New York: Oxford University Press, 1980).
(55) This criticism is made by Jeffrie Murphy in ‘Retributivism, Moral Education, and the Liberal State’, Criminal Justice Ethics, 4 (1985), 3–11.
(56) For response, see Moore, ‘Moral Worth of Retribution’, 150–1; for the controversy, see Scanlon, ‘Significance of Choice’; and Samuel Scheffler, ‘Liberalism, Desert, and Reactive Attitudes’, Philosophy and Public Affairs, 21 (1992), 299–323.
(57) Even when the state is apparently the victim, as in tax fraud or treason, we must keep in mind that it has no tax or security interests of its own, but only those of its citizens. This derivative status renders its position, normatively at least, wholly unlike the positions of unmediated victimhood.
(58) For representative statements of this view, see Sanford Kadish, ‘Luck of the Draw’, in Blame and Punishment; J. Feinberg, ‘Equal Punishment for Failed Attempts’, Arizona Law Review, 37 (1995), 117–34; Stephen Schulhofer, ‘Harm and Punishment: A Critique of the Emphasis on the Results of Conduct in the Criminal Law’, University of Pennsylvania Law Review, 122 (1974), 1497.
(59) See Moore, ‘The Independent Moral Significance of Wrongdoing’, in Placing Blame, 191–248; Leo Katz, ‘Why the Successful Assassin is More Wicked than the Unsuccessful One’, California Law Review, 88 (2000), 791–812.
(60) For the deeper philosophical discussion of moral luck, see Thomas Nagel, ‘Moral Luck’, in Mortal Questions (New York: Cambridge University Press, 1979), 24–38; and Williams, ‘Moral Luck’.
(61) See Sidgwick, Methods of Ethics, 280–1.
(62) The American Legal Institute's Model Penal Code is Wechsler's lasting legacy; for Williams see his Criminal Law: The General Part (London: Stevens and Sons, 2nd edn. 1961).
Where strict liability still obtains, in the regulatory domain mentioned above, or in specific doctrines of, for example, publicans' vicarious liability, the penalties are generally comparatively mild, and liability can often be tied, through the exercise of prosecutorial discretion, to some failure of role responsibility.
(63) See Sandoval v People, 117 Colo. 558, 192 P. 2d 423 (1948) (‘it matters not how short the interval between the determination to kill and infliction of the mortal wound, if the time was sufficient for one thought to follow another’).
(64) See e.g. Brinegar v United States, 338 US 160 (1949) (‘[M]uch evidence of real and substantial probative value goes out on considerations irrelevant to its probative weight but relevant to possible misunderstanding or misuse by the jury’). Note that evidence of a defendant's criminal history is often brought in after conviction, at the sentencing stage, as an aggravating factor.
(65) See Immanuel Kant, Groundwork of the Metaphysics of Morals, trans. Mary Gregor (Cambridge: Cambridge University Press, 1998; 1st pub. 1785), ch. 2, 407: ‘We are pleased to flatter ourselves with the false claim to a nobler motive, but in fact we can never, even by the most strenuous self-examination, get to the bottom of our secret impulsions; for when moral value is in question, we are concerned, not with the actions which we see, but with their inner principles, which we cannot see.’ That said, Kant's mechanism of evaluating potential maxims does seem to presuppose a large amount of self-knowledge.
(66) See e.g. Barbara Herman, ‘Moral Deliberation and the Derivation of Duties’, in The Practice of Moral Judgment (Cambridge, Mass.: Harvard University Press, 1993), 132–58; Richard Brandt, A Theory of the Good and the Right (New York: Oxford University Press, 1979); and John Harsanyi, ‘Rule Utilitarianism and Decision Theory’, Erkenntnis, 11 (1977), 25–53.
(67) One can, of course, take on a new identity as a result of imprisonment, just or unjust. My point is that one cannot return naïvely to the normal positions of an equal citizen. One becomes either an antisocial rebel, or a sentimental citizen.
(68) Douglas Husak is an example of a legal theorist who may distinguish too sharply between the two. He argues that questions of the difficulty of proving mens rea do not bear upon the justice of a mens rea requirement. Husak, Philosophy of Criminal Law, 59–60. Although I agree with Husak that intentionality requirements should not easily be compromised for the sake of easing the burden of proof, I do not think the position generalizes to the autonomy of liability rules from epistemic questions.
(69) For a more thorough examination of the relevant self-governing capacity, see Fischer and Ravizza, Responsibility and Control; and Moore, ‘The Legal View of Persons’, in Law and Psychiatry (Cambridge: Cambridge University Press, 1984), 44–112.
(70) Since it demands that one's actions flow from a process of reasoned action, it does seem incompatible with any notion of physical indeterminacy that would make action or decision essentially random.
(71) For discussion see Fischer and Ravizza, Responsibility and Control, 44–51; Stephen J. Morse, ‘Diminished Capacity’, in Shute, Action and Value, 239–78; Scanlon, ‘Significance of Choice’, 61–4.
(72) This is again to emphasize the special responsive position of the state, and to deny, contra Moore, that it should take the role played by God in a godless world.
(73) Another small philosophical literature has emerged concerning the ‘actual’ blameworthiness of corporate actors, where this question is usually approached through the metaphysical question whether corporations are moral persons. See, e.g., Peter French, Collective and Corporate Responsibility (New York: Columbia University Press, 1984); and Larry May, The Morality of Groups (Notre Dame, Indiana: Notre Dame University Press, 1987). The metaphysical discussion seems to me largely to miss the genuinely central question. Given that all groups are groups of people, what are individuals' responsibilities for what their groups do? Examining complicity seems to me the better way to approach the problem, than to inquire in what sense General Motors can be properly blamed. Of course, there are important questions about the effectiveness of different measures directed at corporate groups. For an excellent discussion, see Brent Fisse and John Braithwaite, Corporations, Crime, and Accountability (Cambridge: Cambride University Press, 1993).
(74) Regina v M'Naghten, 8 Eng. Rep 718 (1843).
(75) Model Penal Code § 4.01.
(76) Durham v United States, 214 F.2d 845, 862 (D.C. Cir. 1954).
(77) See Moore, ‘The Legal Concept of Insanity’, 245, for this surely correct observation.
(78) See John Deigh, ‘Empathy and Universalizability’, Ethics, 105 (1995), 743–63; see also Fischer and Ravizza, Responsibility and Control, 76–81.
(79) There would be a problem for instrumentalists if defendants could easily demonstrate insanity, then simulate a quick recovery, thus avoiding both punishment and lengthy incapacitation; the deterrence system would clearly be undermined. For better or worse, this does not seem an actual problem, as confinement of the criminally insane is notoriously lengthy.
(80) See e.g. Hart, ‘Legal Responsibility’.
(81) Both recklessness and negligence raise questions of individualized versus normalized standards, questions I treat shortly.
(82) This description, in fact, only works for the individual-actor paradigm. In cases of complicity, the accomplice has a culpable mens rea, namely an intention to further the principal's criminal objective, but his actions fall well short of the actus reus component of the offence. For a superb discussion, see Sanford H. Kadish, ‘Complicity, Cause, and Blame: A Study in Doctrine’, California Law Review, 73 (1985), 323–410. As with the harmless torturers mentioned above, the justification in liability has to be grounded almost entirely in the accomplice's participatory intention, not in his causal contribution. See my Complicity, 220–36.
(83) See e.g. Michael Moore, Act and Crime (New York: Oxford University Press, 1993); and Antony Duff, Intention, Agency, and Criminal Liability (New York: Oxford University Press, 1990).
(84) Alvin I. Goldman, ‘Action and Crime: A Fine-Grained Approach’, University of Pennsylvania Law Review, 142 (1994), 1563–86; Donald Davidson, ‘Agency’, in Essays on Actions and Events (New York: Oxford University Press, 1980), 43–61.
(85) For discussion of this point, see Ripstein, Equality, Responsibility, and the Law (New York: Cambridge University Press, 1999), 14; and Jennifer Hornsby, ‘On What's Intentionally Done’, in Shute, Action and Value, 55–74.
(86) These facts are loosely taken from R. v Hyam  Q.B. 99 (C. A) (defendant set fire to house hoping to frighten romantic rival, killing the rival's two daughters). Similar issues are raised by crimes that make specific beliefs part of the definition, notably rape, which in some formulations makes the defendant's belief that the victim does not consent, an element of the crime. In the notorious English Morgan case, it was held that a defendant's unreasonable but sincere belief that a woman had consented to intercourse could be a complete defence to the rape charge; now, it is often held that the defendant's belief about consent must be reasonable. R. v Morgan  A.C. 182. (The defendants in question were supposedly persuaded by Morgan that his wife enjoyed intercourse under duress.)
(87) The relation between deterrence and the excuses raises the important distinction between conduct rules, addressed to citizens, and decision rules, addressed to adjudicators. Excuse rules are arguably decision rules alone, and should not play a role in citizens' deliberations. See Meir Dan-Cohen, ‘Conduct Rules and Decision Rules: On Acoustic Separation in Criminal Law’, Harvard Law Review, 97 (1984), 625–77.
(88) See Ripstein, Equality, 163–70. Ripstein's argument is much subtler and more complex than I show here.
(89) Compare Scanlon, ‘Significance of Choice’, 89–96.
(90) Hart, ‘Legal Responsibility’.
(91) As I mentioned above, the practice of criminal law might come to include a reparative or restitutionary elements, through which the state oversees a response from agent to victim. The pure restitutionary model, such as that suggested by Randy Barnett in The Architecture of Freedom (New York: Oxford University Press, 1998), simply replaces criminal with tort law; it ignores the reasons to repair other relations sundered by criminal acts, including relations to the state.
(92) For a recent reassessment of New Zealand's alternative to a tort system, see Bryce Wilkinson, ‘New Zealand's Failed Experiment with State Monopoly Accident Insurance’, Green Bag, 2d 2 (1998), 45–55.
(93) I borrow these terms, along with much else in my discussion, from Jules Coleman; see his ‘Second Thoughts and Other First Impressions’, in Brian Bix (ed.), Analyzing Law (New York: Oxford University Press, 1998), 257–322, 301–6. But borrowing so much from Coleman is merely to keep company with much of the rest of tort theory in the last two decades.
(94) See e.g. Guido Calabresi, The Costs of Accidents (New Haven: Yale University Press, 1970); William N. Landes and Richard A. Posner, The Economic Structure of Tort Law (Cambridge, Mass.: Harvard University Press, 1987); and Steven Shavell, Economic Analysis of Accident Law (Cambridge, Mass.: Harvard University Press, 1987).
(95) See Coleman, ‘Tort Law and the Demands of Corrective Justice’, Indiana Law Journal, 67 (1992), 349–79.
(96) Such an approach is suggested by Ronald Dworkin in Law's Empire (Cambridge, Mass.: Harvard University Press, 1986), 276–312. In his ‘What is Equality? Part 2: Equality of Resources’, Philosophy and Public Affairs, 10 (1981), 283–345, Dworkin lays the ground for an alternative approach, according to which accident costs should allocated to individuals when those costs reflect, in a specified sense, choices of those individuals; and otherwise should be allocated so as to preserve an egalitarian distribution of resources. This approach is further developed in Eric Rakowski, Equal Justice (New York: Oxford University Press, 1991), 227–43.
(97) This approach is suggested by Hart's theory of criminal liability, discussed above.
(98) Richard Epstein's early view, in ‘A Theory of Strict Liability’, Journal of Legal Studies, 2 (1973), 151–204, is still taken as exemplary of libertarian theories of tort. While Epstein still considers himself a libertarian, his more recent work, in Simple Rules for a Complex World (Cambridge, Mass.: Harvard University Press, 1995), seeks a utilitarian foundation for tort law. Whether or not Epstein can reconcile libertarianism with utilitarianism, his newer view is clearly allocative—which may reduce actually held, genuinely libertarian theories of tort to a null set.
(99) As a corollary, if no agent is responsible for the victim's loss, finding compensation in a mandatory insurance pool would be an illegitimate taking.
(100) Ronald Coase, ‘The Problem of Social Cost’, Journal of Law & Economics, 3 (1960). For a rich development of this point, see Stephen Perry, ‘The Impossibility of General Strict Liability’, Canadian Journal of Law & Jurisprudence, 1 (1988), 147–71.
(101) See the comprehensive New York Times series of reports on the disaster, ‘The Bhopal Disaster: How it Happened’, in the 28, 29, and 31 Jan. 1994 issues, all p. A1. The Bhopal plant was operated by Union Carbide India, whose shares were 50.9% owned by Union Carbide Co. (US), 22% directly by the Government of India, and the remainder among 23,000 Indian citizens.
(102) Ernest Weinrib, A Theory of Private Law (Cambridge, Mass.: Harvard University Press, 1995).
(103) For a suggestive development of a Kantian view of tort liability, see Barbara Herman, ‘What Happens to the Consequences?’, in Practice of Moral Judgment, 94–112.
(104) Jules Coleman, Risks and Wrongs (Cambridge: Cambridge University Press, 1992), 314–18.
(105) Coleman, Risks, 324.
(106) Coleman, ‘The Practice of Corrective Justice’, in Owen (ed.), Philosophical Foundations, 63, 53–72. To say that the principle of corrective justice is contingent, however, is not to say that it is merely contingent, in any sense of ‘merely’ beyond the metaphysical. For the principle that agents have special reason to repair their wrongs resonates throughout a set of ideals of agency, concern for others, and personhood. The principle, in other words, depends on a broad set of relational practices. Echoes of Strawson are strong.
(107) Coleman, Risks, 386–406. The state may not do justifiably nothing, at least if it has a defensible ideal of distributive justice.
(108) See Perry, ‘Responsibility for Outcomes’; Perry, ‘The Moral Foundations of Tort Law’, Iowa Law Review, 77 (1992), 449–513; Tony Honoré, ‘Responsibility and Luck’, The Law Quarterly Review (1988), 530–53.
(109) Perry, ‘Moral Foundations’, 498. Perry suggests that outcome-responsibility is grounded only in one's responses to one's own outcomes. But, being the social beings we are, surely our reactions to what others do, and theirs to us, contribute to the phenomenological importance of causation and control.
(110) See Jules Coleman and Arthur Ripstein, ‘Mischief and Misfortune’, McGill Law Journal, 41 (1995), 91–130; see also Ripstein's independent exploration in Ripstein, Equality, Responsibility.
(111) In the version of liberal political theory Coleman and Ripstein defend, the proper allocation is one that ensures an equal allocation of security, thus deploying normalized standards of liability, and reflects a deliberate ranking of the relative value of different activities. Coleman and Ripstein, ‘Mischief’, 126–9.
(112) Since Coleman and Ripstein make attribution subsidiary to allocation, their argument may seem to threaten to collapse corrective justice into distributive justice, as Perry argues. Perry, ‘Mischief’, 154. But if corrective justice is distinguished from distributive justice by its generation of agent-specific obligations, then Coleman and Ripstein have indeed put forth a corrective justice view, albeit one rooted in distributive justice. Alternatively, they could be read as showing that distributive justice must be understood in terms of both agent-general and agent-specific reasons. See Coleman, ‘Second Thoughts’, 312–16.
(113) It would be equally a mistake to rely on a Strawsonian invocation of ‘natural’ patterns of response, the ‘natural’ being clearly a product of the social.
(114) The English case Holmes v Mather  10 Ex. 21 contains the first prominent claim that negligent or wilful misconduct is a necessary element in a legal claim for compensation. Morton Horwitz argues that the move to negligence in American law expressed a deliberate social policy of subsidizing emerging industries. He notes also, however, that jurists focused on fault as a useful tool for determining liability in cases of joint collision, of which there were suddenly many. Morton Horwitz, The Transformation of American Law: 1780–1860 (Cambridge: Mass., Harvard University Press, 1977), 85–99. See also Lawrence M. Friedman, A History of American Law, 2nd edn. (New York: Simon & Schuster, 1985).
(115) See e.g. Justice Roger Traynor's famous concurrence in the products liability case Escola v Coca Cola Bottling Co., 24 Cal. 2d 453, 461, 150 P.2d 436, 440 (1944), in which he defends enterprise liability in terms of insurance and incentive effects.
(116) The social transformation may come to be viewed generally positively, as is arguably the case in the domain of industrial accidents, with the shift from no employer liability without fault, to no-fault workers' compensation.
(117) Coleman and Ripstein are, as I have noted, particularly conscious of this dimension of tort theory.