II. FEATURES OF A MODEL

Let us start by specifying, first, what makes something a model and, second, what makes something a good model. Here are five observations about models.

First, a model in the relevant sense is a person, body, or work that exemplifies certain observable traits, activities, or behaviors. Typically, if something is an exemplar, the behaviors it demonstrates are neither an accident nor a one-off display. Instead, a model offers repeatable, if not routinely performed, patterns of specialized behavior. That said, an accident or one-off display could serve as a model if it truly exemplified in that moment what a standard, repeatable model would exemplify, and did so in a way that followers could internalize. For example, at the 1968 Mexico Olympics, all things aligned for the track-and-field athlete Bob Beamon to set a phenomenal world record in long jump that stood for twenty-two years, was caught on film, and was studied by teams for training purposes.

Second, a model in the relevant sense is a person, body, or work that we can reasonably propose to emulate. That is, a model's behaviors are ones that we can reasonably try to follow if we wish to engage in those behaviors. This reasonableness test is not very demanding. While it would be unreasonable for a runner to try to model her sprinting style after a comet, it could be reasonable for an engineer to try to model her aircraft designs after birds in flight.

Third, as the bird example implies, a model need not be the same kind of thing as its followers. The family is a common and credible model for the state even though families and states are different things.

Fourth, the language of “exemplar” and “emulation” has positive connotations that suggest that a model is a perfect embodiment of something worthy of emulation. But neither perfection nor value is required for modeling. For something to be a model, it must meet a certain standard of competence in that thing but it need not meet an absolute standard of brilliance in that thing and it need not model anything worth emulating. An expert assassin could be a model for a novice assassin even though he models nothing worth emulating. What he must do is model that activity well enough to be a genuine model. If the novice assassin chooses to copy a lousy assassin who never hits his mark, then the novice picks someone who is not actually a model. The novice may model himself after the lousy assassin, but the lousy assassin does not model how to assassinate. If the novice chooses to copy a middling assassin who occasionally succeeds in hitting the mark, then the novice does indeed have a genuine model to follow, just not a very good one.

For something to be a good model of something, it need not meet an absolute standard of brilliance at that thing, but it must meet a standard of competence that is appropriately higher than that of its followers. If a model is either just barely more competent than its followers or too competent relative to its followers, it will fail to be a good model for them. In the latter case, it will fail to be a good model for them because its followers will be unable to engage in the behaviors it demonstrates. If we are beginner chess players, a competent amateur will be sufficiently proficient to model for us how to play and will model it for us better than a grandmaster would do, since the amateur's decisions will be more accessible to us. The grandmaster is an ideal for us to admire and to aspire to emulate but she is not a practicable model for us as beginners. She is, however, a practicable model for the new professional.

This point about relative competence implies that if we improve sufficiently in a field, then we can live up to or even surpass our model, in which case it will no longer be a model for us.Footnote ⁵ This is true even if our model is a different kind of thing from us. For instance, although a state can probably never be a family in the true sense, it could come to embody better than many families do those organizational and attitudinal traits of families that it sought to emulate. This point about relative competence will be helpful in what follows to understand how the law might model virtue for us.

Fifth, as the above points suggest, modeling is conative. It is about observable behaviors and performative knowledge of how to do something. A master chess player who glances at a board and sees the best move has performative knowledge of how to play chess. If we could emulate what the chess player does, then we would play chess well too. This is the emulation account of expertise. The emulation account contrasts with the advice account of expertise, which is about propositional knowledge of what to do. If a chess master could explain to us what she does to arrive at the best moves, then she would be able to advise us on what to do. The best models most likely combine performative knowledge with propositional knowledge.

With these five features of a model in hand, let us look at what is involved in modeling virtue.

III. MODELS OF VIRTUE

Some thinkers will be skeptical about a virtue-ethical approach to understanding our relation to law. This skepticism can come from two opposite directions. One objection is that, unlike deontology and consequentialism, virtue ethics fails to offer a codifiable set of norms by which to live. In focusing on disposition and character, virtue ethics fails to be action-guiding.Footnote ⁶ The second objection, from the other direction, is that virtue ethics is indeed action-guiding—it instructs us to act as the virtuous person would act. But in being action-guiding, it is derivative of or collapses into a form of deontology.Footnote ⁷

There are good responses to both of these objections. Briefly, as virtue ethicists have argued, much action guidance can be found in attending to virtue and vice terms and doing what is honest, charitable, generous, kind, and so on.Footnote ⁸ Moreover, this action guidance does not present morality in terms of strict rules, absolute principles, perfect duty, judgment, blame, and guilt, which shows not only that it cannot be reduced to deontology but also that it has advantages over deontology, since these deontic terms come with heavy prices. They can be alienating and depersonalizing. A virtue-ethical approach avoids these costs by seeing morality more positively, charitably, and creatively in terms of settled perspectives, dispositions, and patterns of exemplary moral goodness that track ethical precepts that are not applied absolutely.

One advantage of this is psychological. Often, we are likely to be more receptive to incentives and good modeling than we are to threats, demands, and absolute dictates.Footnote ⁹ If this is correct, then a second, instrumental advantage follows: general adherence to moral norms may improve if we frame our moral world in terms of virtue rather than duty. A third, related advantage is cognitive. A virtue-ethical approach gives us more scope in which to exercise practical reason than standard deontological approaches do because it asks us not just to use first-order reasoning and discernment but also to internalize fundamental values as part of our character and dispositions, to observe the patterns of cause and effect in behavior, to witness good examples of moral virtue, and to develop settled inclinations to set good examples ourselves.Footnote ¹⁰

Furthermore, if the skeptic is not persuaded by these points, that need not trouble us in this discussion because our purpose here is not to dismiss the traditional duty-oriented debates about our moral relation to the law, but instead to highlight that there is another, equally interesting dimension to our moral relation to the law that lies in the law's ability to model virtue for us in beneficial ways.

The most modest model of virtue is, I take it, a model of isolated virtuous acts. A slightly more comprehensive model is a model of patterns of virtuous action. An even more comprehensive model embodies a particular virtue or set of virtues (as a count noun). A virtuous person, in the mundane, small-v, small-p sense, falls into this category. Finally, the most comprehensive model would be a model of Virtue (as a mass noun). This last model is often characterized in terms of the so-called Virtuous Person, who is a purported ideal of immutable, full, and perfect virtue.Footnote ¹¹ The Virtuous Person is characterized as having a deeply entrenched, multifaceted mind-set and disposition to act and react in the right way at the right time on the basis of the right reasons with the right attitude, intentions, and expectations.Footnote ¹²

This taxonomy of virtue models rejects any strong version of the unity of virtue thesis embraced by Aristotle and Plato, according to which a being must possess all virtues in order to possess any one virtue. This taxonomy also rejects a weaker unity of virtue thesis, such as that embraced by Susan Wolf, which says that in order to possess any one virtue, a being must fully understand the value of that virtue relative to other virtues even if the being does not embody those other virtues.Footnote ¹³ This taxonomy allows that a being can model virtuous conduct without embodying the virtue that aligns with it, and a being can embody one virtue without having complete understanding of the value of that virtue in relation to other virtues.

Now, of the four models of virtue just outlined, the credibility of the most comprehensive model—the Virtuous Person—faces at least four objections.Footnote ¹⁴

The first is the limitlessness-of-virtue objection. The notion of the Virtuous Person assumes that there is something that is “full and perfect virtue.” In other words, it assumes that there is a limit to each and every virtue. But there is no reason to think that all virtues have a built-in limit. Virtues such as kindness or compassion may be limitless. Indeed, in the Buddhist tradition, which gives a rich articulation of loving kindness and compassion, these virtues are said to be boundless. Hence, speaking of such virtues in terms of “full and perfect virtue” would be like speaking of the largest number in the unbounded infinity that is the natural numbers. If that is so, then there cannot be, even in principle, a being that could fully and perfectly embody all Virtue.

In reply, a defender of the Virtuous Person might grant the possibility of limitlessness and say that the Virtuous Person satisfies a certain threshold of Virtue beyond which further cultivation of virtue is really just gravy. This response is concessive and likely to be unappealing to many who would defend the Virtuous Person as a being with a settled disposition to act in the right way at the right moment for the right reason.

A more credible reply might be that the limit of a virtue is given by the limit on the opportunities to demonstrate that virtue in actions. And those opportunities are limited nontrivially by the fact that a given virtue is not called for in every moment of decision. That is to say, the truly Virtuous Person in the Aristotelian sense acts and reacts in the right way on each occasion, and it is not the case that each occasion calls for each and every virtue.Footnote ¹⁵ So, in a case where compassion is not called for, there will be a “limit” on the virtue of compassion.

This reply is orthogonal to the limitlessness-of-virtue objection, as it appeals to a different sense of “limit” from the objection. Whereas the objection challenges the presumption that there is an upper bound to the possible cultivation of a given virtue, this reply simply speaks of a practical check on the appropriateness of deploying a virtue in a given moment.

The second objection to the Virtuous Person is the pluralism objection. The credibility of the Virtuous Person may be doubted on value-pluralism grounds. If one accepts that morality is fundamentally pluralistic, then “full and perfect virtue” is impossible. Embodying one virtue will generally come at the expense of embodying some other virtues.Footnote ¹⁶ Of course, this does not cast doubt on the possibility of there being virtuous persons, in the small-v, small-p sense, who are mutable beings committed to struggling to improve morally.

The third objection is the impossibility-of-aspirations objection. Irrespective of its conceptual intelligibility, the Virtuous Person appears to be unintelligible as a moral model for us because this being necessarily lacks an aspiration to improve morally.Footnote ¹⁷ Moral improvement is not possible for such a being. Yet, having an aspiration to improve morally is part of a morally good life for us as persons and thus is a requisite element of a plausible moral model for us. Moreover, without the possibility of aspirations to improve morally, the Virtuous Person begins to look like a deep impossibility. It is difficult to imagine what form a being would take if it possessed full and perfect virtue and consequently lacked moral aspirations.

In reply, a defender of the Virtuous Person might ask why moral aspirations need to be a requisite feature of a plausible model for those of us seeking to be virtuous. The defender might point out that, in my own view, someone can be a model in one respect and not another. And therefore, even if we ought to have moral aspirations, it does not mean that a being without them cannot be a good model of virtue for us.

This reply is unconvincing because, if we accept that having an aspiration to improve morally is part of a morally good life for persons, then the Virtuous Person is at best an imperfect moral model for us, since she cannot model moral aspirations. This is not a problem as far as her status as a moral model goes because, as just noted, something can be a model in one respect and not in another. But the imperfection of the Virtuous Person as a moral model is a problem for the idea that the Virtuous Person is a being of full and perfect virtue. If part of being a Virtuous Person is being a perfect model of virtue as well as a perfect embodiment of virtue, then the Virtuous Person must be able to model moral aspirations. This result actually shows that the Virtuous Person is paradoxical. Either she can have and model moral aspirations, in which case she is not perfectly virtuous, or she is perfectly virtuous, in which case she is not a perfect model of virtue for us.

The fourth objection is the inaccessibility objection. Even if the Virtuous Person were not a deep impossibility and, indeed, could have moral aspirations, nevertheless she would still not be a practicable moral model for us in the sense of model specified in Section II. The reason for this is that her behaviors would not be ones that we could emulate. To us beginners, she would be the chess grandmaster whose reasoning and decisions we cannot understand. Therefore, a much more modest display of virtue is needed for something to be a good model of virtue for us.

There is a notable payoff in this conclusion. It puts to rest one possible objection to the idea that the law could model virtue for us, which is the objection that by nature, the law falls far short of true Virtue. As shown above, this objection is a nonstarter, since true Virtue is suspect on several grounds. There is, however, a weaker version of this objection, which we might call the embodiment problem.

IV. THE EMBODIMENT PROBLEM

This problem says that the law cannot model virtues because virtues have affective and cognitive dimensions, not just conative dimensions, and the law has no affective and cognitive dimensions.

In reply, first, as John Gardner notes, the capacity for moral agency is a capacity not only of adult human beings but also of the institutions that they create and inhabit.Footnote ¹⁸ The law is a human product and a human institution and hence, through its participants, has a capacity for moral agency. It has the capacity to operate in accordance with moral reasons. This reply takes care of the cognitive dimension of virtue and (depending on one's metaethical views on motivation) the affective dimension, too.

Second, even if no affective states or character dispositions can be attributed to law, law has an analogue in its history and legal tradition as well as in the wider culture, commitments, and values of the society. Its animating purposes and attitudes are evident in its substance and history and in the mythology that the society has built around it.

Third, intuitively, at least some virtues such as justice and civility are virtues of good law. Indeed, justice and law go hand and hand. Morally speaking, justice is what the law is supposed to achieve. Therefore, in principle, the law can model some virtues, if not all.

Fourth, models are conative. They can only model observable things such as individual actions and patterns of actions. They cannot model intentions, emotions, feelings, or characters. As far as models go, the nonobservable elements of virtues are irrelevant. So what the law could offer as a model would be the same, in principle, as what a virtuous person could offer, which is a model of conduct.

V. THE LAW’S VIRTUES

The law is not just a human product but a product of a state and, more specifically, a government and its legislators, executives, and judicial officials. When we are considering the law's credentials as a model of virtue, we can look at specific laws and policies as well as the constitution, the system of government, and the network of institutions that comprise that regime.

First, the most straightforward and likely way for the law to model virtue is through specific laws and policies. Here are a few examples:

1. 1) The law can model generosity to strangers in need by enacting a substantial foreign support budget, making generous contributions to global public funds, and having immigration policies that are as welcoming as possible to asylum seekers.

2. 2) The law can model benevolence by having inclusive visitor and migrant worker programs.

3. 3) The law can model kindness and compassion by helping other communities when they face disaster, and doing so without strings attached.

4. 4) The law can model forgiveness and charity by adopting a criminal justice system that emphasizes mediation and restoration and is merciful to people who offend against the law.Footnote ¹⁹

5. 5) The law can model decency through policies that are supportive of the people who are most vulnerable, such as policies of unemployment insurance, basic income, universal preventive health care, education and reeducation opportunities, community housing, effective systems for redress and compensation, contract laws that favor vulnerable parties, unconscionability doctrines, and constitutional protections of human rights.

6. 6) The law can model respect by being accommodating of people who are politically engaged even if they dissent.Footnote ²⁰

7. 7) Finally, the law can model open-mindedness by allowing people to pursue different ways of living and by protecting them in those pursuits.

Second, more generally, the law models virtue in showing that disputes should be resolved through discussion, not force. A key thing that the law models (in the Anglo-American tradition) is the giving of reasons.Footnote ²¹

Third, more comprehensively, the law models virtue by practicing what it preaches when what it preaches is morally defensible. As I argue elsewhere, the law can set a good moral example by embodying itself the moral norms that it has enacted. For instance, a legal system that instructs its members not to kill can set a good moral example of this by outlawing capital punishment and by refusing to sanction unjust wars. Similarly, a legal system that instructs its members against invidious discrimination based on sex, age, ethnicity, disability, nationality, or sexual orientation can set a good moral example by ensuring that its own institutions are nondiscriminatory.Footnote ²²

Fourth, even more comprehensively, morally sound laws and policies can become entrenched in a society's legal tradition so that the society celebrates them as a valued part of their cultural identity. And, at that point, they not only model patterns of virtuous action, but embody those virtues as entrenched dispositions of the society and legal culture.Footnote ²³

Fifth, equally comprehensively, the various foundational structures and institutions of a legal system can model virtue:Footnote ²⁴

1. 1) A society's constitution is one indicator of its potential to embody virtues. That potential is high if the constitution reflects a commitment to justice, truth, morality, integrity, decency, peace,Footnote ²⁵ respect, honor, and freedom, which provide a foundation on which the society can embrace the virtues it espouses.

2. 2) A society's system of government is a second indicator of its potential to embody virtues. That potential is high if the structures allow for ongoing development and refinement, are responsive to changes in circumstances, and foster a culture of decency, honesty, and dependability amongst its officeholders. Term limits are one way to check the pressures that elected officials can face. Strict limits on financial contributions to political campaigns are another.

3. 3) A society's interlocking web of official positions is a third indicator of its potential to embody virtues. This potential is high when legislators, executives, and judicial officials as well as teachers, doctors, nurses, police, prison officials, and soldiers view themselves as public servants and hold themselves accountable to the interests of the community.

   1. a. The kinds of virtues that particular institutions can embody depend on how they are structured. If courts are well-structured, they could model the moral trait of listening carefully to both sides, showing all persons equality before the law, attending to the relevant facts impartially, and reserving judgment until all appropriate information has been presented.Footnote ²⁶

Finally, there may be core virtues that the law can make it possible for us to realize, even if the law does not model those virtues explicitly. One example might be sociability.

While I trust that these various examples are intuitively appealing, there is a potential problem that I wish to note but leave largely unaddressed, partly because it is a problem that confronts any effort to apply moral theorizing to concrete cases. Briefly, a critic might argue that, in the abstract, talking about the virtues of law is fine, but in practice there is the problem of specifying how those virtues are to be realized through policy. For example, what kind of policy actually models the virtue of respect for life? Would a staunch, uncompromising antiabortion policy be a good model of respect for life? When I suggest above that the law could honor its own command to its people not to murder by itself not murdering, should it extend that to not permitting abortion? Briefly, in reply, although I embrace moral pluralism and grant therefore that different, competing policies can each in their way honor a given virtue, I do not embrace the political liberal view that on some issues there can be no one answer that is better than others about which policy to adopt. Some policies come with very high costs—an uncompromising, exceptionless antiabortion policy would be one such policy—that detract from or color the moral modeling they might otherwise offer.

Now, there are some steps missing between showing that the law can demonstrate virtue, on the one hand, and showing that we can recognize it as a model of virtue to emulate, on the other. This discussion so far has taken the first step of showing that the law can demonstrate virtue. But this step is not particularly difficult. Good law is supposed to display the moral virtues. The difficult steps are to show, first, that other, less salient aspects of the law do not poison whatever virtue the law is demonstrating, and second, that the law's demonstration of virtue is something that we can recognize as a model of virtue and can emulate. Let me take them each in turn by addressing first the poisoning problem and then the emulation problem.

VI. THE POISONING PROBLEM

There are several dimensions to the poisoning problem. One dimension arises from what might seem to be an asset of law in the effort to make people act virtuously. Aristotle's view of the law is that it is there for people who cannot otherwise be persuaded to act virtuously. The law has a coercive power that it can use to increase virtuous conduct. The poisoning problem here is that this coercive power of the law can model things for people other than virtue. It can model a dehumanizing attitude toward offenders as potential objects for use. And it can model a general willingness to harm people in order to deter them from stepping out of line.

A second, more general dimension of the poisoning problem is that although the law may model generosity in one policy and decency in another, it is still in the nature of law to require morally problematic conduct from its members as a matter of course even in a reasonably good society. The law necessarily calls on its members, as officials and citizens, to do things such as threaten people, attack people, make laws that harm people, lie to people, detain people, isolate people, charge people with offences, make judgments on people's guilt, sentence people to be punished, impose punishments on people, deprive people of their resources, and perhaps, in extreme cases, incarcerate and possibly kill people. Carrying out (many of) these tasks is unavoidable even in a reasonably good society. The best that the law can do is minimize as much as possible the moral burdens it places on its members in doing its bidding.Footnote ²⁷

These realities about law are sharpened by the fact that law can be seen as a distrustful enterprise whereby we rope each other in with the law because we do not trust each other to behave well otherwise. This image casts a shadow over the law's credentials to set a moral example. Just as a profit-driven enterprise makes a poor model for generosity even if it gives generously to charities, so, too, a distrust-driven enterprise like law makes a poor model for trust, openness, and decency even if the law is applied honestly and openly. The nature and purpose of the enterprise of law colors any moral example it sets.Footnote ²⁸

Other features worth noting in this context are the following. First, legal systems have difficulty admitting past mistakes. So whatever virtue goes with being willing to admit past mistakes is one that legal systems probably will not embody. Second, the law typically emerges from highly unvirtuous processes. Despite the ideal scenario of the just, democratic legislative process, in practice, laws are like sausages: you do not want to know how they are made.Footnote ²⁹ Third, the law is a detached, depersonalized institution that has difficulty being context-sensitive. Fourth, the law is the paradigm of a rights-based framework, which complicates any effort to view its operations as a model of virtue. Although the law can display virtuous attitudes towards people, the ways that it implements these attitudes in practice are through the creation of rights and duties. If we model our moral practices on the law, then we may well end up respecting people's rights and demanding that they fulfill their duties instead of adopting the more sensitive, flexible approaches recommended by virtue ethics.Footnote ³⁰ Put starkly, good law picks out what we can properly demand of each other: it is about a moral minimum.

Such features show just how difficult it can be for us to see models of virtue in the law. This difficulty is heightened by the fact that we have a love-hate relationship with the law. It can protect us but it can also strike out at us, and it has a reputation for being an antagonist. Why would we want to model ourselves after something so hot and cold, clunky, and inflexible?

These are hard charges to answer. One reply is that the distrust model of law is not the only one on offer. The law may be seen more optimistically as a coordinating enterprise rather than a distrustful enterprise. In this view, law would exist even if we were all well intentioned and trusting.Footnote ³¹ Or the law may be seen as evidence of our wish to respect each other reciprocally as equal moral agents and equal members of a community.Footnote ³²

Second, to model virtue for us, the law need not meet an absolute standard of brilliance in virtue. It need only be morally competent relative to us, its followers, in those areas in which it is modeling virtue. So even if the law were inescapably a poor model of some virtues, it could be a pretty good model of others, such as fairness, tolerance, justice, mercy, and generosity.

One worry with this response is that it is a bit like saying that the Nazi prison guard who is loving to his children can be a good model of a loving parent despite his profession, and that we can look to him as a model for how to act virtuously as parents even though we cannot look to him as a moral model for anything else.

One answer to this worry is the optimistic but not entirely implausible empirical claim that such a person just could not be a model of a loving parent because his profession and his attitudes would necessarily bleed into his family life through his manner, his moods, and his teachings to his children. And by analogy, a legal system that is brutal, cruel, or vicious in some domain just could not model virtue well in any other domain because the viciousness would bleed into all of its institutions and processes.

A second answer bites the bullet and says that, yes, the Nazi parent or the brutal regime could model virtue in a particular domain provided that the domain could be observed by followers independently of other domains. Presumably, a stranger who comes to the Nazi prison guard's house for dinner and has no idea about his profession could look to his loving care for his children and genuinely respond to it as a good model of loving parental behavior.

This second answer leads to a third, which is that such a person or such a legal system would not be a good model for the people who would actually be well placed to emulate it, such as friends in the case of the guard and citizens in the case of the legal system. Unlike the stranger coming to dinner, friends or citizens have knowledge of the model outside the domain in which he or it is virtuous, and that knowledge colors their perception, understanding, and response to the conduct offered for emulation.

Therefore, although the law need not satisfy a standard of excellence to model virtue for us, it does need to satisfy a general standard of decency and reasonableness in its operations so that its modeling of individual virtues is not tarnished when viewed in combination with its activities as a whole. This is a potential test for a legitimate state: that it satisfies a test of general decency that gives credibility to the moral modeling its legal system offers to its people.Footnote ³³ When it does meet that test, then it can be a very useful model for us. When we observe this sufficiently decent legal system being merciful to offenders, we may learn to forgive more quickly in our own day-to-day interactions with friends and family. When we observe it being tolerant of persons’ individual choices, we may become more accepting of different people's ways of living. And when we observe it being generous to strangers and compassionate to its own vulnerable members, we may learn something about the nature of generosity and compassion. We may see ways that we can incorporate similar kinds of goodness in our own actions. We may also come to make more comprehensive changes to our own outlook; for instance, we may learn to view strangers not as threats but as vulnerable people seeking acceptance.Footnote ³⁴

Now, even once we recognize that the law can offer a good model of virtue for us, a problem remains. This is the emulation problem.

VII. THE EMULATION PROBLEM

Even though the law can model moral virtue for us in a way that we can recognize and emulate, we will not act virtuously if we emulate that model. In brief, we face two horns of a dilemma. The first horn is that if the law's model of virtue is operative in our reasoning and motivation when we act, then we are motivated by the wrong reason and hence do not act virtuously. When we are motivated to follow the law's model, we are motivated by the reason that the law has modeled it and not by the right reasons that would make the act virtuous. But here is the second horn. If the law's model of virtue does not actually figure in our reasoning and motivation when we engage in a virtuous act, then the law is not actually serving as a model for us.

This dilemma is related to what we might call the problem of calculative elusiveness.Footnote ³⁵ Virtue, like spontaneity and many other moral qualities, is calculatively elusive because it cannot be deliberatively internalized or calculated over at a first-order level without self-defeat. As Philip Pettit notes with the example of spontaneity, to be spontaneous is to be uncalculating, and no calculative pyrotechnics can allow spontaneity to be deliberatively internalized.Footnote ³⁶ Similarly, we cannot deliberately aim to act virtuously, since to do so would undermine the virtuousness of the act. In Jon Elster's phrase, virtue is an essential by-product, not something that can be pursued in our individual choices.Footnote ³⁷

The above dilemma shows that virtue also cannot be calculated over at a second-order level without self-defeat. We cannot aim to do as the law models in order thereby to act virtuously because to do so would undermine the virtuousness of the act. Virtue cannot be indirectly pursued in our individual choices. This means that, even though the law can model virtue for us, we will not act virtuously if we emulate the law in order to act virtuously.

The emulation problem in itself is not a distinct problem for law. It is a problem for any relationship between a virtue model and its follower. The answer that is standardly given in virtue ethics is that over time, after modeling herself after the virtue model enough times, the follower becomes able to act virtuously in her own right and for the virtuous reason and not for the reason that the model has set an example that she emulates. The master models by rule of thumb, which over time the disciple learns to approximate more and more precisely.

This standard reply is not entirely satisfactory when the model of virtue is the law because, on one common view at least, the law presents itself as a duty-generator (even if it does not present itself only as that). For those people who view the law as giving them a moral duty to do as it says because it says so, there will always be the problem that when they follow the law for that reason, they act for the wrong reason in terms of virtue. They act for the reason that the law orders them and that they believe they have a duty to do as the law orders because it orders it and not for the reasons that would make their act virtuous.Footnote ³⁸ Moreover, it might seem to be in the interest of the law as a model that many, if not most, people do respond to it in this way because the law can model virtue for us only when enough of us follow the law enough of the time. It is only when there is sufficient adherence (and possibly sufficient unreflective adherence) to the law that the law can continue to manifest patterns of behavior that can give us models of virtuous conduct. And even if we reject the idea that law must be seen principally as a duty-generator in order for it to garner enough general support to function as a model of virtue, nevertheless it remains the case that during our period of learning, our attention is oriented toward the law and not toward the right reason.

But of course, here, too, the law is not the only virtue model that confronts this problem. Effective parents of young children tend to present themselves and tend to be seen by their children as duty-generators among other things, and for many years their children act for the reason that their parents have ordered it. Yet the usefulness of this (initially) unreflective adherence to parental dictates does not prevent those children from internalizing the behaviors over time with an appreciation for the genuinely good reasons behind them. In short, while we are learning from our model, we may disregard the significance of our being motivated by the right reasons when we engage in virtuous acts. In the case of the law, therefore, it is enough, for the kind of model that law can be, that it brings about increased virtuous conduct on our part even if that conduct is not motivated by reasons that would make it truly virtuous.

If these answers are satisfactory, then we may conclude that the law can model virtue for us in beneficial ways that enable us to cultivate morality. This conclusion is surprising, given that, unlike most of the other social enterprises that can model virtue for us, such as stories, movies, plays, clubs, teams, and religious practices, the law is deeply morally complicated. Moreover, this conclusion is noteworthy and indeed comforting, since the law has the potential to shape much of our moral thinking.