Conceived of as a contender to other theories in substantive ethics, virtue ethics is often associated with, in essence, the following account or criterion of right action:

VR: An action A is right for S in circumstances C if and only if a fully virtuous agent would characteristically do A in C.Footnote 1

There are serious objections to VR. These take the form of counter-examples in which we are confronted with cases where less than fully virtuous persons would be acting rightly in doing something that no fully virtuous agent would characteristically do in the circumstances. In section 2 of this paper, I present four such counter-examples to VR. I then proceed to consider various proposals for how to revise VR in order to avoid these counter-examples (sections 3 to 6). I argue that in so far as the revised accounts really manage to steer clear of all the counter-examples to VR, something which it turns out is not quite true for all of them, they instead fall prey to other damaging objections. I end by discussing the future of virtue ethics, given what has come to light in the previous sections of the paper (section 7). In particular, I sketch the outlines of a virtue ethical account of rightness that is structurally different from VR. This account also confronts important problems. Still, I suggest that further scrutiny is required before we are in a position to make a definitive decision about its fate.

First of all, however, I will provide some background to the discussion.

1  

Much work in contemporary virtue ethics relies on a basically Aristotelian account of virtue. Ethical virtues, on this account, pertain to human character. They are constituted by well entrenched dispositions or traits to deliberate and make decisions in accordance with correct reason, as well as to reliably act in appropriate ways and to be properly affected in desire and feeling. In order to get all of these things right, something which is the distinguishing mark of a fully virtuous agent (an Aristotelian phronimos), it is essential that one is equipped with practical intelligence or wisdom (phronesis).Footnote 2 Practical wisdom provides its possessor with an ability to see things as they really are and to appreciate the salient features determining what would constitute the appropriate thing to do or feel in particular situations. This is important not only in order to ensure reliability in good or virtuous behaviour. Rather, the exercise of practical wisdom is itself constitutive of a fully virtuous response in a given situation. For a person to manifest full virtue in his behaviour, it is not enough that the outcomes of his actions are of a certain kind. According to Aristotle, certain conditions about the agent doing the actions must also be fulfilled: “first, if he does them knowingly, secondly if he decides to do them, and decides to them for themselves, and thirdly if he does them from a firm and unchanging disposition” (Aristotle 2002, 1105a31-b1). In addition, the person should characteristically enjoy or take pleasure in his virtuous behaviour rather than, say, having to overcome an inclination to behave otherwise.Footnote 3

Full virtue is clearly meant to constitute an ideal. It is an ideal that we are urged to aspire towards in order to achieve flourishing human lives (eudaimonia).Footnote 4 But Aristotelian virtue ethics would surely be elitist to the extreme if it entailed the claim that in order for anyone to act rightly, all of the above mentioned conditions for fully virtuous behaviour need to be fulfilled. Given the present state of our characters, most of us are simply incapable of embarking upon an ideally virtuous life here and now. We might need to improve with regard to our practical wisdom in order to ensure that we reliably know what would be the virtuous thing to do. Or perhaps we need to work on our desires in order to more consistently take pleasure in our virtuous actions. Neither of these things, it seems, can be rectified immediately at will, but require sustained aspiration to improve.

Contemporary defenders of virtue ethics often respond to this particular problem by drawing a distinction between doing what a fully virtuous agent would characteristically do in the circumstances, and doing it as a fully virtuous agent would characteristically do it, where only the latter requires that all of Aristotle’s conditions for fully virtuous behaviour are fulfilled.Footnote 5 Armed with this distinction virtue ethicists can go on to say that right action should be restricted to the former, which is something that we all, allegedly, can comply with in each particular situation. And then they seem indeed to have arrived at something like VR, that is, the claim that an action A is right for S in circumstances C if and only if a fully virtuous agent would characteristically do A in C. There are a few passages in the Nicomachean Ethics that might be taken to indicate that this corresponds to Aristotle’s own view of the matter. For example, Aristotle says that we become virtuous by doing what a virtuous agent would do (or by doing virtuous actions).Footnote 6 He also argues that there is no independent account or criterion by which we can determine what would constitute virtuous actions in different circumstances; virtuous actions are rather said to be “determined by rational prescription and in the way in which the wise person would determine it” (Aristotle 2002, 1107a1-2). I think it is a fair assumption that these passages are partly responsible for why so many virtue ethicists have subscribed to VR. But even more important, of course, is the fact that VR carries initial plausibility in its own right. It seems to be an important part of our ordinary conception of a virtuous agent that she is someone who characteristically performs right actions.

Before moving on to show why VR should nevertheless be rejected, a few points of clarification about VR are in place.Footnote 7 First, understood merely as an equivalence claim, according to which the class of right actions is the same as the class of actions that a fully virtuous agent would characteristically do, VR could conceivably be accepted by consequentialists and deontologists as well as by virtue ethicists.Footnote 8 On the virtue ethics picture, however, VR is meant to constitute an account of right action that distinguishes virtue ethics from other ethical theories. It is assumed that there is no prior account of rightness that could be used in order to flesh out the notion of doing what a virtuous agent would characteristically do. Instead we should begin by investigating what it is to be a virtuous agent, and then, in the light of that investigation, flesh out the notion of right action in terms of the actions that are characteristic of such an agent. Many critics of virtue ethics have objected that an attractive specification of a virtuous agent will inevitably contain some reference to an antecedent notion of rightness.Footnote 9 Thus, the objection continues, it is not possible for virtue ethics to provide the kind of specification of what it is to be a virtuous agent that is needed in order to turn VR into a distinctively virtue ethical account of right action. For the purposes of this paper, however, we need not get bogged down in this debate. Most of my arguments below retain their force irrespective of how this debate is ultimately resolved.

The second point is that VR, at least as I understand it, is intended primarily as an account or criterion of right action, and not as a decision-procedure. Defenders of VR might very well grant that in everyday life there often are other, more effective ways of finding out how to act rightly, than actually counselling VR itself.

Third, VR says that an action is right if (and only if) it is what a fully virtuous agent would characteristically do in the circumstances. This is because even a fully virtuous agent could conceivably act out of character, and if she does, then her action may well be wrong.Footnote 10

Fourth, VR states that an action is right if (and only if) it is what a fully virtuous agent would characteristically do in the circumstances. “This”, as Robert Johnson puts it, “avoids complications arising from the possibility that in the absence of the other virtues, a courageous person, say, might be led by his courage to act unjustly; a just person may be led by justice to act unkindly; and so on” (Johnson 2003, p. 812). A fully virtuous agent is someone who is in possession of all the virtues, and right actions, according to the claim that we are considering here, are those actions that such an agent would characteristically do.Footnote 11

2  

2.1  

Those of us who do not already lead a life of perfect virtue will occasionally find ourselves in circumstances that a fully virtuous agent would never be in.Footnote 12 For example, suppose that Jones has hurt Smith’s feelings, and that Jones has done so in a way that no one leading a fully virtuous life would have done. Intuitively, we may find it quite clear that in these circumstances it would be right for Jones to (at the very least) apologize to Smith, but this is not something that can be accounted for in terms of what a fully virtuous agent would do, simply because no fully virtuous agent would have done what Jones did to Smith in the first place.

This may not seem a very serious objection to VR. Even a fully virtuous agent could sometimes act out of character. As we mentioned earlier, VR is couched in terms of what a fully virtuous agent would characteristically do in the circumstances precisely because we wanted to avoid the implication that just anything that a fully virtuous agent would do, even if it goes against her overall virtuous character, will have to be assessed as right. Hence, in response to the first counter-example it might be said that while a fully virtuous agent (at least under normal circumstances) would not characteristically hurt another person’s feelings, this may still be something that she could do.

However, even if it is conceivable that a fully virtuous agent occasionally acts out of character, there are limits on what she could do if we are not to withdraw our judgment that she is in fact fully virtuous. Hurting someone’s feelings in the heat of an argument, say, may be among the things that a virtuous agent could do—even though she characteristically would not—without having her virtue put in question. On the other hand, following Rosalind Hursthouse we can imagine a man who has induced two women, A and B, “to bear a child of his by promising marriage”, but who “can only marry one” (Hursthouse 1999, p. 46). This man seems to have got himself into circumstances that no fully virtuous agent could ever be in. His circumstances are the result of behaviour during an extended period of time of a sort that would disqualify anyone from counting as fully virtuous. But suppose now that the man realizes what a mess he has created and that he suddenly wants to act rightly. Furthermore, let us assume with Hursthouse that there is a clear sense in which it would be “worse to abandon A than B” (Hursthouse 1999, p. 47).Footnote 13 While we might be able to say that in these circumstances it would be arrogant, callous, cruel, and perhaps irresponsible for the man to go ahead and marry B instead of A, we cannot, if we hold on to VR, say that it would be right for the man to marry A, just because that is not what a fully virtuous agent would characteristically do in the circumstances. Indeed, nothing the man can do in the circumstances would meet this condition, since his are not the kind of circumstances that a fully virtuous agent could ever be in. And this, it seems, really constitutes a problem for VR. Maybe (I am far from confident about it) there are cases in which there is no available course of action that would count as right, but the case that we are considering here does not seem to be such. The man undoubtedly “merits … blame, for having created the circumstances that made it necessary for him to abandon B” and therefore “should be feeling ashamed of himself, not proud, and so on” (Hursthouse 1999, p 47). But if we just focus on the question which action that now would be right for the man to perform, it appears as if the answer is marrying A.

It is worth noting that Hursthouse, who indeed is the most well-known champion of VR, says that she is prepared to accept the implication that the man does not act rightly in marrying A.Footnote 14 Still, in so far as it would be arrogant, callous, and cruel of the man to marry B instead of A, Hursthouse argues that marrying A is the right decision, even though acting in accordance with the right decision in these circumstances cannot be assessed as right action. I find it counter-intuitive to say that the man in the example that we have been discussing would not act rightly in marrying A. Furthermore, Hursthouse obviously owes us a separate account or criterion of right decisions. Unfortunately, the only suggestion Hursthouse has to offer as for what such an account might look like is that a decision is right if and only if it is what a virtuous agent would decide in the circumstances.Footnote 15 This will not do, of course, since Hursthouse is granting that there might be right decisions to make even in circumstances that no virtuous agent could be in.

2.2  

It is arguable that VR is not sufficiently sensitive to how people’s current character flaws might affect whether they should do what a fully virtuous agent would characteristically do in certain circumstances.Footnote 16 Suppose that John is going away for some weeks to visit his parents. Since he is expecting a very important letter, John asks his colleague Peter to forward his mail to his parents’ address. Given the importance of the letter, it is crucial that if Peter promises to do what John is asking of him, then John should be able to trust that Peter actually fulfils his promise. Now unless, say, Peter is going away himself, it seems as if promising to help John is what a fully virtuous agent would characteristically do in Peter’s circumstances. However, while a fully virtuous agent could be trusted to keep his promise to help John, Peter is well aware that he cannot be trusted to do this. In fact, Peter is very likely to quickly forget about the whole thing. Because of this it seems intuitively correct to say that Peter would not act rightly in promising to help John. On the contrary, the right thing for Peter to do is rather to turn down John’s request in order for John to be able to make other arrangements.

2.3  

A third counter-example to VR could focus on the practice of asking other, hopefully more virtuous, persons for guidance about what one should do. As even defenders of VR often point out, it is quite likely that those of us who are not yet fully virtuous sometimes will be unable to see for ourselves what a fully virtuous agent would characteristically do in our circumstances. However, while it seems plausible to say that if we find ourselves in a case of uncertainty about what to do, and there is a virtuous agent around that we could ask for guidance, then asking the virtuous agent for guidance would be the right thing to do, this is not something that could be accounted for in terms of what a fully virtuous agent would characteristically do in the circumstances. A fully virtuous agent has practical wisdom and she would, therefore, characteristically be able to figure out for herself what to do.

2.4  

Finally we shall consider an objection according to which VR does not leave proper room for seemingly plausible requirements of moral development.Footnote 17 This problem could also be said to result from the fact that VR is not sufficiently sensitive to the relevance of people’s current flaws of character. This time, however, the objection does not so much concern cases in which our current character flaws seem to oblige us to avoid doing what a fully virtuous agent would characteristically do in the circumstances (as in (ii)), or require us to ask someone wiser than ourselves for guidance (as in (iii)). The present objection rather puts into question whether VR can accommodate requirements of actively trying to make ourselves more virtuous.

Think of a person who has developed a firm habit of telling small lies. Without giving it much thought, he tends to make up stories about places that he has been to, people that he has met, and so on. Suppose now that this person becomes aware of his bad habit and wants to get rid of it. To succeed in this task, it seems as if he can try a number of different strategies. For instance, he could try to making it a general principle to stick to the truth at all times. But granted that lying has become a deeply engrained habit for him this may not be very effective. So, in the words of Robert Johnson, maybe the person instead,

after consulting with a therapist … decides to begin writing down lies that he tells, no matter how insignificant, to become more aware of his habits and to keep track of improvements. Further, whenever he is aware of temptations to lie, he tries to do develop a concrete idea of what would happen if he told the truth. Who, exactly, is protected by my lie? Why do I want to protect her? Who, exactly, would be dismayed if my lie were discovered? (Johnson 2003, p. 817)

Now whatever strategy the person in the end chooses in order to get rid of his undesirable habit, it seems reasonable to think that it would be right for him to engage in some such self-improving behaviour. But, of course, no ideally virtuous agent would characteristically engage in self-improving behaviour of this kind. Such an agent, after all, is already fully virtuous and therefore does not need to improve.

3  

The counter-examples considered in the previous section strongly suggest that even though VR might seem initially attractive, it cannot in the end be quite right. If one is less than fully virtuous, then it will sometimes be right for one to do something that no fully virtuous agent would characteristically do in the circumstances. Could VR be revised in some way to accommodate this?

Valerie Tiberius has recently argued that “the best way for a virtue ethicist to think of right action is the following … An action A is right for S in circumstances C iff it is the action in accordance with the reasons that would guide the action of a completely virtuous person acting in C” (Tiberius 2006, p. 248). I will refer to this account as VRT.

Recall the last counter-example to VR above, that is, the case of the habitual liar who, in order to get rid of his bad habit, consults a therapist and is advised to start writing down his lies and how he thinks that his lies affect people in his surroundings. Tiberius suggests that

the important feature of context in this case is that the liar believes he is falling short in ways that he wants to remedy. The fully virtuous can be in such circumstances because the fully virtuous agent is concerned about the development and upkeep of her character. We should not conceive of the fully virtuous as someone who does not worry about temptation or who does not think she might fall short of her standards, rather, the ideally virtuous person is someone who is engaged in a project of character development and who is deeply committed to self-improvement. Now this fact does not help Hursthouse’s [VR] because the virtuous person will have no reason to do what the habitual liar needs to do since she does not need to overcome a terrible habit. But it does help [VRT] because the virtuous person’s reasons include the ideal of self-improvement and this means that the habitual liar can act on these reasons in order to take self-improving action (Tiberius 2006, p. 253).

Thus, a fully virtuous agent can be in circumstances that are relevantly similar to the circumstances of the liar. The reasons that would guide the conduct of the fully virtuous in circumstances where the important feature is that she wants to improve her character include reasons of honesty: “lying is disrespectful and unfair to others, it takes advantage of a good social practice without contributing to it, and so on” (Tiberius 2006, p. 254). In the light of these reasons, the fully virtuous would act honestly. She would tell the truth, keep her promises, etc. This is of course very different from starting to write down one’s lies with a view to overcoming a bad habit, but Tiberius wants to say that in following his therapist’s advice, the liar would indeed act in accordance with the same reasons that would guide the fully virtuous agent to acting honestly in the circumstances.

Essentially the same line of argument, according to Tiberius, can be used to show that VRT also escapes the other counter-examples to VR that we looked at earlier. For example, if we assume “that the important feature of context” in the case where John asks Peter to help him with the mail is that Peter is in possession of a character flaw that needs to be remedied (namely that he cannot be trusted to fulfil his promises), then even though a fully virtuous agent in relevantly similar circumstances would be led by reasons of (say) kindness and honesty to promise to help John, the best way to act in accordance with those reasons for someone like Peter is instead to refrain from making the promise to John. “Similarly”, writes Tiberius, “an imperfectly virtuous person who asks for guidance may be acting in accordance with a maxim of honesty, generosity or the like, because given her ignorance about what reasons require, asking for guidance is the best way to act in accordance with them” (Tiberius 2006, p. 256).Footnote 18

From a virtue ethics perspective, perhaps the most obvious problem with VRT is that it seems to constitute a reasons-based, rather than a virtue-based, account of right action. One may wonder why we could not simply drop the reference to the fully (or “completely”) virtuous agent in VRT and instead account for right actions as actions in accordance with the relevant reasons, that is, reasons of honesty, generosity, justice, and so on. Tiberius acknowledges this problem, but suggests in response (referring to McDowell 1997) that the reference to the virtuous agent is important because “the reasons of the virtuous cannot necessarily be grasped by those without the virtues” (Tiberius 2006, p. 259). However, this just seems to mean that the correct application of reasons of honesty, generosity, justice, and so on, in many situations requires judgment of a kind that is characteristic of the fully virtuous, and that is something many moral theorists who do not subscribe to virtue ethics would agree with.

Furthermore, I fail to see why we should accept Tiberius’ claim that a fully virtuous agent and, for example, the habitual liar can be in circumstances that are relevantly similar. It seems correct that a virtuous ideal should be concerned about the upkeep of her character and be committed to continuously exercising her virtues throughout her life, something which perhaps might give her reason to occasionally rehearse the considerations for why she should behave, say, honestly. But this is different from saying that the fully virtuous could ever have reason to improve her character. In what respects could the fully virtuous improve? It is tempting to think that if a person finds that he needs to improve his character, then that shows that he is not yet fully virtuous.Footnote 19

4  

Consider now instead the following alternative:

VRA: An action A is right for S in circumstances C if and only if a fully virtuous agent would characteristically advise S to do A in C.

The possibility of moving towards an advice model of rightness in virtue ethics has been acknowledged by several writers.Footnote 20 It is often compared to moves made in other contexts of practical philosophy. For example, with respect to practical reasons Michael Smith has argued that “[t]he content of our reasons is thus fixed by the advice we would give ourselves if we were fully rational” (Smith 1995, p. 112). And in a well-known paper, Peter Railton defends the view that a person’s good is a matter of what a fully informed version of the person would want for himself as he actually is.Footnote 21 The comparison between VRA and Smith’s and Railton’s accounts is of some importance since it points to something that should be made explicit: in order to ensure that the advice is properly grounded in facts about the advisee’s actual circumstances, it must be assumed that the virtuous advisor in VRA is fully informed about the advisee’s situation, including the strengths and weaknesses of the advisee’s character.

On the face of it, VRA seems quite promising. Adopting VRA helps virtue ethics to escape the four counter-examples to VR. Even though there are circumstances which no fully virtuous agent could get into, there is nothing to prevent the fully virtuous from advising other, less virtuous people about what they should do in such contexts. In Hursthouse’s example with “the distinctly non-virtuous man who has induced two women [A and B] to bear a child of his by convincing each that he intends to marry her” (Hursthouse 1999, p. 50), a fully virtuous advisor would presumably advise the man to marry A, since if the man were to marry B instead, then his action would be arrogant and callous. Furthermore, a fully virtuous advisor, as we have said, must take into account facts about the deficiencies of other people’s characters in determining what they should do. Thus, a fully virtuous advisor would plausibly advise someone like Peter to turn down John’s request for help, direct us to ask for guidance when we are ignorant about what to do, and tell the habitual liar to follow his therapist’s directions and to start writing down his lies.

In addition, VRA arguably steers clear of at least two independent objections that have been launched against it. First, there are many decisions about how to act in everyday life that we are, or at least should be, capable of making ourselves.Footnote 22 Indeed, even in not so ordinary circumstances, there are decisions to be made which seem so personal that it is more fitting that one makes those decisions oneself, rather than on the basis of someone else’s advice, regardless of concern for whether the latter is more likely to lead to right action. In these sorts of cases, fully virtuous agents might reasonably think that it would be wrong for them to advise others about what they ought to do. But this does not constitute a problem for VRA, simply because VRA does not entail that the fully virtuous will (or would) actually impart their advice about what we ought to do in all cases.

Second, it has been objected to VRA that by ascribing something such as a full information component to the virtuous ideal, “the view collapses into a version of an ideal observer view” and [t]he virtue element would no longer be doing work in the account beyond providing some general goodwill that is already present in ideal observer views in the form of universal benevolence (Johnson 2003, p. 828). It is true that by adopting VRA as its account of right action, virtue ethics is moving in the direction of a kind of ideal observer view. But the characterization of the virtuous ideal that we find in virtue ethics is certainly different from the standard characterizations of the ideal observer.Footnote 23 For one thing, (universal) benevolence does not figure on Aristotle’s list of virtues. Contemporary virtue ethicists might of course want to revise Aristotle’s list in this respect, but even if they do they will also ascribe a number of other virtues to the virtuous ideal, such as courage, justice, temperance, and practical wisdom, and these virtues will indeed do a lot of work in determining the conduct and advice that would be characteristic of the fully virtuous agent.Footnote 24

As it stands, VRA is nevertheless untenable. The reason for this is that there are cases in which a fully virtuous advisor might advise another person on strategic grounds not to do what is right.Footnote 25 For just one example, when confronted with someone who is quite obstinate, the fully virtuous might choose to advise that person to do not-A, even though the person should really do A, if that is the advice that is most likely to lead to the person’s acting rightly (that is, to his doing A).

This particular problem can perhaps be avoided if we just add a qualification to VRA saying that in order for an action A to be right for S in circumstances C, it must be the case that A is what a fully virtuous agent would characteristically advise S to do in C, were S a reliable follower of advice.Footnote 26 Trouble remains for VRA, however, even if a qualification of this kind is put in place. To see why, consider the case of Julius whose one character flaw is that he is not a reliable follower of advice; no matter how good the advice is, he is quite likely to do the opposite to what others advise him to do. It seems plausible to think that it would be right for Julius to do something in order to get rid of this character flaw. But if what it is right for Julius to do is determined by what a fully virtuous agent would characteristically advise him to do, were he a reliable follower of advice, then it would not be right for Julius to take steps towards ridding himself of his bad habit of not listening to other people’s advice. Why? Well, simply because no fully virtuous agent would characteristically advise Julius to take such steps if he were a reliable follower of advice.Footnote 27

5  

At this point, virtue ethicists might turn instead to:

VRAP: An action A is right for S in circumstances C if and only if a fully virtuous agent would characteristically approve of S’s doing A in C.Footnote 28

VRAP is structurally very similar to VRA.Footnote 29 In contrast to VRA, however, VRAP seems fit to deal with cases in which a fully virtuous agent would characteristically advise another person not to do the right thing. Even if the fully virtuous would characteristically approve of the other person’s doing A, the virtuous agent could still characteristically advise the person to do something else instead, if that is the advice that is called for in the circumstances.

Unfortunately, VRAP is afflicted by other problems. Here we shall mention two. First, VRAP fails to discriminate between actions that are right or permissible, and actions that are supererogatory.Footnote 30 In cases where there are different courses of action which seem morally acceptable, but there is one course of action that is clearly better than the others, the fully virtuous might approve only of our choosing the best. Second, VRAP does not seem quite capable of dealing with cases in which a fully virtuous agent would characteristically be indifferent towards a person’s choice between two courses of action. Suppose, for example, that Jones can choose between two equally good dishes for lunch (pasta and fish, say). It appears awkward to say that a fully virtuous agent would characteristically approve of Jones’ choosing one or the other of these dishes, even though either choice would be right (permissible or acceptable).

In the light of the second problem for VRAP, we can try:

VRAP*: An action A is right for S in circumstances C if and only if a fully virtuous agent would not characteristically disapprove of S’s doing A in C.Footnote 31

Since a fully virtuous agent would not characteristically disapprove of either choice in Jones’ case above, VRAP* entails that both dishes are permissible options for Jones. So far so good, but does VRAP* fare any better than VRAP with regard to respecting the difference between actions that are merely permissible or morally acceptable, and actions that are supererogatory? I think the answer to this question is “No”. The reason for this is simply that the fully virtuous would characteristically disapprove of our doing anything but the (or perhaps a) best action in the circumstances. Perhaps someone wants to deny this claim on the ground that the fully virtuous is not concerned about other people’s behaviour to the extent that she would be busy disapproving of what others do as soon as they do anything less than the best. In response, however, we may note, first, that the fully virtuous must at least be concerned about what other people do to the extent that she characteristically disapproves of those actions that we intuitively would consider to be below the threshold for what is permissible or morally acceptable. If the fully virtuous is someone who is not concerned even to this extent, then VRAP* would still have to be rejected. Second, and for our purposes more important, we need to keep in mind that the fully virtuous agent is an ideal that possesses all the virtues to the fullest extent, and has full information of the circumstances. In real life, of course, a virtuous person will usually not be informed enough to properly judge whether others have done the best they could or not (nor is it very likely that she has actually developed the virtues to the full). But if the virtuous person had full information (and all the virtues to a maximum degree), then it seems quite plausible that she would characteristically disapprove of others not acting as well or excellently as they are capable of.

6  

The revised versions of VR that we have been looking at so far have all retained the reference to a fully virtuous agent. But maybe virtue ethics should lower the standard a little and adopt an account of rightness such as the following:

VRD: An action A is right for S in circumstances C if and only if a decent person would characteristically do A in C.Footnote 32

One virtue of VRD is that it leaves room for the category of supererogatory actions. If we think of a decent person as “simply defined by being nonvicious” (Brännmark 2006, p. 595), then we can say that while such a person’s characteristic actions will not always be as admirable or good as the actions of the fully virtuous, they are nevertheless morally all right; they constitute the lower limit, as it were, for what we are allowed to do. Furthermore, VRD seems to avoid the counter-examples to VR found under (ii), (iii), and (iv) in section 2. A merely decent person might turn down a request to help someone like John with the mail just because he knows that he is likely to fail in his task if he promises to help, he might ask for guidance in cases where he (precisely because he is not perfectly virtuous) is unable to figure out what he ought to do, and he might choose to act in ways that are conducive to the moral development of his character.

VRD will not save the day for virtue ethics, however. For one thing, VRD entails that if someone has landed himself in circumstances that no decent person could be in, then whatever the person does will count as wrong. But that is, at least for many cases, intuitively implausible. For example, the right thing to do for the man in Hursthouse’s example is to marry A, but we cannot explain that by reference to what a decent person would characteristically do in the circumstances, since no decent person, just as no fully virtuous person, could ever have got himself into the man’s situation to begin with.Footnote 33

For another thing, even though it is true that decent or nonvicious people, while acting in character, might aspire towards moral improvement, they need not do so. It is not an essential feature of being nonvicious that one characteristically aspires to become better. According to VRD, it is therefore permissible for everyone not to aspire to become better. But this seems wrong. We intuitively want (don’t we?) to say that most people really are at least pro tanto obligated to try to improve their characters in different respects, but VRD does not allow for that.

Of course, one may ask how far reaching this obligation to improve is. From an Aristotelian virtue ethics perspective, I believe it is very far reaching: it applies to every person who has not yet achieved the ideal of a fully virtuous character. In the light of this, perhaps someone wants to suggest that it is a virtue of VRD that it does not entail that we are required to aspire towards full virtue; that this feature of VRD in fact relieves virtue ethics of an unattractive perfectionist element.Footnote 34 However, we should note, first, that VRD does not even entail that we are obligated to aspire towards acquiring an overall decent character. It is true that in order for our actions to be right, according to VRD, they must conform to what a decent person would characteristically do in the circumstances, but we do not actually have to be, or aspire to become, overall decent to fulfil that requirement.Footnote 35 Furthermore, for my own part, at least, I must admit that I just do not see anything unattractive or unduly perfectionistic about the idea that we are (at least pro tanto) obligated to strive towards full virtue; on the contrary, I find it quite compelling to think that we indeed are obligated to do so.Footnote 36

Now, having introduced the notion of the decent person, we might (as an anonymous referee pointed out to me) consider replacing VRD with, for example:

VRDAP*: An action A is right for S in circumstances C if and only if a decent person would not characteristically disapprove of S’s doing A in C.

In contrast to VRAP*, which we considered in the previous section, VRDAP* seems to allow for a difference between the merely acceptable and the supererogatory, since a decent person, presumably, would not characteristically disapprove of our performing actions that are morally ok, even if not the best possible actions in the circumstances. It might perhaps also be suggested that VRDAP* avoids the two problems pertaining to VRD: if the man in Hursthouse’s example does not marry A, then a decent person would characteristically disapprove of the man’s behaviour, and maybe it is constitutive of being a decent person that one also would characteristically disapprove of people who do not aspire towards acquiring at least a decent character.

However, as I have indicated just above, I believe the requirement to improve oneself goes much further than what VRDAP* allows for; the requirement does not just stop at decency. Another serious problem for VRDAP* is that a decent person might not disapprove of other people’s wrongdoings. It seems entirely compatible with being a decent person that one characteristically fails to muster any disapproval towards the wrongful actions of others (or at least towards certain kinds of wrongful actions that others perform; maybe there are actions that are so bad that anyone who characteristically fails to disapprove of them would fall short of decency). For example, even if a decent person would not himself characteristically insult other persons, he might characteristically remain indifferent when hearing two strangers insult each other in morally unacceptable ways.

Finally, it is worth noting that even though VRD and VRDAP* seem to leave room for supererogatory actions, it remains to be shown what a virtue ethical account of such actions more specifically might look like. We cannot just combine VRD or VRDAP* with the view that an action is supererogatory if and only if it is what a fully virtuous agent would characteristically do in the circumstances. It is hardly plausible that everything a fully virtuous agent would characteristically do is supererogatory.

7  

While VR at first glance seems a very natural way of accounting for right action within the framework of virtue ethics, we have seen that it is subject to several damaging counter-examples. We have furthermore considered and rejected a number of different proposals for how to revise VR in the light of these counter-examples. It is true that there still might be some way of revising VR that would be more successful than the alternatives we have looked at in this paper, but I think our inventory suggests that no such account is forthcoming.

Should we then give up the idea of virtue ethics as a distinctive position in substantive ethical theorizing? That is a possibility that needs to be taken very seriously. At some point we may just have to accept that even though any complete ethical theory should contain a component dealing with virtue, it should also contain a criterion of right action and the latter cannot be understood in terms of the former.Footnote 37

Another possibility might be for virtue ethicists to take sides with moral particularists and reject as implausible the whole project of searching for an account or criterion of rightness that holds true without exceptions. At least some virtue ethicists might be fairly comfortable with this move. Drawing on Aristotle, they might argue that moral life is such a complex matter that in philosophical ethics we often have to rest content with what is true for the most part, and maybe we should not be particularly surprised to find this to be the case with regard to right action.Footnote 38

But is there really no hope of finding an account of rightness that is structurally different from VR (and thereby also from the various revisions of VR that we have considered), but nevertheless distinctly virtue ethical? I can think of one such account. According to virtue ethics, at least of an Aristotelian bent, we should all aspire towards becoming fully virtuous. In book II of the Nicomachean Ethics, Aristotle argues that acquiring virtue is in many respects similar to acquiring a practical skill.Footnote 39 We might thus learn something important about how we become virtuous by considering what is involved in, say, becoming a builder or a piano player. The beginning builder or piano player is required to do a number of things which an expert in either of these fields would not need to do. The beginner has to learn by following the example of a role model, he has to practice very basic things, he will from time to time make mistakes which require him to do something all over again, and so on and so forth. Of course, as he gets better, what is required of him will change in different ways. Assuming that Aristotle’s skill-analogy is correct, then something very similar is true with regard to the acquisition of a virtuous character. It might be suggested that virtue ethics therefore should consider adopting what Julia Annas has recently called a “developmental account” of virtue and right action.Footnote 40 Instead of saying that the rightness of an agent’s action is determined by whether it is what a fully virtuous agent would characteristically do (advise, approve, etc.) in the circumstances, the developmental account says that the rightness is determined by whether the action is in accordance with, or perhaps appropriate at, the agent’s own specific level of development in virtue.

An account of this kind might also provide the ground for an interesting explanation of the phenomenon of supererogation. David L. Norton, for example, has argued that in many situations it is possible for us to act in ways that go beyond what we are required to do in the light of our specific levels of virtue development.Footnote 41 On this picture, what counts as supererogatory for one person may be morally required for someone else who has reached a higher level of development in virtue.

The developmental account is intriguing, but it has not yet received any very detailed treatment in the literature.Footnote 42 While a detailed treatment unfortunately is beyond the scope of this paper as well, I will end by drawing attention to some of the obstacles confronting the account.

To begin with, it remains to be firmly established that there indeed are important similarities between the development of virtue and acquiring a practical skill, something which requires further research not only in philosophy, but also, I suppose, in developmental psychology.

Second, in order not to entirely exclude some people from the possibility of performing right actions, the developmental account presupposes that it is always possible to begin developing a virtuous character. This is not uncontroversial, however. Aristotle, for example, argued that before a person can acquire virtue, she “must in a way already possess a character akin to it, one that is attracted to the fine and repulsed by the shameful” (Aristotle 2002, 1179b29-31). But whether someone indeed is in possession of a character “that is attracted to the fine and repulsed by the shameful”, depends on whether she has received a proper upbringing.Footnote 43 Defenders of the developmental account need to make a case for thinking that, pace Aristotle, even those of us who were not correctly brought up always have it in our power to somehow initiate the process of developing virtue.Footnote 44

Perhaps the main worry about the developmental account, however, is that it seems too vague or uninformative to constitute much of an account of right action at all. The claim that right action depends on our individual levels of development in virtue may be important in so far as it urges us to keep in mind that what is right for a person at one specific level of development is not necessarily the same as what is right for a person at a different level of development. But on the face of it, at least, the claim does not really tell us anything substantive about what is actually right at a specific level of virtuous development. It is hard to see how the developmental account could ever be used to assess whether a particular action is right or not.

These are serious obstacles, but it is too early to say whether they are insurmountable. We need to know much more about what the developmental account specifically amounts to before we are in a position to determine whether it constitutes an interesting contender in substantive theorizing about right action, or if it rather should be set aside as another dead end in virtue ethics.