The Structure of Virtue
Julia Annas

I

            As a long-time worker in the field of virtue ethics I cannot but be pleased by its recent increased importance in ethical theory, and also by the recent emergence of a form of virtue epistemology which takes its inspiration from virtue ethics. This has been due to the work of Linda Zagzebski, especially her book Virtues of the Mind, [i] , which has opened up many exciting paths of research exploring the links between virtue ethics and epistemology. This is a gripping and seminal book, which will surely change the contours of its field, and bring together two areas which have functioned in mutual isolation and can only gain from the discovery of their links. We all owe Zagzebski thanks for her pioneering work and its effects.  My own contribution comes from the direction of virtue ethics, and I shall be exploring two aspects of the structure of virtue, as that has developed in the ancient virtue ethics tradition, which have implications for the relevance of virtue to epistemology. I shall have less to say about the details of the application, since epistemology, at least modern epistemology, is not my area of specialization ; but I am fairly confident that they are central to the project of using a rich notion of virtue to illuminate epistemological issues.

            The issues I shall focus on are those of virtue and skill, and virtue and success. In both cases we get a clearer picture if we look at the whole ancient virtue tradition, rather than emphasizing Aristotle. For contingent historical reasons Aristotle’s has been the theory on which most philosophers focus when they turn to virtue. [ii] But treating Aristotle as authoritative for virtue ethics fails to do justice even to the ancient tradition. For hundreds of years different theories were proposed within the framework of happiness and virtue, and there was extensive inter-theory debate. As a result, we can separate the framework and main assumptions of virtue ethics from the specificities of Aristotle’s own theory. Sometimes this can turn out to make a large difference as to what is implied by the use of a ‘virtue ethics’ approach, and I shall be arguing that for these two issues it does. In both cases, if we look at the whole virtue tradition, we find important implications for the relation of the moral to the intellectual virtues, and, hence, for the relation of ethics to epistemology.

II

            Aristotle rejects the idea that virtue is a skill. (Virtue here is moral virtue, as indeed is standardly assumed in ancient ethical discussion [iii] ; we shall get to intellectual virtue shortly.) This may strike us as unsurprising, indeed mere common-sense. But it is significant that Aristotle is a lone standout here. The ancient virtue ethics tradition followed Plato and the Stoics in holding that virtue is a skill. That is, it is a kind of skill, there being other kinds as well; virtue is, as the Stoics put it, the skill of living. The claim that we should follow the ancient tradition rather than Aristotle may at first sound rather academic, but this issue of whether virtue is or is not a skill is not of merely historical interest: it raises philosophically crucial issues about the intellectual structure of virtue.

            Aristotle says that there are many points of difference between a virtue and a skill. This is obvious enough, and modern writers have developed and modified the list of differences which he sets out. [iv]   For example, skill involves a mere capacity, can be forgotten and is less precise than virtue is. [v]   However, we can reasonably ask how much such differences matter: the thesis that virtue is a skill is a claim that virtue is one kind of skill, and thus that the idea of skill is central in helping us to understand what virtue is. Against this claim, pointing out obvious differences between virtues and skills is ineffective. [vi]

            How can virtue be a kind of skill? It has the intellectual structure of a practical skill. Even Aristotle recognizes this; it enters right at the start of his account of virtue, where he compares learning to be just with learning to be a builder. You have a role-model, and first you copy what he or she does, then come to understand for yourself what the point is of doing what that person does. Increase of understanding goes with increased autonomy of reflection and action. This is why, for both skill and virtue, you need a teacher to begin with, but then become able to act on your own independently.

            Underlying this simple fact is a connected set of epistemologically interesting points about skill.  Firstly, a skill or expertise is teachable.  There is some intellectual content to be conveyed, not just picked up by external mimicking. Where there are teachers and learners, we have something which at least in principle is an expertise, not just a matter of an empirical ‘knack’ to be picked up. Secondly, the expert is someone who has an understanding of her subject-matter as a whole. This is a demanding condition. Someone learning, say, a language, will pick up bits of the subject here and there - the future tense, vocabulary and so on. The expert in the language will have mastery of all that is needed to understand the language, and, moreover, will see how it is all unified. Similarly, someone learning a practical skill like building will pick up bits of know-how and technique here and there; the expert, however, will have mastery of everything relevant to that kind of building, and will have unified that mastery so as to be able to understand his own and others’ successes and mistakes, and to be able to apply his skill in new situations without further learning being required. And thirdly, an expert is able to articulate her understanding of her subject, able to ‘give an account’ of it, logon didonai, in the ancient way of looking at it. She is able not only to unify the various judgements she makes within her field, and the actions she does, but to explain them and, if necessary, justify them, in terms of whatever general principles are needed to express understanding of the subject.

            These conditions are not independent of one another, since a teacher can scarcely teach if she is completely inarticulate about her subject, and what is taught must be a unified body of practical knowledge rather than a bunch of unconnected practical tips, if it is an expertise that we have. [vii]

             How is virtue a skill? It shares the intellectual structure of a skill in these three ways. This is visible in the account Aristotle gives of the acquisition of virtue, [viii] but it can be seen most lucidly in the Socratic dialogues of Plato. There, Socrates challenges people who appear to be experts about some virtue, such as courage, but fail to unite their isolated beliefs [ix] and to offer any articulate unified understanding of the matter; this shows that they lack understanding of the virtue in question. Laches, in the dialogue named after him, can give examples of courage by pointing to men fighting in battle. But at first he can provide no account at all of what this kind of action might have in common with other kinds of brave behaviour - coping bravely with illness or poverty, for example - and when he does, his suggestion is obviously hopeless at explaining how and why all the very diverse kinds of brave action are brave. Laches, who is supposed to be an expert in bravery, has failed to convey any articulate understanding of bravery; clearly he is, despite appearances, no expert in it. The same undermining of claims to expertise in moral matters occurs in other dialogues, and Plato scholarship has for some time now recognized that the kind of knowledge or understanding which is required and found lacking is the kind of knowledge that an expert possesses. [x] Indeed, Socrates is always appealing to practical skills such as those of the navigator, doctor or farmer to illustrate the kind of practical understanding that he seeks in moral matters.  Nor is this concern pecular to Plato. The Stoics take over the thesis that virtue is a skill and develop it explicitly and at length, for the same reasons which appear in the Socratic dialogues. [xi] It became so standard in ancient ethical theory that it could be taken for granted in any serious debate between ethical theories. 

            There are common objections at this point from modern philosophers. Kinds of practical expertise have ends which are fixed ; it is clear and uncontroversial what counts as success in a skill like navigation, or car repair. And these ends are conditional in their hold on our motivation; our interest in exercising them depends on our concern to obtain their ends. Can it then be reasonable to think of virtue as having the structure of a practical expertise, given that the end that virtue aims to achieve - a well-lived life - is one which is neither clear nor uncontroversial, and also one from which we cannot similarly become motivationally detached?

            The answer to this in the ancient virtue ethics tradition is clear, and best put by the Stoics (though it can also be found in Plato). Virtue is ‘the skill in living’, and living your life is an end which everybody has, and which, short of suicide, is non-detachable. By the time you start to reflect about your life and the best way to live it, you already, as we put it, have a life. You already have a family context, for example, and a socio-economic context, with some kind of employment and income. You already are the production of some kind of education, including moral education, and have certain values and priorities. For the ancient virtue tradition, all this is your raw materials, on which you get to work as you develop virtue, aiming to make your life a product of understanding rather than conformity, something unified around pursuit of good values rather than driven by isolated desires or run by the values of others. Virtue, then, is a global expertise in your life, and will always differ from local kinds of expertise in just these two ways, namely that the end we seek in becoming virtuous is not antecedently fixed in the way that the end of car repair is, and also that living our life is not an end that we can cease to care about, as we can cease to care about having the car fixed. We can choose, of course, to live our lives in a thoughtless and random way rather than to live them in a way which tries to improve them in the light of unifying understanding; but this does not make the end of virtue detachable in the way the end of local skills are. [xii]

            Moral virtue, then, is a skill in the ancient virtue tradition; it is an expertise, a kind of practical knowledge. Local, mundane skills serve as examples of the kind of unified practical understanding which, if we become virtuous, will order our lives in a unified way based on understanding.

            How, then, does moral virtue relate to intellectual virtue? The right answer has been elegantly stated recently by Paul Bloomfield. [xiii] Moral virtue is one kind of skill, intellectual virtue is another. Moral virtue, as explained, is a kind of practical knowledge which is illuminated by practical kinds of expertise. Intellectual virtue is another kind of skill. Neither should be seen as a sub-kind of the other - although of course any realistic account of the moral life will find many complex connections between them. [xiv]

            It is only to be expected that intellectual virtues should have a strong intellectual structure unified by understanding. Must an intellectual virtue, however, have the same intellectual structure as the moral virtues that we have seen sharing the structure of practical skills? If expertise is marked by the three conditions discussed above, of teachability, unified understanding of the field as a whole and articulate ability to give an account of what is understood, then intellectual and moral virtues will share this structure. Intellectual virtues, however, appear to be more various in their structure than moral virtues are, in a way that doubtless owes something to the fact that theoretical skills are more various in their structure than practical skills are. Aristotle’s intellectual virtues in Nicomachean Ethics VI are highly diverse. The virtue aiming for demonstrative knowledge, for example, is different in structure from the virtue aiming for non-demonstrative knowledge. The same goes for the Stoic sub-divisions of wisdom, and for Plato’s collection of intellectual virtues. [xv] The structure of an intellectual virtue will naturally depend on the scope and type of the relevant intellectual skill; it would seem that we might have several differently structured intellectual virtues which all met the conditions for expertise.

            It could be argued that the moral virtues essentially involve emotions and feelings in a way not true of the intellectual virtues. Indeed, moral virtue crucially involves in its development the progressive control and finally transformation of the person’s emotive side. [xvi] But it would be a mistake to hold that development of an intellectual virtue like perseverance or intellectual honesty never involves such control and transformation of recalcitrant, not purely intellectual elements of the person. Moreover, development of the intellectual virtues may straightforwardly require such transformation of the emotions and feelings by way of the development of a moral virtue. Honesty in some research, for example, requires that the person not be under the influence of greed for money; indeed, honesty seems to be the same moral virtue whether applied in handling money matters or in conducting research.

            The real distinction emerges when we consider that moral virtue is essentially practical; it is the skill of living, where living, in the virtue tradition, is seen as essentially active, shaping your life so that it is ordered from within. The way you live is seen as actively reflecting and expressing your character and hence your choices. Intellectual virtue, on the other hand, is not essentially practical; it is theoretical in that it is directed at achieving aims other than good action. Particularly if we think of intellectual virtue as aimed at achieving truth, we can see that its aim is going to be distinct from that of moral virtue.

            Of course, there might still be a close connection between the two kinds of virtue, and most virtue theories have thought that there is. One view frequently found attractive is that the intellectual virtues, whose aim is truth, deepen the understanding which is the basis of the moral virtues. After all, the moral virtues are aimed at doing the right thing, and this can scarcely allow indifference to the truth of your beliefs about the matter. Even if the intellectual virtues enable us to discover truths about matters which are recondite and abstract, still our increased grasp of truth will serve to broaden and deepen the understanding at the basis of the kind of practical knowledge which is moral virtue.  As Zagzebski puts it, ‘[I]f it turns out that the ultimate end of truth and the ultimate ends of the moral virtues are all components of a life of eudaimonia, then the moral and intellectual virtues do not even differ in their ultimate ultimate ends.’ [xvii]

            What, though, if this turns out not to be the case? In the virtue tradition there are two conflicting lines of thought on this, both of which are found appealing by both Plato and Aristotle. While they think most of the time that seeking truth will form part of a life well-ordered by moral virtue, they both at some points express a contrasting thought: seeking truth can become an end indifferent to or even conflicting with the end of living according to moral virtue. [xviii] The attractiveness of the intellectual search for truth, and the intrinsic appeal of its objects, can lead humans away from the aim of living a morally ordered life. It can lead them to aim to devote their energies entirely to the search for truth, to the point of wishing to transcend the boundaries of human life altogether and to try to ‘become immortal’, as Aristotle famously says in this connection. In this case the pursuit of happiness in a morally unified life will have been disrupted. Someone who seeks truth in a way which is indifferent to or conflicts with living a morally virtuous life is still, however, exercising the intellectual virtues. It is unconvincing to claim that someone whose intellectual pursuit of the truth conflicts with leading a moral life must really be lacking in intellectual virtue. (Indeed, it is likely to be the intellectually virtuous achievers, rather than the intellectually faulty, who have this problem.) The intellectual virtues can, though they need not, have a differing aim from the moral virtues, since the theoretical aim of truth can come into conflict with the aim of moral virtue, which is a practical type of knowledge.

            The relation between intellectual and moral virtue that emerges from the virtue ethics tradition, at least in its developed ancient form, is that both are kinds of skill or expertise, whose aims can but need not converge. Taking virtue seriously in the epistemological framework of the intellectual virtues, then, does not give support to thinking of intellectual virtues as a subset of moral virtues, nor to taking epistemology to be properly subsumed under ethics. [xix] Taking both kinds of virtue seriously, however, may be fruitful in other ways. For one thing,  taking moral virtue seriously reveals how intellectual a structure it has, and this suggests that virtue ethics might get aid from epistemology, as well as epistemology benefitting from virtue ethics. Ethics and epistemology can produce mutual benefit from mutual study.

III

            The second issue is that of virtue and success.  Zagzebski frequently insists that ‘virtue is a success notion’, [xx] and in this she and others are following not only Aristotle, who insists that the virtuous person is successful (katorthotikos) but the Stoics, who call a virtuous action as performed by a virtuous person a success (katorthoma). [xxi]

            The success element in virtue is important for anyone wishing to develop an epistemology in which virtue plays a basic or foundational role. For knowledge is a success term if any is. Knowledge is not the state you achieve by doing your best though you fail, but the state in which you actually succeed in getting your claim right, and succeed in meeting the required conditions, whatever these may be.  Zagzebski defines knowledge as a state of cognitive contact with reality arising out of acts of intellectual virtue, [xxii] and obviously ‘act of virtue’ must be a success term here, or it would not be knowledge that we were defining. Virtue, then, must be a success term in virtue epistemology.

            When we look at the virtue ethics tradition, however, we find that the relation of virtue and success contains complexities, and that when these are examined we find that we must also introduce complexities into any attempt to make use of virtue in reaching a definition of knowledge.

            The virtuous person must have the right motivation, and must also reliably succeed in what she does. An act which fails to achieve its aim can be said to ‘lack[s] something morally desirable’. [xxiii] But what is the virtuous person’s aim in acting? She has two. One is her telos or overall aim, of living virtuously and acting from motives of virtue. [xxiv] Virtue, after all, is a settled state of the person, with the overall aim of making the person’s life as a whole be one way rather than another, virtuous rather than evil or complacent. (Living virtuously, further, either constitutes, or contributes to, happiness; but that is a distinct issue.) The virtuous person’s other aim is what the Stoics call her skopos or immediate target, which is what is aimed at in any particular case of acting virtuously. The target of a just distribution will be everyone’s getting what they are entitled to, that of a brave rescue will be the safe conveyance of people out of the burning building, and so on.

            Plainly, someone can succeed in achieving the immediate target of an action on a particular occasion without achieving the overall aim of living virtuously. This will be the case if the person is not virtuous, and so does an action which is the kind of action which a virtuous person would characteristically do, but does not do it as a result of the virtuous person’s motivation. [xxv] Equally plainly, a virtuous person can succeed in achieving the overall aim of living virtuously by performing a virtuous act, even if, through no fault of her own, she fails to achieve the immediate target. If the brave rescuer does everything he can, takes the appropriate precautions and so on, but the victims die anyway because they are shot on the way out by a deranged gunman who happens to be there, then the brave action has failed to achieve its immediate target, but not in a way which implies that the brave person has failed to achieve his overall aim of living virtuously and so acting, in this, case, bravely. It is crucial, therefore, in examining a virtuous act, to ask what kind of success is in question - success in achieving the overall goal or success in achieving the immediate target. For achieving the overall goal is a matter of having the right motivation (something, of course, which in a virtue ethics is the result of a lengthy and demanding process), and this is up to the agent, since it is she who makes her life be one kind of life rather than another. But success in achieving the immediate target may not be in this way up to the agent, and may depend on various kinds of moral luck. 

            From a virtue ethics point of view, which is the success that matters? Virtue ethics is concerned with the person’s life as a whole, with character and the kind of person you are. The right perspective on an action, therefore, will for virtue ethics be the one which asks about success in achieving the overall goal, rather than success in achieving the immediate target. What matters is what the person’s motivation was, and how this relates to her developed character and life as a whole; for this is her achievement, what she has made of her life. To the extent that success in achieving the immediate target depends on factors over which the person has no control - moral luck of various kinds - it will be of less interest to virtue ethics. Success or failure in achieving the immediate target will affect various judgements we make about the action, but if, like the Stoics, we distinguish clearly between the immediate target and the overall aim, it is achieving the latter, not the former, which will make the action a success, a katorthoma.  Here virtue ethics parts company with theories like (most forms of) consequentialism, for which it is the actual results that matter for our evaluation of the agent, [xxvi] and stands with Kantianism, for which what matters is the agent’s motivation. [xxvii]

            Virtue is a skill, in the virtue tradition, the skill of living your life in a way which turns your raw materials into a life lived with and from understanding. It is a global skill, as we have seen, and this explains why it is compatible with failure to foresee some particular circumstances. Our ordinarwhich is judged by its products, but is more like performance skills, such as acting or dancing, where the excellence that is judged is the excellence in the activity and not in some separable result. For with virtue it is not the results which define success: ‘actions initiated by virtue are to be judged right beginning from their first inception and not in their completion.’ [xxviii]

            This is an issue where privileging Aristotle can lead to confusion, since on this point he is confused. On the one hand, he insists on the praiseworthiness of virtue, and the importance of choice as opposed to action in distinguishing characters. [xxix] But on the other hand he also stresses success in achieving the target in the practice of various virtues, sometimes in cases where this is explicitly not up to the agent, as with the ‘virtue’ of magnificence (the ‘virtue’ of spending money on civic projects), which only a rich person can exercise. [xxx] There is an unresolved internal tension in his theory as a result, mirroring his uncertainty as to the role of external goods in the virtuous life generally. [xxxi] We can see the problem if we ask about the role of a ‘virtue’ like magnificence. To exercise it the person needs not only external goods, in this case money, whose possession is a matter of moral luck; he needs to be actually successful in his exercise of tasteful spending, producing what succeeds in impressing the audience without overwhelming them, and so on. Aristotle, however, also believes in the mutual reciprocity of the virtues; to have one you have to have them all. But he clearly does not believe that if you are fully brave, you have all the other virtues, and therefore have magnificence, and therefore have magically acquired lots of money and taste.  This example shows that, since the virtues are mutually reciprocal, none of them can depend for their exercise on moral luck; hence magnificence, which does so depend, is not a real virtue. Aristotle fails to draw this conclusion because he is too respectful of conventional views which think of the activities of rich people, like magnificence, as virtues.

            On this issue the Stoic view is much clearer and more defensible than Aristotle’s. Of course it is often not up to me whether my action achieves the immediate target; but is it up to me whether I succeed or fail in acting virtuously - that is, with the right motives, from a developed disposition and with the right reasoning? If it is not, then it is not up to me whether or not I can become a moral person; and the Stoics are not alone in finding this an unacceptable position. [xxxii]

            It is sometimes urged that we feel more admiration for the act which, as well as being virtuous and thus succeeding in the overall aim, also actually does get its target; that this is the sense in which the act that fails here is morally lacking. But this seems not to be true. Take Socrates’ defence speech, the Apology, [xxxiii] in which he uncompromisingly defends the values that he has lived by, and refuses to pander to the jury’s values even at the risk of being executed for not doing so.  Do we admire Socrates less because in fact he failed to swing the crucial thirty votes? [xxxiv] Do we think of him as a pathetic loser because he failed to express the degree of deference to the jury that would have secured his acquittal?  Surely, rather than finding his action morally lacking, we admire him all the more for refusing on this occasion to compromise his values - if anything, his knowing refusal to do what was required to secure his immediate target makes us more convinced that he succeeded in achieving his overall aim of living a virtuous life.

            It can be suggested [xxxv] that we continue to praise the agent, but give the act less praise. This distinction can do work in some kinds of ethical theory, but in a virtue theory is problematic. For the suggestion here would be that we praise Socrates for being virtuous, living the life he does and having the character he does, but on this occasion we fault his action. Why, however, do we fault it?  For being so uncompromising as to lead to failure in worldly terms. But this is to say that we fault it for being just the kind of act which this kind of person would do! To fault what Socrates did for its lack of success in achieving the target precisely is to fault Socrates for being the person he is, and for acting accordingly.  Of course we can deplore the actual results of the action on this occasion. Virtue ethics can account as well as other theories for the fact that often we wish that the world had gone well and been improved in a way that did not happen. [xxxvi] We, as well as Socrates, can regret that the thirty votes went the wrong way; there is no reason to think that virtue ethics is more indifferent to the results of actions than other theories are. But the relevant point cannot be put, within a virtue ethics, by separating assessment of the agent from assessment of the act. Insofar as it was a virtuous act, done by a virtuous person for reasons of virtue, it cannot be faulted from the virtue point of view. It is the jury we wish had been different, not Socrates’ action.

            It is also doubtless true that virtue is generally reliable in producing success in getting the immediate target. As Terence Irwin puts it, ‘It is easy to see why, in favorable external conditions, virtuous people will have more objective success than other people will have. For they will have done all that can reasonably be expected of them; and if they do that, they will have tried to find all the relevant information that they could reasonably be expected to find, taken proper care, and so on. It is not surprising that action on these principles will often result in objective success.’ [xxxvii] A virtue ethics approach can take all this into account; but when the virtuous person fails to get her target through no lack or fault of her own, a history of usual success here is not to the point. We have to choose which kind of success matters, and any virtue ethics in which the issue is clearly faced comes down on the side of success in achieving the overall aim, which is compatible with failure to achieve the immediate target.

            How does this matter for the application of virtue theory to epistemology? As I mentioned, knowledge is a success term, and so a theory which defines knowledge in terms that feature virtue must take virtue to be a success term. Knowledge will, of course, be defined in terms of intellectual, not moral virtue, but for a theory which holds that these have the same structure or that the intellectual virtues are a subset of the moral ones this issue will be the same. So we must ask: what kind of success in intellectual virtue will be required for a workable definition of knowledge?         

            Here it looks as though the answer we get from ethics about moral virtue is the wrong one for epistemology. We surely do not want to define knowledge in terms of an overall disposition to succeed which is compatible with particular failures due to ‘epistemic bad luck’. It would have the implication that I could have knowledge as a result of being intellectually virtuous, even though on this occasion, through no fault on my part, I am in fact wrong.  A viable definition of knowledge in virtue terms must surely avoid this, and so must be aiming at success in achieving the immediate target. I have knowledge if I am right, say the truth (and also meet some further conditions). But, as we have just seen, this is not what a theory of ethical virtue demands; what is of most interest for it is success in achieving the overall goal. It looks as though the requirement for virtuous success in ethics - achieving the overall goal -is precisely what a virtue epistemology has to reject. Similarly, the requirement for virtuous success in epistemology - success in achieving the immediate target - is precisely what a virtue ethics has to reject.

            It is because of this point, I think, that problems arise for attempts, such as Zagzebski’s, to define knowledge as a state arising from acts of intellectual virtue. What is an act of virtue, in this theory? It is defined as follows:

            ‘An act is an act of virtue A if and only if it arises from the motivational component of A, is an act that persons with virtue A characteristically do in the circumstances, and is successful in bringing about the end of virtue A because of those features of the act.’ [xxxviii] Zagzebski claims that ‘the concept of an act of virtue is something we would want in an ethical theory anyway’, [xxxix] apart from its application in epistemology. This may be true, but what I am now concerned with is the issue of how well it fits a virtue ethical theory in particular.

            Firstly, you do not need to have the virtue in question to perform an act of that virtue in the technical sense. [xl] In the virtue ethics tradition, this would mean that the act is not an act of virtue at all; it is just an ‘appropriate action’ such as anyone can perform when trying to become virtuous. An act performed by a non-virtuous person could not be ‘success’, a katorthoma, and so could not be a starter as an act of virtue in any sense.

            The impact of this, however, is somewhat softened by Zagzebski’s requirement that to perform an act of virtue you need to have the virtuous motivation in question, at least the motivational component of it, although you are not virtuous. The idea is presumably that you have the virtuous motivation, but do not (yet) have it in a sufficiently robust, reliable and integrated way to be virtuous. Hence you can perform an act of virtue in the technical sense, without having the virtue.

             Here the problem is that this is the wrong story about virtuous motivation. For all versions of the virtue ethics tradition, the motivation of the non-virtuous person is different from that of the virtuous. Aristotle says that the virtuous person is motivated by the ‘fine’ (kalon) while those who are not yet virtuous are not - they are motivated by advantage or pleasure.  Becoming virtuous is not a matter of already having a small amount of the right motivation which, so to speak, spreads and grows bigger; it is a matter of learning to change. Becoming virtuous is learning to acquire the right motivation;if you already had it, even on a small scale, becoming virtuous would be easier than it is. A virtue, in the virtue ethics tradition, is a complex matter, and coming to judge rightly and to be rightly motivated go in tandem, develop slowly and involve the person’s becoming responsive to considerations which precisely do not resonate with the non-virtuous. [xli] Someone without the virtue in question, then cannot possess its motivational component. If an act can be performed by a person without the virtue in question, then it can be performed without being motivated in the way appropriate to that virtue. The technical notion of an act of virtue tries to combine the point that it can be performed by someone lacking the virtue in question with the point that it is nevertheless brought about by means of that virtue, since it is to be brought about by that virtue’s motivational component. But these two points cannot be reconciled within a virtue theory, at least one in the virtue ethics tradition.

            Secondly, an intellectually virtuous person, in this kind of virtue epistemology, may reason in a way that is virtuously motivated and follows the appropriate intellectually virtuous reasoning; but they may fail, because of epistemic bad luck, to get the right result, and in that case we do not have an act of virtue, in the technical sense. That is, a virtuous person may successfully achieve her overall aim, but fail to perform an act of virtue in the technical sense because she fails to achieve her immediate target. The person is virtuous, and acts virtuously, from virtuous motivation and reasoning. Yet the act fails to be an act of virtue in the technical sense because of facts that have nothing to do with virtue. This is surely a strange and undesirable result. Admittedly an act of virtue in the technical sense is a term of art; but surely it is strikingly odd that to succeed in performing one you need precisely what virtue does not supply.

            This second problem makes it particularly clear that an act of virtue in the technical sense contains tensions which arise from the conflicting demands here of ethics and of epistemology. To serve in a definition of knowledge, the act of virtue in this sense must guarantee success in achieving its immediate target. But just this makes it utterly unlike acts of virtue within virtue ethics, for which the relevant kind of success is success in achieving the overall aim, even when the immediate target is unavoidably missed.

IV

            I have stressed two aspects of the structure of virtue, as that figures in the ancient virtue ethics tradition as a whole and not just Aristotle. I have not here, of course, had the scope to develop them in a way adequate to make them appear appealing, still less compelling, though a fuller exposition would, I think, show that they both have individual advantages and form part of a powerful and attractive type of moral theory.

            I have argued that if we take these two aspects of virtue - skill and success - seriously as they figure in the virtue ethics tradition, we find reason to doubt that virtue as it figures in that tradition can unproblematically be used as a basis for a traditional definition of knowledge.  Tensions emerge between the structural requirements of virtue in the virtue ethics tradition and the structural requirements of virtue as it figures in virtue epistemology. In itself this is far from fatal to virtue epistemology, or to the project of defining knowledge in terms of intellectual virtue. It simply shows that virtue epistemology will have some problems as long as it works with the notion of moral virtue which comes from the ancient tradition of virtue ethics.

            There are many possible responses to this. One might be to retain the goal of using virtue as the basis for a definition of knowledge, but to work with a notion of moral virtue which is explicitly a product of modern reflection, cutting ties to the ancient tradition. Some modern theorists of what they call ‘virtue ethics’ have already developed such notions. [xlii] Another version of this response might be to retain emphasis on moral virtue as it figures in the virtue ethics tradition, but to separate moral and intellectual virtue, treating them, for example, as distinct sub-kinds of skill, and to give an account of intellectual virtue which makes more of its differences from moral virtue. On either of these accounts, intellectual virtue as the basis for a definition of knowledge would not suffer from having to reflect the facts about success which make moral virtue in the ancient tradition problematic in application to epistemology.

            Another response [xliii] might be to think of virtue not as the basis for a definition of knowledge guaranteeing success in achieving  immediate targets, that is, appropriately achieved truths, but rather as the component of knowledge corresponding to justification or more broadly rationality, the analogue to the ethical overall aim of being virtuous. Intellectual virtue would on this view be what guarantees a global aim of being justified or having a rational approach, and would be thought of, like ancient moral virtue, as having succeeded if this overall aim were achieved, even if on occasion the immediate target were missed in circumstances of epistemic bad luck. This would mark a considerable divergence from traditional accounts of knowledge, for which the point of seeking knowledge is to guarantee the achievement of truth, in the right conditions. A radical approach might claim that these traditional accounts have proved unprofitable and that we would get more insight into knowledge by trying the above kind of intellectual virtue approach. A more moderate approach might claim that there is more to knowledge than traditional accounts have produced, and that we will gain from pursuing both approaches - aiming for the achievement of truths and aiming for overall rational justification, and taking them to be complementary rather than competing.

            Intellectual virtue could on these approaches be seen to be of value in itself as constituting the epistemologically well-lived life. And it might also be of value as being our best strategy for success in achieving our immediate targets. The interest of this suggestion will of course depend on a number of factors: whether intellectual virtue can have both these roles, and, if it cannot, whether it can have either, and which is of greater epistemological interest. Since, as I have stressed, epistemology is not my own field, I will leave these speculations to the epistemologists to criticize.

            In the ancient, long and rich virtue ethics tradition, moral virtue is not a more general concept which includes intellectual virtue, and the kind of success that virtue requires in ethics is radically different from the kind of success demanded by a notion of virtue which could serve to base knowledge. This shows only, of course, that in the virtue ethics tradition ethics did not serve as a basis for epistemology. [xliv] The idea that it might, and also the weaker idea that ethics might be illuminating for epistemology, are exciting ideas. I think, however, that they would be more defensible if we introduce, from the ancient virtue ethics tradition, complexities which make the application of ethics to epistemology more problematic and complicated, without reducing its interest, or its fruitfulness for further explorations.