II. DRIVER'S CONSEQUENTIALIST THEORY OF VIRTUE

Aristotle, Hume and Kant specify various psychological conditions that must be met by virtuous agents and virtuous actions. Aristotle – the classical source of virtue theory – gives a richly detailed account of virtue, claiming that virtues are admirable character traits, and that the exercise of virtue requires knowledge of the good, right feelings, and pleasure in the virtuous agent. Moreover, Aristotle maintains that virtue is a necessary but not sufficient condition for eudaimonia, or ‘human flourishing’.Footnote ¹ In contrast, Hume offers a far more basic characterization, according to which virtues are traits of which we approve, contemplation of which produces pleasure in us, and which we would desire to have attributed to ourselves. Hume notes that virtues have a tendency to produce good consequences, and bring ‘happiness to human society’,Footnote ² although he denies that the sole merit of virtues lies in their tendency to produce good consequences.Footnote ³ Kant differs from both Aristotle and Hume, claiming that virtue is a settled disposition to act out of respect for the moral law, despite any inclinations to the contrary.Footnote ⁴ Kantian virtue requires self-mastery and fortitude, as well as knowledge of the right.

In Uneasy Virtue, Driver argues that none of these psychological conditions are necessary for virtue. First, consider what Driver calls the ‘knowledge condition’ on virtue. Aristotle claims that virtue requires correct perception, including knowledge of the good, and Kant maintains that virtue requires at least knowledge of the good, although not correct perception of non-moral facts. In what I have labelled point (1), Driver rejects the knowledge condition, arguing that some virtues – modesty, blind charity, trust and impulsive courage, for instance – are ‘virtues of ignorance’, which depend on a systematic lack of knowledge.Footnote ⁵ Less controversial, I think, is Driver's argument that some virtues do not require ignorance, but are compatible with ignorance. In point (2), Driver draws on Bennett's discussion of Huckleberry Finn,Footnote ⁶ suggesting that, in aiding Jim's escape, Huck exhibits the virtue of sympathy, despite holding false beliefs about what is good, and hence lacking knowledge of the good.Footnote ⁷ In point (3), Driver argues against Aristotelian conditions of psychological congruence within virtuous agents, and against Kantian conditions of psychological conflict within virtuous agents.Footnote ⁸ Both Aristotle and Kant, it seems, add superfluous conditions to their respective accounts of virtue.

Driver has more sympathy for Hume's theory of virtue, although she does note that his claims about virtue are not particularly clear, and appear to be inconsistent. Driver identifies two competing, inchoate methods by which Hume characterizes the virtues: first, the virtues are traits the contemplation of which gives us pleasure,Footnote ⁹ and second, the virtues are traits that produce actions with good consequences.Footnote ¹⁰ In what I have labelled point (4), Driver argues against Hume's first, ‘projectivist’ method of unifying the virtues. If we see virtue as being whatever traits find approval, either from the individual agent or from broader groups, then we must judge that it is impossible for an individual or a broader group mistakenly to disapprove of a virtue. Yet such mistakes do seem possible.Footnote ¹¹ In light of this possibility, Driver prefers Hume's second method: a consequentialist account of virtue. If virtues are whichever traits promote good consequences, then individuals and whole societies can mistakenly disapprove of traits that really are virtues. Moreover, this consequentialist account would explain points (1)–(3). Virtue is compatible with ignorance, including ignorance of the good, and is compatible with inner congruence and inner conflict, because it is possible that ignorant, congruous or conflicted agents regularly and reliably act so as to promote good consequences.

Although I have not defended it in any detail, I find Driver's argument up to this stage convincing. Points (1)–(4) really do seem to undermine the psychologically rich theories of virtue offered by Aristotle, Hume and Kant, and Driver's psychologically minimalist, consequentialist alternative does account for (1)–(4). For the purposes of this article, let us assume that this is the case, for it is the next stage of Driver's proposal that I wish to challenge. As we shall see in section III, Driver goes on to argue that points (5)–(8) also count in favour of consequentialism about virtue.

III. ADVANTAGES OF CONSEQUENTIALIST VIRTUE

Driver notes, in my point (5), that there are deep and persistent disagreements between Aristotelians and Kantians over the psychological conditions required for virtue. Aristotelians admire desires that are aligned to the good, whereas Kantians stress the value of self-control. According to Driver, consequentialism explains this disagreement, while allowing us to move beyond it:

Driver explains the disagreement by recognizing the validity of both intuitions in particular cases, but refusing to generalize them. Whilst denying that the virtues are marked out via shared psychological features, though, Driver does not offer a brute, particularist account of the virtues. The mistake, she thinks, was to look to motive rather than to consequences as the common ground amongst virtues.

In point (6), Driver argues that well-known problem cases, such as the scapegoat case, produce deeply conflicting moral intuitions within most of us. Consider the unwilling sheriff, who refuses to put the innocent man to death in order to prevent a murderous riot. For standard consequentialist reasons, Driver claims that the unwilling sheriff acts wrongly here.Footnote ¹³ Yet, she claims, he acts wrongly because he is a good person; he possesses a virtue – say, a sense of justice – that prevents him from performing the right action.Footnote ¹⁴ Driver aims to construct an account of virtue that allows for the possibility that, in some circumstances, a good character trait will prevent right action. As Driver argues, some indirect versions of virtue consequentialism deny this possibility. Indirect virtue consequentialism suggests that right actions are those that flow from virtues, and an inability to kill innocents is a virtue, as that character trait usually promotes good consequences. Hence, on this view, the unwilling sheriff who refuses to kill the scapegoat performs the right action and is virtuous.Footnote ¹⁵

How can Driver, from within her consequentialist account of virtue, allow that there is a clash between good character and right action in the scapegoat case? She advocates a direct version of consequentialism, where actions are right if they promote the good, but also where traits are virtues if they promote the good. What would this lead us to say of the sheriff? According to Driver, it is plausible that the right action will be murdering the scapegoat, but also, that the unwilling sheriff's disposition not to murder innocents would be a virtue. How could a disposition not to murder innocents be a virtue, if, in the scapegoat case, it leads the sheriff to perform the wrong action? Driver's solution is to reject the claim that virtues are only those traits that maximize good consequences, and to fall back to the weaker view that a trait counts as a virtue if it usually produces good consequences.

We get the requisite conflict in the scapegoat case, Driver concludes, by judging that it is right for the sheriff to kill the scapegoat, but admitting that a strong aversion to killing innocents in any situation is a virtue – albeit, perhaps, not the best possible virtue – since that aversion usually produces good consequences. Hence, Driver thinks that her version of consequentialism about virtue accounts for point (6).

Another reason for favouring a consequentialist account of virtue, as Driver argues in point (7), is that a trait can lose its status as a virtue. Sometimes we might judge that a once-admired trait – piety, for instance – never really was a virtue to begin with. Yet in the case of chastity, Driver argues, what really was a virtue is no longer. How can we explain such a change in status? It is plausible that chastity once promoted good consequences, but, due to social and environmental changes, it no longer does so. For this reason, Driver concludes, we no longer judge that trait to be a virtue.Footnote ¹⁷ If this reasoning generalizes, as Driver thinks it does, it supports a consequentialist account of virtue.

So far, Driver seems to have mounted a strong defence of her consequentialist account of virtue. In point (8), though, she admits that many of us have strong intuitions that push us towards valuing certain character traits in spite of their failure to promote the good. Driver notes:

Given the weight of evidence in favour of her account, Driver argues that we should override these intuitions. After all, she suggests, they are merely intuitions, and once we adopt the critical perspective of assessing our intuitive reactions, we see that ‘virtues’ that do not promote the good are not really valuable after all.Footnote ¹⁹ Once we have reflected critically, our subjective attitudes regarding virtues will catch up with our objective perceptions about consequentialist value.Footnote ²⁰

Driver has argued that points (1)–(8) are best accounted for by her consequentialist theory of virtue. In the next section, I will set out an alternative account – disjunctive minimalism – and begin to argue that this account does a better job of explaining these intuitions than does Driver's view.

IV. DISJUNCTIVE MINIMALISM ABOUT VIRTUE

All of the various accounts of virtue under consideration in this article agree with the basic claim that virtues are valuable or good character traits. Minimalism about virtue is the view that, in giving a general definition of virtue, we should not add further psychological conditions to that basic claim; all we can say in general is that virtues are valuable or good character traits.Footnote ²¹ Minimalism about virtue is compatible with the claim that some particular virtue – say, courage, or generosity – requires a psychologically rich set of conditions. Minimalism is the view that not all virtues require such conditions.Footnote ²² Aristotle, Hume and Kant clearly are not minimalists about virtue, but Driver is a kind of minimalist. Exactly which kind of minimalist is Driver?

Generally, there are two ways in which something can have value: it can be one of the intrinsic goods, which count as good in and of themselves, or it can be an instrumentally valuable means to securing what is intrinsically good. This distinction between types of value generates further distinctions between types of virtue theory. Virtue theorists must endorse one of the following options:

1. • Intrinsicalism: virtues are character traits that are intrinsically valuable/good.

2. • Instrumentalism: virtues are character traits that are instrumentally valuable/good.

3. • Disjunctivism: virtues are character traits that are either intrinsically or instrumentally valuable/good.

Driver's ‘consequentialism about virtue’ includes a commitment not just to psychological minimalism, but also to instrumentalism. She argues that character traits are virtues because they produce intrinsically good effects, and claims that traits which cease to produce such effects lose their status as virtues.Footnote ²³ So, Driver is an ‘instrumental minimalist’ about virtue. In contrast, Thomas Hurka offers an intrinsicalist account of virtue, explicitly considering and then rejecting the view that merely instrumentally valuable character traits might count as virtues.Footnote ²⁴ (Hurka is not an intrinsic minimalist, though, because he is not a minimalist. He maintains that all virtues have a common psychological structure: they consist in loving that which is intrinsically good and hating that which is intrinsically bad.) In the remainder of this article, my aim is to argue that the third of these options – disjunctivism – produces the most plausible minimalist account of virtue.

The first attractive feature of disjunctive minimalism is that it is compatible with both consequentialism and non-consequentialism. Not all consequentialists must be ‘consequentialists about virtue’, in Driver's sense. Consequentialism is the view that we should act so as to promote or maximize the occurrence of the things that are intrinsically good (plus, perhaps, the condition that the goods are agent-neutral).Footnote ²⁵ Many intrinsicalists and disjunctivists will reject consequentialism, maintaining that we must honour or embody the intrinsically valuable virtues, rather than act so as to maximize the presence of good things in the world. But neither intrinsicalism nor disjunctivism commits one to non-consequentialism. Intrinsic and disjunctive minimalists might recommend that we act so as to maximize the occurrence of the intrinsically good things, including the intrinsically valuable virtues. Hurka is one clear example of a consequentialist who is not an instrumentalist about virtue. Notably, utilitarian consequentialists cannot be intrinsic or disjunctive minimalists, because utilitarians limit the intrinsic goods to positive states of mind, such as pleasure, happiness or desire satisfaction, thus excluding virtues from the class of intrinsic goods. For the purposes of this article, I shall assume that utilitarianism is less plausible than richer forms of consequentialism, as has been well argued elsewhere. One of the attractive features of disjunctive minimalism, then, is that it is compatible with the most plausible forms of consequentialism and with non-consequentialism.

The compatibility of disjunctive minimalism with non-consequentialism might not seem particularly appealing to committed consequentialists. Nonetheless, there are other reasons why consequentialists, including Driver, should accept disjunctive minimalism. Let us return to points (1)–(8), which Driver takes to support her instrumental minimalism. Clearly, disjunctive minimalism holds many of the advantages of Driver's own view, simply because it too is psychologically minimalist. Like Driver, but unlike, Aristotle, Hume and Kant, disjunctive minimalists can account for points (1)–(4). People can possess admirable character traits that require ignorance, or at least are compatible with ignorance. Huck Finn's sympathy can count as good even though he judges it to be bad, either because his sympathy is intrinsically good, or because it produces good effects, or both. Huck himself can count as virtuous even though he is internally conflicted, does not enjoy his virtuous action, and performs that action out of inclination. Moreover, people and whole societies can mistakenly view a virtue with displeasure, if they mistakenly think that it is neither intrinsically nor instrumentally valuable. Disjunctive minimalism is at least as flexible as Driver's view in allowing for psychological variety in the traits that count as virtues.

Disjunctive minimalism also respects point (5). The deep disagreement between Aristotelians and Kantians over the psychological conditions required for virtue, Driver suggests, is the result of overgeneralizing the features of examples of particular virtues. According to disjunctive minimalists, Driver is guilty of overgeneralizing as well. She moves from the observation that some virtues have merely instrumental value, and hence lose their status as virtues once they lose their typical good effects, to the general claim that all virtues are of this type. Disjunctive minimalists maintain that some virtues might remain as virtues even if they regularly fail to promote good consequences, because they are intrinsically valuable character traits.

Disjunctive minimalists’ refusal to generalize the features of virtues will seem worrying to some. Whilst Driver ditched the psychological conditions held to be necessary for virtue by Aristotle, Hume and Kant, at least she pointed to some other shared feature which explains why seemingly disparate character traits all count as virtues. Notably, Driver proposed the consequentialist account as the more preferable of Hume's two proto-theories of virtue.Footnote ²⁶ When we approach the matter through Hume, we seem to face a choice between identifying the virtues via projectivism, in which case we forfeit objective realism, or identifying the virtues via their consequences. Yet Hume's two alternatives are not the only alternatives available. Despite rejecting Driver's definition of virtues, disjunctive minimalists do not have to choose projectivism, and claim that whichever traits are admired or valued count as valuable, and hence as virtues. Rather, they claim that traits which are admirable and valuable are virtues, and they can make this as realist as they like. Disjunctive minimalists think there is no common feature that must be shared by all virtues other than their goodness and their status as character traits. Critics may argue that, by ascribing intrinsic value to some character traits, disjunctive minimalists are trafficking in mystery. Yet every moral realist must ascribe intrinsic value to some things. It seems no more mysterious that a kind of character trait might be good in and of itself than that knowledge or beauty or an untouched wilderness might be intrinsically valuable. As ‘objective list’ consequentialists attest, it is perfectly respectable to defend a pluralistic account of the things that are intrinsically good.Footnote ²⁷

Recall, though, that Driver does not adopt instrumental minimalism simply in order to avoid projectivism. She notes in point (7) that some virtues like chastity cease to count as virtues when they cease to produce good effects. Does point (7) give us a reason to favour instrumental minimalism over disjunctive minimalism? It strikes me as plausible that chastity was a virtue and that, given changes in medical technology, contraception and social attitudes, it is not a virtue in our current society. This example does pose a problem for intrinsic minimalists. They would have to say either that chastity still is a virtue, or never was a virtue, or that chastity has somehow ceased to be intrinsically good. Yet disjunctive minimalism can account for this change in the status of chastity just as easily as can Driver. Disjunctive minimalists judge that some virtues – presumably including chastity – count as virtues only because they are instrumentally good, and hence lose their status if they no longer promote good consequences. Driver's conclusion is that a similar transformation would occur for all current virtues: if a virtue ceases to promote the good, then it is no longer a virtue. In contrast, disjunctive minimalism suggests that some virtues would remain virtues despite not promoting good consequences. They remain virtues because they are intrinsically valuable. So, if we were to favour Driver's broad conclusion, we would need much more evidence than the case of chastity.

Something that looks like evidence is offered in Driver's discussion of modesty and blind charity, both of which she defends as virtues that require ignorance. Why are they virtues? According to Driver, one sign is that they are ‘pleasing to a high degree’ both to herself and to various characters in novels.Footnote ²⁸ More fully:

But this explanation fits neatly with disjunctive or intrinsic minimalism. Perhaps these traits seem to be virtues simply because they are intrinsically admirable. To avoid disjunctive or intrinsic minimalism, Driver must argue that they are virtues only in so far as they promote good consequences. This she does, by inventing stories according to which modesty promotes happiness by limiting our tendency to rank ourselves relative to others,Footnote ³⁰ and blind charity brings happiness to others (or at least, is reported to do so by characters in novels).Footnote ³¹

These stories sound plausible, but if Driver really believes her own account of virtue, we should expect her to take a very different approach. Driver should be highly uncertain as to whether modesty and blind charity are virtues, as their actual beneficial effects are unclear.Footnote ³² We can invent equally convincing stories with opposite conclusions. Perhaps modesty lowers overall happiness in virtually any society because it prevents gifted people from taking on difficult projects of which they modestly believe themselves to be incapable. Even more plausibly, it could be that blind charity lowers overall happiness in virtually any society because it leads people to be insufficiently wary of dangerous criminals and con-artists. Which stories are true? Disjunctive minimalists think we might not need to answer this question in order to defend the claim that modesty and blind charity are virtues. If these character traits are intrinsically valuable, they count as virtues regardless. But for instrumental minimalists, the question must be answered before we know whether modesty and blind charity really are virtues.

Driver's easy confidence in the value of modesty and blind charity, and her use of fiction as a source of vindication, sit very uneasily with the implications of her own account of virtue. I am not suggesting that novels are an inappropriate source of insight into admirable character. On the contrary, in fiction we find many fine-grained and insightful dissections and evaluations of motive, virtue and vice. What we do not find are empirically grounded claims about the effects of certain character traits on overall happiness in specific environments. But it is precisely this, and only this, that matters for judgements about virtue, according to Driver's instrumental minimalism. Instead of pursuing this knowledge, I suspect that Driver relies on intuitive assessments of the intrinsic worth of character traits, and trusts that the empirical data will fall into line. Of course, this is an ad hominem criticism. One could agree with Driver's account of virtue and argue that she ought to be much more careful in basing her judgements about virtues on empirical assessments of the consequences of character traits. Still, it is revealing that even advocates of instrumental minimalism find it natural to defend the status of some virtues simply in terms of what strikes us as pleasing in moral character. This counts as evidence, albeit weak, in favour of disjunctivism or intrinsicalism.

If we are to favour disjunctive minimalism, we need a stronger reason for thinking that some virtues are virtues because they are intrinsically good. Such a reason becomes apparent when we consider what we would do if a virtue that once promoted good consequences ceased to do so. In some cases, such as chastity, I think we are happy to change our attitudes towards the trait in question. Most of us are not tempted to change the world so that a lack of chastity has the often dire consequences that it once had. Yet, in other cases, this is precisely what we would strive to do. Suppose that honesty or sympathy have ceased to promote good consequences, perhaps because honest and sympathetic people are vulnerable to exploitation, and the likelihood of exploitation has risen dramatically. If instrumental minimalism were true, then we should be happy to say that honesty and sympathy simply are no longer good; so long as overall happiness remains the same in a world filled with defensive dishonesty and lack of sympathy, there is nothing better or worse about the world after this change. But would this really be our attitude? I think that we would fight to recreate a world in which dishonesty and lack of sympathy were costly, and honesty and sympathy could be expressed without cost. There is something better about this world. It contains more good things; namely, more honesty and sympathy. When we can, we structure society so that the intrinsically valuable virtues are allowed to flourish and intrinsically disvaluable vices are not, and this is compatible with disjunctive minimalism but not with Driver's instrumental minimalism.

Disjunctive minimalists’ insistence that some virtues are intrinsic and not instrumental goods also allows them to make sense of the conflict referred to in point (6) in a way that Driver cannot. Recall that Driver explains our conflicting intuitions in the case of the ‘unwilling sheriff’ who refuses to kill the scapegoat by claiming that his deeply ingrained aversion to killing innocents is a virtue, because it raises (but does not maximize) the level of good in the world, but that his refusal to kill the scapegoat is wrong. I think that this explanation does not adequately capture the depth of the conflict. After all, if Driver agrees that killing the scapegoat would maximize the good, then she should admit that an ‘aversion to killing innocents except in scapegoat cases’ would count as a better virtue than a simple aversion to killing innocents.Footnote ³³ Although she can claim that the unwilling sheriff did the wrong thing because he was a good person, what can Driver say about a ‘cautious but willing’ sheriff, who has an aversion to killing except in scapegoat cases, and who actually kills this scapegoat? Driver must say that the cautious but willing sheriff did the right thing, and did it because he was a good person – a better, more virtuous person, than the unwilling sheriff. Hence, it is both more virtuous and right to kill the scapegoat. If Driver's theory of virtue is correct, then, when considering the case of the cautious but willing sheriff, our intuitions should not be conflicted. Yet, I contend, our intuitions are no less conflicted in this case. For instance, even if we agree that the cautious but willing sheriff did the right thing, we still judge that he has done something terrible, and that he ought to feel guilt, or what Bernard Williams calls ‘agent regret’.Footnote ³⁴

Unlike Driver, consequentialists who are disjunctive minimalists can account for conflicted intuitions even in the case of the cautious but willing sheriff. Suppose that these consequentialists agree that the cautious but willing sheriff's action maximizes good consequences, and hence is the right action. Also, suppose that his character trait of killing innocents only in scapegoat cases maximizes good consequences, and hence is a virtue, because it is instrumentally valuable. But suppose that this character trait, while instrumentally virtuous, is intrinsically bad. Perhaps there is something intrinsically horrible about the person who can kill scapegoats, even when it is right to do so. Is the willing sheriff's character trait a virtue? Disjunctive minimalists say that it is both a virtue and a vice; an instrumental virtue and an intrinsic vice. Admittedly, this is very messy. But it is messy in the exactly the right kind of way, I suggest, to match our conflicted intuitions about these cases. As I shall argue later, we can live with this kind of messiness about virtue, so long as we place it within an appropriate theory of right action. For now, my claim is that, unlike Driver, disjunctive minimalists can make sense of the fact that our intuitions are conflicted even in the case of the cautious but willing sheriff.

The conflict in the cases mentioned in point (6) is best understood simply as that conflict which prompts the dispute between consequentialists and non-consequentialists. Sometimes it seems we can maximize good things by failing to be maximally good ourselves. Should we be less good than we could be in order to promote good, or should we be as good as is possible, even if this fails to promote good? Driver's instrumental minimalism about assessments of character denies us this natural formulation of the conflict, since her view implies that we have a maximally good character only in so far as we maximize good consequences. Even malicious intent to harm others could, in bizarre environments in which that trait reliably promoted happiness, count as a virtue and not a vice on this view. To be a good person, for Driver, is simply reliably to promote good. In contrast, disjunctive minimalism allows us to express the conflict in its most convincing form: should we violate intrinsically valuable virtues so as to promote good consequences, or should we be intrinsically virtuous even when this fails to promote the good? Disjunctivism doesn't commit us to either of these answers – some disjunctivists are consequentialists, others are non-consequentialists – but at least it allows us to recognize the very real conflict in such cases. It allows us to say that there is something bad about malicious people, even in worlds in which malice promotes good consequences.

I have argued that points (5)–(7) do not give us a reason to favour Driver's instrumental minimalism over disjunctive minimalism. What of point (8): that we have strong and persistent intuitions that some virtues are virtues even though they do not promote the good, and these intuitions often carry over into a practical refusal to neglect those virtues? Prima facie, these intuitions fit neatly with the disjunctivist claim that some virtues are intrinsically valuable, and hence there is something bad – perhaps prohibitively bad – about failing to exhibit those virtues, even in cases when such a failure would actually promote the good. These intuitions do not fit at all well with Driver's instrumentalism. Driver could respond to this problem by claiming that she offers a normative theory of the virtues: she is telling us how people ought to think about the virtues, not how people in fact think about the virtues, and hence her account is not undermined by its failure to align with everyday intuition and practice. Yet, elsewhere, Driver takes ordinary moral practice to count as evidence in favour of her theory, as when she notes the changes in our attitude towards chastity.Footnote ³⁵ Hence, Driver needs to explain why the intuitions and practice that accord better with disjunctivism should be overridden, whereas the intuitions and practices according merely instrumental value to certain virtues ought to be respected and generalized.

The explanation Driver offers is unconvincing. She maintains that judgements that virtues have intrinsic value are based on mere intuitions or subjective attitudes that do not survive critical scrutiny.Footnote ³⁶ It seems that Driver assumes the only way that a prima facie virtue can withstand critical scrutiny is if it is shown to be ‘valuable objectively (because it produces good consequences)’.Footnote ³⁷ But this test cannot be valid in general for critical reflection on what is valuable. If anything passes Driver's test, then something must be intrinsically good, for we cannot have an endless regress of merely instrumental goods. Furthermore, there is no reason to suppose that everything which is intrinsically good is also instrumentally good. Hence, it is possible that there are objective goods that are merely intrinsically good, even though these would not pass Driver's test. Disjunctivists suggest that, just as happiness or desire satisfaction count as good in and of themselves, so too do some of the virtues. Disjunctivists argue for this view by resting on intuitions as to what is valuable, but no defence of intrinsic value can do otherwise. We might have reason to override widespread intuitions about what is intrinsically valuable if those intuitions are based on some kind of ignorance or irrationality, or if they fail a test of coherence. Driver has not shown that valuing character traits intrinsically is the product of ignorance or irrationality. Moreover, as I have suggested, the degree of tension and internal conflict allowed by disjunctive minimalism accords with that in our moral intuitions – including the intuitions in (8) – more closely than does Driver's instrumental minimalism.

Still, cohering with our tangled moral intuitions might not be sufficient justification for favouring disjunctivism. Perhaps those intuitions themselves are so tangled as to be inconsistent with each other. In such cases, a theory that fully respects our moral intuitions will be incoherent itself. The final challenge in defending disjunctive minimalism is showing that it can fit into a sufficiently coherent general ethical theory.

V. DISJUNCTIVE MINIMALISM AND RIGHT ACTION

There are three reasons why those who agree that our moral intuitions support disjunctivism might reject disjunctivism nonetheless. First, because disjunctivism does not map the class of virtues onto a sufficiently neat natural kind; second, because it implies that a single trait could be both a virtue and a vice; and finally, because it is not clear that disjunctivism gives us a coherent recommendation as to how we ought to act. In this final section, I shall argue that these problems are not reasons to reject disjunctive minimalism.

One aim of philosophy is to carve reality at the joints, but another aim is to elucidate our everyday concepts. When an everyday concept fails to carve reality at the joints, philosophers face a choice: either revise the everyday concept into a more appropriate form, or conserve the boundaries of the everyday concept whilst noting its failings. The revisionary option seems more appropriate when the class picked out by the everyday concept contains a natural kind plus some radically different extras that belong to another natural kind. In such cases, the concept can be drawn into line with natural kinds whilst preserving most of its everyday extension. The concept of ‘fish’ was revised, quite appropriately, to exclude whales and dolphins on these grounds. The conservative option seems more appropriate when the everyday class wholly contains more than one natural kind. Here, excluding one of these subclasses would radically alter the extension of the concept. For instance, it is appropriate that we still use the concept of ‘jade’ to refer to the distinct natural kinds of jadeite and nephrite, because neither one of the natural kinds has a better claim really to be jade, and because excluding either one would be an arbitrary and jarring alteration to the extension of the concept.

I have argued that the everyday concept of virtue refers to any valuable character trait, but that this class fully includes two distinct but partially overlapping subclasses: intrinsic virtues and instrumental virtues. Both intrinsicalists and instrumentalists revise the concept of virtue to match what admittedly are important natural kinds within ethics, each of which slots into ethical theory in a different way. What makes the revisionist option even more tempting in this case is the high probability that the subclasses substantially overlap – most intrinsic virtues probably are also instrumental virtues in common environments – so excluding either subclass from the whole preserves most of the original extension.

Yet the fact that we could revise the concept of virtue so that it maps onto an important natural kind whilst preserving most of its extension does not show that we should do so. I have suggested that we ought to take the conservative option in the case of virtue. We ought to conserve the disjunctive extension because we have no reason to favour one of the subclasses over the other. Revisionists could go for intrinsicalism, and say that chastity never was a virtue, or they could advocate instrumentalism, and maintain that honesty is no longer a virtue if it ceases to promote good consequences, but the choice between these two options seems completely arbitrary. The concept of ‘virtue’ is analogous to the everyday concept of ‘parent’, which refers disjunctively to the partially overlapping subclasses of ‘biological parents’ and something like ‘primary caregivers with guardianship of the child’. This might seem deeply problematic, as someone could be a ‘biological parent’ but not a ‘guardian parent’. To escape this problem, we could choose to revise the concept of parent to fit with one of the natural kinds in its everyday extension, but which? With ‘virtue’, as with ‘parent’, we should stick with the everyday disjunctive extension, whilst noting that it contains two importantly distinct natural kinds with substantial overlap in their extensions. And we should not forget that all members of the disjunctive set of instrumental and intrinsic virtues have something in common: they are all good character traits.

Yet there is an extra problem with a disjunctive conception of virtue. If we accept disjunctivism, we must judge that a single trait could be both a virtue and a vice. Suppose that blind charity is intrinsically good but lowers overall happiness. Blind charity thus qualifies as an intrinsic virtue, which is enough to make it a virtue, full stop. But blind charity also qualifies as an instrumental vice, which is enough to make it a vice, full stop. So, even if disjunctivism fits best with our intuitions about which traits are virtues, it violates what we might call the ‘principle of mutual exclusivity’: that a trait cannot be both a virtue and a vice. Switching to Driver's instrumentalism (or, perhaps, to intrinsicalismFootnote ³⁸) would allow us to preserve the principle of mutual exclusivity. If we think that the principle of mutual exclusivity is more important than the intuitions supporting disjunctivism, perhaps we should follow Driver and revise the concept of virtue.

There are several possible responses to this problem, short of endorsing instrumentalism or intrinsicalism.Footnote ³⁹ For instance, we could claim that a virtue is a character trait that has more total value, either intrinsic or instrumental, than it has disvalue.Footnote ⁴⁰ According to this view – call it ‘totalizing disjunctivism’ – no trait is both a virtue and a vice, since no trait has both a positive and a negative total value. Suppose blind charity is intrinsically valuable and instrumentally disvaluable. In this case, totalizing disjunctivism implies that blind charity will be a virtue if its intrinsic value outweighs its instrumental disvalue, or a vice if its instrumental disvalue outweighs its intrinsic value. Hence, blind charity cannot be both a virtue and a vice at once. Perhaps such weighing of intrinsic against instrumental value will be problematic. Still, totalizing disjunctivism is an attractive view in many respects. If we have reason to retain the principle of mutual exclusivity, I think we should accept totalizing disjunctivism. However, I think we have no reason to preserve the principle of mutual exclusivity.

The minimalist definition of virtues is that virtues are good character traits, or valuable character traits. If we are austere in our minimalism, then we ought to endorse a principle of mutual exclusivity in relation to virtue only if a corresponding principle applies in relation to the concept of ‘good’ and the concept of ‘valuable’. Is it the case that something cannot be both good and bad, nor both valuable and disvaluable? Arguably, we ought to allow that something can be both good and bad, valuable and disvaluable, because there are two different ways something can be valuable – intrinsically or instrumentally – and it is plausible that something could be valuable in one way and disvaluable in the other. This should be accepted, I contend, even by those who are not sympathetic to a disjunctive account of virtue. Everyone must admit that some things could count as intrinsically good but instrumentally bad. For instance, utilitarians must admit that the public humiliation of a widely hated person is both intrinsically bad, because it is painful to the humiliated person, but instrumentally good, because that pain causes a larger amount of pleasure in others. So, mutual exclusivity cannot apply to good and bad, or to valuable and disvaluable. Since disjunctive minimalists wish merely to transfer this feature into their account of virtue, we would need a special reason to reject the disjunctive minimalists’ claim that a trait can be both a virtue and a vice.

Just such a special reason is offered by those who forge a tight link between virtue and right action. If a straightforward practical recommendation flows from the fact that a trait is a virtue, then calling a trait both a virtue and a vice is tantamount to recommending that one both possess that trait and not possess that trait. If disjunctivism forced us to give inconsistent recommendations, we would have a good reason to reject the view, or, at least, advocate the totalizing version.

A tight link between virtue and rightness is appealing to many. For instance, McDowell notes that virtue reliably leads to right action,Footnote ⁴¹ and Pettit identifies as a ‘widely shared belief about rightness’ the view that ‘The virtuous person is reliably disposed to recognise right options and to choose only such options.’Footnote ⁴² The tight link, though, could be explained in two different ways. For some, including Driver and Pettit, considerations independent of virtue determine what counts as right action, and then virtues are defined as whichever character traits (reliably) produce right action.Footnote ⁴³ This approach, however, denies intrinsic value to virtues, and hence does not threaten the principle of mutual exclusivity. What concerns us at this point is the combination of the view that at least some virtues are intrinsically good, and that it is right to be virtuous or perform virtuous actions. The combination of these views is, arguably, the definitive characteristic of ‘virtue ethics’.Footnote ⁴⁴ Slote defends this kind of virtue ethics, suggesting that we can simply derive judgements as to what is right action from our logically prior judgements about virtue:

But if we accept disjunctivism, we cannot simply derive claims about right action from the fact that a trait or action is virtuous, for, according to disjunctivists, a single trait could be both a virtue and a vice. It seems that virtue ethics does require the mutual exclusivity of virtue and vice, and hence does not fit with disjunctive minimalism.

Yet virtue ethics itself is not plausible. The idea that it is right to act out of a trait because that trait it is a virtue is very difficult to defend. A terrorist's action might be plainly wrong, for instance, even though it is courageous. A bureaucrat's action might be sympathetic yet unjust. Virtue ethicists go to great lengths to deny such claims. The usual strategy is to claim that the virtues are deeply unified; either, along with Aristotle, that one cannot possess a single virtue without possessing all of the virtues,Footnote ⁴⁶ or, along with McDowell, that all virtues are part of one complex sensitivity to value.Footnote ⁴⁷ The right action is the action that would be performed by the virtuous person,Footnote ⁴⁸ or is the action that is virtuous simpliciter, and, if the virtues are unified, any action flowing from any particular virtue will also be virtuous simpliciter. But the doctrine of the unity of the virtues is untrue to the introspective phenomenology of virtue, as well as to empirical psychological research into character traits.Footnote ⁴⁹ One can be honest without being generous, courageous without being patient, and modest without being friendly. In everyday life, to call an action generous, or courageous, or patient, or modest, or friendly is not necessarily to commend it as the right action. It is merely to say that the action flowed from a character trait that is good. Thus, there is no tight link between individual virtues and right action, and this means that there is no special reason for upholding the principle of mutual exclusivity. To call a trait both a virtue and a vice is not to issue an incoherent practical recommendation, because to call a trait a virtue or a vice is not to issue a practical recommendation at all.

Disjunctivism plus virtue ethics would result in incoherent recommendations, but, I have argued, we ought not to accept virtue ethics. What should disjunctivists say about the relationship between virtues and right action? Obviously, the fact that a trait is a virtue is not irrelevant to the question of how we should act. Virtues are good, and considerations of what is good are a crucial part of determining what is right. Yet, of course, there is much dispute between consequentialists and non-consequentialists over the relationship between the good and the right. Those who accept disjunctivism might take either side in this dispute. A non-consequentialist might claim that, although some traits are instrumental virtues, they ought not to be embodied, because they are incompatible with deeply important intrinsic virtues, or because acting on those traits would lead us to violate perfect duties. On the other hand, disjunctivists might argue that we should promote what is intrinsically good, including the intrinsically valuable character traits. Yet consequentialism implies that sometimes it is wrong to embody intrinsic virtues. Whichever way we go on this issue, there is no straightforward relationship between the good and the right. According to disjunctivists, though, the question of whether a trait is a virtue is settled by whether it is either intrinsically or instrumentally good; it is not a question about right behaviour.

Disjunctive minimalism is a relatively messy ethical theory. Since virtues are simply good character traits, and since something can be both instrumentally good and intrinsically bad (or vice versa), disjunctive minimalists say that a trait can be both a virtue and a vice. In contrast, Driver's instrumental minimalist account of virtue has the virtue of simplicity. Yet, to say that a theory has the virtue of simplicity is not to say that it is the right theory, and that you ought to believe it. It is just to say that there is something good about the theory. I have argued that disjunctive minimalism fits better with points (1)–(8) than does Driver's own account. In other words, disjunctive minimalism fits better with the way we talk about virtues in everyday life. Moreover, I have argued, disjunctive minimalism is not incoherent, nor so messy as to warrant philosophical revision. Even consequentialists should be disjunctive minimalists about the virtues.