Open Research Online To Live Outside the Law You Must Be Honest

( valence R and ( reasons behave like R . For my someone who that nearly all reasons are almost completely invariant (though none is 100 % invariant) more like a generalist than someone who believes that just a handful of reasons are 100 % invariant but nearly all other reasons are very variable indeed. Yet Dancy’s definitions the first person a particularist and the second a generalist.

Section i states Sandis's view in his paper (particularism for actions, generalism for dispositions); Section ii describes and begins to criticize Dancy-style particularism; Section iii applies these criticisms to Sandis's view; Section iv delineates an alternative view (my own) about actions, dispositions, and the particularism/generalism debate; Section v raises and considers a further puzzle, about how in general we should understand virtue ascriptions anyway. I Sandis's View. Constantine Sandis (2021) argues both for a generalist virtue ethic about character traits, 'where these are understood as reliable dispositions to feel and act in certain ways', and for a moral particularism about actions. His generalistic virtue ethics gives us action-relevant instructions of the form 'Be kind', which we may rewrite more fully as 'Have the ceteris paribus disposition to act kindly as and when the situation calls for it'. But then, 'Be kind' doesn't V C 2021 The Aristotelian Society This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited. reduce to 'Act kindly', at least not if that means 'Always do the kind thing'. Thus generalism at the level of virtues doesn't translate into generalism at the level of action, but into particularism: 'While moral thought is ultimately hindered by principles concerning which actions we should perform, it may nevertheless be aided by principles concerning which character traits we ought to possess. ' So, for instance, 'being honest is compatible with lying on certain occasions . . . an honest person may-and perhaps sometimes even must-tell a lie. . . . That a particular linguistic act may be all the better for being an instance of lying is . . . not incompatible with the thought that honesty is a virtue' (Sandis 2021, pp. 000). And it is a mistake to think that so to speak bottom-up identifications of actions as (for example) 'the honest thing' will get us to a class of actions that has any automatic moral status; if moral generalism is committed to such bottom-up identifications, then moral generalism is mistaken. To understand whether an action is in accordance or not with (for example) the virtue of honesty, we need to understand not whether it is 'the honest thing'-which may well be irrelevantbut whether it is 'what the situation calls for'. And that is not something that can be determined bottom-up: particularism is already built into the holistic notion of 'what the situation calls for', in a way that it is not already built into the atomistic notion of 'doing the honest thing' or 'acting honestly'.

II
What Is Particularism? What, then, is moral particularism, and what is its opposite, moral generalism? 1 Jonathan Dancy's answerand I take it that Dancy's particularism is close to canonical-is that moral particularism says that moral judgement, rationality, deliberation, and explanation are necessarily particular, not general. The particularist denies the existence of moral principles, or says that we can do normative ethics without them, or both; whereas moral generalism asserts the existence of moral principles, or says that we cannot do normative ethics without them, or both. 'The rationality of moral thought and judgement in no way depends on a suitable provision of' moral principles, says Dancy in his most recent setpiece exposition of the view (Dancy 2017).
So next, what are moral principles? And how are the moral principles that are at the heart of generalism distinct from the items that are central to Dancy's own particularism, moral reasons? The answer is that moral principles have invariable valence. Moral principles can be 'absolute' or they can be 'contributory' (unlike Dancy's moral reasons, which are only contributory). An absolute moral principle will say, for instance, 'Necessarily, 2 each and every action of breaking a promise is a wrong action'; a contributory moral principle will say, for instance, 'Necessarily, if an action involves breaking a promise, that counts against it'-meaning always and essentially counts against it. But either way, the distinctive feature of a moral principle is this necessary universal invariability, which is supposed to be what makes it explanatorily potent. That something is a promise-breaking is always, according to Dancy's generalist, some kind of strike against it. And it will count as a decisive strike against it if either (a) the generalist holds an absolute moral principle against promise-breaking, or (b) we have a case where the negative contributory reason of its being a promise-breaking is not outweighed by other contributory reasons.
What is notable about moral generalism so defined is how few of us hold it. 'Always' is a very big word. It takes a lot of anti-sceptical confidence to hold out for any absolute moral principles-for the claim that there are any substantive, non-question-begging reasons such that every possible action on that reason is a wrong action because it is on that reason. It does not take much less confidence to hold out even for contributory moral principles-for the claim that there are any substantive, non-question-begging reasons such that it inevitably counts against 3 every possible action on that reason, that it is on that reason. There are some rather familiar and not very edifying cottage industries in this area ('How many babies should you 2 For reasons too complicated to explain fully here, I do not accept Vä yrynen's (2004) characterization of the debate. But I have learned much from talking the issues through with Pekka, and no doubt have plenty more to learn yet. My thanks to him, and my apologies for not working out how to do better justice here to his important work on the topic. 3 Or for. To keep things simple, I leave this out of the main text. But it is an interesting and open question whether we can expect variable but usually positive valence to vary in the same way(s) as variable but usually negative valence. torture to prevent nuclear holocaust, and how much?', and so on). And after all, it is so very easy to come up with examples of variable valence. Jonathan Dancy himself is, naturally, a dab hand at such examples; and here is one of my own. Satine is in love with Christian; so for her, it normally counts in favour of an action that it is a loving action towards Christian. However, Satine now knows, first, that she herself is dying fast, and secondly, that if Christian doesn't leave the Moulin Rouge he will very soon be killed by the Duke's goon Warner. So as Zidler tells her, what Satine now urgently needs to do-out of love for Christian-is 'to hurt him to save him': to convince Christian that she does not love him, so that he will leave the Moulin Rouge at once and escape being murdered by Warner. Hence for Satine, being 'a love-displaying action towards Christian' now counts against performing that action, where before it counted very strongly in favour; and being 'a contemptand hate-displaying action towards Christian' now counts in favour, where before it counted very strongly against.
If all we need to get moral particularism off the ground, and to stymie moral generalism, is a list of cases like this as inductive support for the general claim that no reason always has the same valence, then it doesn't look very hard to do. Maybe a few philosophers will have the anti-sceptical confidence to hold out for the 'contributory principles' version of generalism, insisting that there are some action descriptions that always count at least to some degree against doing any action to which they apply. For example, I would, with 'torture' and 'rape' as instances. (I apologize for using these rather horrible examples; they or something equally unpleasant are regrettably necessary for my story.) Maybe some will even hold out for 'absolute principles' generalism, and insist that there are some action descriptions that always count decisively against: for example, I would, with 'torture for fun' and 'uncoerced rape' as instances.
However, let's leave aside my possibly over-confident, and among philosophers somewhat unusual, belief in principles both absolute and contributory. I want to revert to two words that I used two paragraphs back: 'so defined'. Suppose we agree that we are (except for the absolutists among us) moral particularists and not moral generalists as Dancy defines the terms, because we don't accept that there are any moral principles in Dancy's sense: there are no action descriptions of completely invariable valence. Still, mightn't there be action descriptions of largely invariable valence? Action descriptions such that, even if we can't say they always have the same valence, they nearly always do?
To say this is quite compatible with at any rate the letter of Dancy's definition of particularism. Yet it means we can have a normative ethics of extremely generalist spirit, even though it is officially a particularist ethics. Explanations of our reasons will make no appeal at all to principles either absolute or contributory. There will be nothing in our explanations that depends on any reason being always of the same valence, and so a principle. Indeed, it seems very likely that typical deliberators simply won't know whether any reason to which they appeal is always of the same valence; which is to say that they won't know whether that reason is just a reason or a principle as well. Dancy is surely right that it will normally make not the slightest difference to their deliberations, nor to the explanatory value of the reason in the case, that they don't know this. As above, always is a very big word, and it would be a strange kind of scepticism to insist that I cannot deploy a reason in moral deliberation or explanation unless and until I know exactly how that reason behaves in absolutely every possible bizarre circumstance.
Even in the absence of generalism, our explanations may appeal to reasons that (so far as we can tell) are nearly always of the same valence. They can include clauses that go something like 'This is the right thing to do in C because there is a reason R that applies to cases like C unless there are some exceptional circumstances; and there aren't any'. As far as I can see, nothing in Dancy's particularism stops us talking in this way of 'default reasons', as he himself calls them, in discussing a proposal of Mark Lance and Maggie Little's that is rather similar to my line here (Dancy 2004, pp. 111-17;cf. Lance and Little 2008). We can agree with Dancy that there are no reasons with invariable valences, yet also talk of reasons as having typical or usual or default valences, and of the frequencies with which these valences occur.
Notice here that, as Lance and Little (2008) emphasize, a default valence is not necessarily a statistically usual valence. Default valences can be like natural tendencies in Aristotelian science. Fish eggs naturally tend to turn into fish-that is the 'default setting' for fish eggs. But that does not mean that fish eggs usually or typically turn into fish: most of them get eaten by predators. As far as the frequencies go, we might even try to identify some, at least roughly. For example, the fact that an action is rude has a good deal more than an even chance of telling against it: rudeness is sometimes appropriate, but usually a bad idea. (As one of my favourite schoolteachers used to say, 'A gentleman is someone who is never rude unintentionally.') Likewise the fact that something is an act of pushing a sharp steel blade deep into someone else's lumbar region: normally a terrible idea, but sometimes not only permissible but required. (I know an academic whose kidney was surgically operated on by a member of his own university's medical faculty; he likes to say that he is the only philosophy professor he knows of who really has been stabbed in the back by a colleague.) So again with more well-worn examples like promise-breaking and lying: normally we shouldn't do such things, but sometimes we may or, as Constantine says, perhaps even must. (For myself, I must confess I have never quite understood the level of philosophical hand-wringing that, at least since the time of Ross, broken promises have occasioned. As for lying, I can't see why anyone has ever thought that people's undeniable general right that I not deceive them, or presume to control their access to correct information, can never be abridged: for example by the circumstance that they are murderous Nazis.) With funniness as a feature of actions, things are rather different: funniness is a good thing, but clearly not even half of all the things I do should be funny, even if I am a professional comedian. And with, say, emphaticness, things are different again: though certainly emphaticness can be a good-making feature of actions, and also a bad-maker, in probably the majority of cases it is morally valenced neither negatively nor positively, but neutral. 4 All this is possibleand all of it looks, at any rate to me, decidedly generalistic, and decidedly like what Dancy has sometimes called a 'subsumptive' explanation of our reasons. The only thing that Dancy's account of moral principles completely rules out is that we should assign any of these frequencies either 100% or 0%. But that is a pretty roomy restriction. As far as I can see, it leaves untouched nearly everything substantive that a generalist might want. 5 Indeed, we can even talk, in rather rule-consequentialist style, of what I'll call agnostic absolutes: of reasons that as far as we can see always make the same contribution to the valence of a possible action. We don't know for sure that there is no possible scenario, perhaps beyond our imaginative limits, in which some such reason R will be valenced the opposite way from usual. What we do know is that we can't see how R could be valenced the other way. For all practical purposes R might as well be a principle for us; but strictly speaking it isn't. Strictly speaking, we don't actually know whether R is a particularist reason or a generalist principle; so strictly speaking, we don't actually know whether we are particularists or generalists (in respect of R; or perhaps more generally). And in practice, it doesn't make any difference. To live outside the law you must, no doubt about it, be honest. But in a lot of cases it isn't actually that easy to tell, in all honesty, whether or not you are living outside the law.
One moral of these reflections is that, pace Hector Barbossa, mere 'guidelines' can be pretty strong determinants of the shape of practical reality even when they're not 'actual rules'. Another, no doubt, is that we try to do too much in ethics with \ and [, and not enough with less ambitious quantifiers like 'for the vast majority of cases', or 'for a small range of exceptions', or Aristotle's hws epi to polu, 'for the most part' (also mentioned in Lance and Little 2008). That a generalization has one exception is a proof in mathematical logic but, usually, not much of a proof of anything in ethics.
A second moral is, as I say, that going at least by the official definitions, far more of us are Dancy-particularists than realize it-and for all that, no worse equipped to provide generalistic and subsumptive explanations of the behaviour of moral reasons to act, with frequency levels for reasons' valences anywhere at all, provided they're in between 0% and 100%. 5 Dancy (2004, pp. 116-17) concludes a six-page discussion of Lance and Little's defence of default reasons by arguing that their view fails either to count as a form of generalism or to establish a stable middle ground between generalism and particularism. It does not count as generalism because it does not meet four conditions, all of which, briefly, are variations on the idea that generalism must be able to explain everything by reference to its own principles. This test seems to me decidedly over-demanding. But in any case, it doesn't undercut the interest of Lance and Little's view-never mind how we classify it. A third moral is that we should be suspicious of the appearance of explanatory anarchism that some expositions of particularism wear-the appearance that particularism 'flattens the landscape', as people sometimes say; that, for any case we encounter, nothing is settled about it until everything is settled about it, and that this is equally the case of every case we encounter. If my reflections in this section are correct, this anarchist air is a mirage. Or at least, it is a mirage unless and until particularism is restated in a more (as Dancy says) trenchant or (as I shall say) fervent form, with stronger reasons given for thinking something closer to what particularists are often supposed to think, even though I doubt they really do, or should, or even could: namely that, in advance of the careful examination of particular cases, the rationally expected frequency for any reason to act to have any given valence in any given case is more or less exactly 50%.

III
Constantine's Combination: Generalism for Traits, Particularism for Actions. The previous section might lead us to a distinction between, as I shall call them, nominal and fervent particularists. The fervent particularists are those who believe in particularism in some strong, perhaps landscape-flattening, maybe even anarchistic sense. The nominal particularists, by contrast, are those of us (though not in fact me: as I said, I am a generalist, I don't live outside the law, because I do believe in some invariant valences) who are particularists merely in the sense that we fit Dancy's specification for particularism in the way I have described. Nominal particularists don't believe that there are any principles, because they don't believe that there are any reasons to act, either contributory or overall, that have invariant valences in absolutely every possible case. Yet I've argued that one can be a nominal particularist and still access pretty well everything that generalism is really after. Above all, a nominal particularist can have nearly everything that there is to have in the way of generality, simply by defending reasons to act that are almost never reversible in their valence and so, as we might put it, are very nearly principles. For sure, the nominal particularist cannot have absolute generality of the kind expressed by [. But she can have 99.999. . .% of absolute generality. And that might well seem enough to be going on with. 6 No doubt, among particularists as among other kinds of believer, there is a spectrum from the nominal to the fervent. How fervent is Constantine? Clearly his paper gives some evidence of fervour: but in one area, and not in another. For he thinks that matters stand differently with our reasons to have virtues (or vices?) and with our reasons to act. In the slogan at the head of this section, his view is generalism for traits, particularism for actions. With actions, he suggests, principles get in the way, which they don't with virtues. So as I gloss him, following Dancy: Constantine believes that we have at least some reasons of absolutely invariant valence to have the virtues, or some virtues; but that all our reasons to act are variable in their valence.
Now if section ii is right, Constantine could have something like the contrast he seeks between action and disposition without being strictly speaking generalist for either action or disposition. He could just be more fervently particularist about actions and more nominally particularist about dispositions; indeed, he might even get a more marked contrast this way. Suppose, for example, that the frequency with which reasons to have the virtues are of variable valence is 15%, and the frequency with which reasons to act are of variable valence is 60%. This is (so to speak) a 45% contrast, and so a sharper contrast between the two kinds of reasons than a contrast between 0% variability for reasons to have the virtues, and 1% variability for reasons to act, even though, given Dancy's definitions, this latter contrast is a contrast between generalism and particularism, whereas the former contrast is only a contrast between a more and a less nominal particularism.
What, though, are 'reasons to have the virtues'? Two readings come to mind. Taken one way, 'We have reason to have the virtues' 6 So I can be an almost-generalist by being a nominal particularist but accepting that there are some reasons that are almost always invariant. The converse possibility is also worth noting: I can also be an almost-particularist, that is, someone who thinks that some few reasons are completely invariant, but nonetheless most reasons are very variable indeed. The degree to which anyone believes in the variable valence of reasons is a product of two quantities: (1) the degree of variable valence that they think any reason R displays, and (2) how many reasons they think behave like R. For my money, someone who believes that nearly all reasons are almost completely invariant (though none is 100% invariant) looks much more like a generalist than someone who believes that just a handful of reasons are 100% invariant but nearly all other reasons are very variable indeed. Yet Dancy's definitions make the first person a particularist and the second a generalist. Clearly Constantine can't think that generalism is true about our reasons to act so as to have the virtues, because that would contradict his particularism about actions. For my money, he's right not to. We do have reasons to act in ways that cultivate virtue, but to think that they're invariant seems implausible. My reasons to cultivate a virtue can easily conflict with my reasons to exercise that virtue, especially if the virtue involves self-forgetfulness, as in fact most virtues do. So 'that u-ing would be a cultivation of virtue V' can be both (i) a reason not to u, and (ii) not a reason to u, as well as (iii) a reason to u; indeed, we can have reasons not to act so as to cultivate V that are based in V itself. (Think of humility here. Or of the baffling injunction 'Be spontaneous'. Or, not entirely unrelatedly, of the difficulty of programming a computer to produce a random sequence.) On the other reading, 'We have invariant reasons to have the virtues' comes roughly to this: 'It is always good to have the virtues' or 'All people should always have the virtues'. In advance of further detail about what the virtues are, this might seem almost trivially true. Yet one way in which it can be disputed, and certainly seems to need at least qualifying, is already there in Plato's Meno: Meno. But Socrates, it's not hard to say what virtue is. If you want the virtue of a man first, that's easy. A man's virtue is to be man enough to run his city's affairs, and to run them so as to benefit his friends and harm his enemies-and make sure that no such harm ever comes to himself. Then if you want a woman's virtue, that's not hard to state either. What she has to do is keep house well, looking after the property and obeying her husband. A child's virtue is another thing, and it is different again depending on whether we mean a boy's virtue or a girl's. And there is a specific virtue for an old man, with further differences depending on whether he is a free old man or a slave. (Meno 71e, my own translation) Whether we are philosophers, historians or sociologists, this is a passage we ignore or skim-read to our cost, full-apart from anything else-of crucially interesting concrete social insights about Athens in about 390 bc. Meno tells us here that it is always good to have the virtues appropriate to your age and station, assuming any are.
(Perhaps for small babies no virtues at all are appropriate, and no vices inappropriate either, unless the Baby Jesus' 'no crying he makes' counts as a virtue.) It may follow that, for all people (who aren't small babies), there are some virtues that they should have. It by no means follows that, for some virtues, those are the virtues that all non-infants should have. Virtue, for Meno, is a time-indexed affair, and a gender-role-indexed and social-role-indexed affair too: what it is to be a good woman is different from what it is to be a good man, and what it is to be a good slave or freeman, or subject or ruler, or wife or husband, or old person or person-in-his-prime, are different again.
I hope it's obvious that I am not pursuing this solely out of nerdy antiquarianism. The point is that, here as so often elsewhere in the dialogues, Socrates' interlocutor at least hints at a really interesting and plausible idea to which Socrates does scant justice. The idea is that '(All) people should always have the virtues' turns out false, at least if that means 'the same virtues'. 'The virtues' is a name for a set of dispositions that are role-indexed in the kinds of ways that Meno indicates. So, for example, Achilles-like public glory-seeking fits a man in his prime, and eloquence like Nestor's fits a man who is old. But what befits a woman, as Pericles' Funeral Oration famously tells us, is obscurity and silence: If I really must bring to mind anything about women's virtue (gunaikeias ti aretês), given that some of you will now be widows, I will indicate all of it in this brief advice. Your great glory (megalê doxa) is to prove no worse than your inherent nature, and to have as little reputation (kleos) as possible among men, either for virtue or for reproach. (Thucydides, History of the Peloponnesian War, 2.45.2, my translation) So, for Thucydides' Pericles-and implicitly for Plato's Meno-the very same disposition can be a virtue in one person but a vice in another. That something is a disposition of glory-seeking, or eloquent loquacity, makes it a virtue in a man but a vice in a woman; with dispositions towards obscurity and silence, it's the other way around. For Pericles and Meno, in other words, that a trait is F is sometimes a reason to have it and sometimes a reason not to have it; indeed to have its opposite.
A more recent example of people displaying variable-valence dispositions needs a trigger warning: it involves horrible violence and murder. (The addendum to it in note 7 below is even worse.) It is Freddie Oversteegen and her sister Truus, a pair of teenage women who [in 1943] took up arms against Nazi occupiers and Dutch 'traitors' on the outskirts of Amsterdam . . . they seduced their targets in taverns or bars, asked if they wanted to 'go for a stroll' in the forest-and 'liquidated' them, as Ms. Oversteegen put it, with a pull of the trigger. 'We had to do it . . . It was a necessary evil, killing those who betrayed the good people.' When asked how many people she had killed or helped kill, she demurred: 'One should not ask a soldier any of that' . . . It was, she said, a source of pride and of pain-an experience that she never regretted, but that came to haunt her in peacetime . . . Ms. Oversteegen often spoke of the physics of killing-not the feel of the trigger or kick of the gun, but the inevitable collapse that followed, her victims' fall to the ground. 'Yes,' she told one interviewer . . . 'I've shot a gun myself and I've seen them fall. And what is inside us at such a moment? You want to help them get up.' (Smith 2018) In their role as resistance fighters, 'soldiers' as they called themselves, the Oversteegens needed not only different dispositions from those that they deployed in happier times; they actually needed opposite dispositions. And for them in their role of honey-trappers, coldheartedness, ruthlessness, murderousness, treacherousness, dishonesty, promiscuity, and a brutally instrumental attitude to their very temporary 'lovers' were the dispositions they needed. The kind of compassion that wants to help a gravely injured fellow human get up, and the remorse that can't sleep at night after cold-bloodedly murdering someone, even a Nazi: such dispositions were a distraction and an impediment. In sufficiently terrible circumstances, and within the confines of specific roles like 'resistance fighter', such stark and awful dispositions can be a kind of virtues, and their humane opposites can have either the opposite valence, that of vices, or at best a neutral valence. 7 So not only are there actions that have 7 In the same terrible year of 1943 when the Oversteegens took up arms, something like the same phenomenon of valence-switching dispositions was noticed from the other side, in an even grimmer way, by Heinrich Himmler: 'And then along they all come, all the 80 million upright Germans, and each one has his decent Jew . . . They say: all the others are swine, but here is a first-class Jew . . . None of them has seen it, has endured it. Most of you will know what it means when 100 bodies lie together, when 500 are there or when there are 1000 . . . to have seen this through and-with the exception of human weakness-to have remained decent, has made us hard and is a page of glory never mentioned and never to be mentioned . . . We have carried out this most difficult task for the love of our people. And

IV
An Alternative View (My Own). As we've seen, I am, unlike Constantine, a generalist about actions, because while I agree that there is a greater or lesser amount of valence-switching in the case of a lot of our reasons for action, I also think that there are some reasons for action, both contributory and absolute, that never switch valence. (Or minimally, never switch valence as far as I can see; call me an agnostic absolutist if you like.) My examples (with apologies for their grim and possibly triggering content) were torture/torture for fun and rape/uncoerced rape. And thinking a little more, if we can bear to, about the Dutch resistance case may help spell out why I don't think that our reasons against torture reverse valence even in a case like theirs. The Oversteegens' terrible actions were intended to have a symbolic and expressive power, to say to the Nazi occupiers something like 'So long as you persist in occupying our country, you will never be able to relax even for a moment; nothing is safe for you here; don't even try to sleep at night, or to go to the bar for a beer, or to look for sex; you are monsters, and we will never let up in our resistance to you, and you deserve to spend every moment that you occupy our country wondering whether you're about to be shot.' It was consistent with that expressive force to murder Nazi officers cleanly and quickly. But it would not have been consistent with it to torture them to death-for then the Oversteegens too would have been monsters. Even for them, torturing their victims would have made no sense.
As well as being a generalist about actions, am I also a particularist about dispositions, so that my position is the mirror image of Constantine's? Not unless I think that all virtues have variable valence. If there are any dispositions-if there is even one disposition-that it is good to have in any circumstances at all, no matter we have suffered no defect within us, in our soul, in our character.' (From Himmler's notorious Posen speech of 4 October 1943, https://www.facinghistory.org/holocaust-human-be havior/himmler-speech-posen-1943. Holocaust deniers sometimes pretend that there is no clear contemporary evidence of mass murder by the Nazis. This speech, for a start, is such evidence.) how squint or difficult or specialized, then generalism about dispositions is still true-at least of that disposition. Now, a large part of the Western tradition of ethics, from Socrates on, has been concerned to identify dispositions that are in this way invariant in valence, apparently with a view to defending both the importance of those invariant dispositions, and also generalism about dispositions. One of the central motivating thoughts of Socratic ethics is precisely that the only true virtue is the virtue that is always a virtue, in any and every conceivable circumstance; and since that turns out to be wisdom or knowledge or intelligence, 'virtue is knowledge'.
But this, it seems to me, is (and is obviously) a pretty dodgy inference. Despite what Socrates, Plato, Aristotle and Augustine are all inclined to think, the fact that some trait T is not a virtue in all possible circumstances, and that in some circumstances T is a vice, does not immediately show that T is not ('really') a virtue at all. Everything depends on which circumstances make T a vice, and which a virtue. From Thrasymachus and Protagoras to Hobbes, it has been argued often enough that justice is only a virtue within civil society, and in the state of nature either neutral or inapplicable or something more like a vice. If these authors are right, then our reasons to be just have reversible valence. But that might well not be a reason to say, 'So much the worse for justice'; it might rather be a reason to say, 'So much the better for civil society'. It is coherent to think that there are situations where justice is something that we have reason (maybe even moral reason) not to have; but that situations like that are terrible situations and it is infinitely better not to be in them, just as it is infinitely better to live in Amsterdam in peacetime than under the Nazi terror. ('If civilization has an opposite, it is war', Le Guin 1969, p. 52.) With other virtues, it might be that they are role-specific: they are not indexed to the circumstances of civil society, but to those of particular stations that we might occupy that bring with them particular duties. And this might be a matter of (pernicious) ideology and of oppression, as with the role-specific virtues for women, slaves and (perhaps) children that Pericles and Meno point us towards. But I see no reason why it must be. Within many kinds of organization, a rather authoritarian social structure is not just permissible, but necessary. Within such a structure, there are different role-virtues for 'officers' and for 'other ranks'. (The scare quotes indicate metaphor, but of course one application of my point is to the case of the military, where there is no metaphor and so no scare quotes.) Officers need the role-virtues that make them good at generating, questioning and refining strategies and plans; other ranks need the role-virtues that make them good at not questioning the officers' plans, but at enacting them efficiently, predictably, and without fuss or commentary or backchat. At least some of the officers' role-specific traits are virtues in them, but vices in the other ranks; and conversely.
Here, no doubt, we will feel again the temptation to say (1) that there will be some traits that are virtues both in officers and in other ranks, and (2) that just because they are the traits that are shared, (3) only these shared traits can really be virtues. Here too my point is that (1) is undeniably true, but that (1) does not get us to (3), and in particular, that (1) does not get us to (3) via (2). We can and should recognize the reality of traits that are sometimes virtues but sometimes vices; which is to say, of reasons to have the virtues that have reversible valence. To repeat, this isn't particularism about the virtues-though if you like, it is particularism about these virtues; but it isn't the claim that it is like this with all the virtues. But as we have seen, when it comes to reversible-valence reasons, the dividing line between particularism and generalism is not necessarily the place where all the interesting action is anyway.
So I don't believe that there are no completely invariant virtues, no dispositions whose moral valence never changes. Certainly some of the most important virtues-the most important dispositions relative to which we normally have reasons of positive valence-are in fact particularistic. This seems true of justice, which (as above) is a virtue in civil society, but either a neutral disposition or a vice in the state of nature-and as I say, so much the worse for the state of nature. The other three cardinal virtues, I suspect, are much more plausible candidates for being invariantly positive dispositions. Whether we are in civil society or the state of nature, it is still-as far as I can see-going to be a good thing to be self-controlled, brave, and intelligent. (Those too were important dispositions for the Oversteegen sisters; though if justice was important for them, it certainly wasn't important in the usual way.) But my denial of particularism, and my affirmation of generalism, about virtues is cautious. And this for at least two reasons. First, I think there is a constitutive unclarity about two notions that I have already made quite an important use of, the correlative notions of a role and of a role-virtue. What is a role, and how do we individuate roles, and how widely do roles run? The questions have, I think, no completely determinate answers. That indeterminacy necessarily infects the notion of a role-virtue too; therefore it also spills over into claims about when and where role-dispositions and other dispositions reverse their valences. I would like to be able to argue that the difference I've just pointed to between justice, on the one hand, and temperance, courage and wisdom, on the other, arises from the fact that justice is a citizens' virtue, and citizen is a role-perhaps the most general role there is: that of being a member of civil society. As things stand, I can't argue this because I don't have a firm enough grip on the notion of a role, nor of what, if anything, a human being is when she is not occupying any role, or when she is considered 'in herself and apart from her roles'. So I can only suggest it-as I hereby do.
The second reason why my affirmation of generalism about the virtues is cautious is because of something else that I don't have a very firm grip on (which is a polite way of suggesting that I doubt any of us has). This is the general notion of virtue ascriptions. As a coda to this paper, I say something about this problematic activity in the following section.
V What Is a Virtue Ascription? The basic problem about virtue ascriptions is that I don't think we always have the same thing in mind when we say, for example, 'She is virtuous' (overall, generically), or 'He is V' (with V for the adjective for some particular virtue), or 'That was a V action'. In fact I think we can have a rather untidy variety of different things in mind, which can more or less overlap with each other-or not. As I shall call them, virtue ascriptions can be commendatory, explanatory, causal, dispositional, motivational, deliberative, justificatory, exclusionary, or iconic: maybe other things too, but certainly these. And it is far from obvious that all these different kinds of virtue ascription point in the same direction even for any single virtue, let alone for all the virtues alike.
To give this some concreteness, consider the case of courage. When I praise you by saying 'What a courageous action that was' or 'That was an act of pure courage', it sounds as though I am not only commending you but also making an explanatory claim, a claim about how your action was caused. Apparently I am saying that what caused your action was that you have a specific, modular capacity in you, the disposition of courage, and this module motivated you to act as you did; perhaps I am also saying that the motivation worked like this-you deliberated on what to do, and did this because it was the brave action; and again, perhaps I am saying that your action happened in this way and not some other way that coincidentally looks like courage; perhaps, finally, I am exhorting other people to take what you have done as an example of virtue, and specifically of the virtue of courage.
The trouble is that, as performers of courageous actions regularly point out themselves when they familiarly say 'I was just doing my duty' or 'what needed to be done', there often doesn't seem to be anything like this story behind any particular courageous action. Courageous acts obviously don't need to be, and usually can't be (there usually isn't the time), the result of a deliberative inquiry as to 'What would be the courageous thing to do here?' But it is not only that: it is also that courageous acts do not necessarily seem to result from any particular kind of 'v-thoughts' as we might call them, following Hursthouse (1999, p. 37)-of motivating thoughts that are specific to the virtue in question.
The closer up one gets to the phenomenon of courage, the more courage looks like an absence of dispositions rather than a disposition or set of dispositions. In typical cases of courage, the most unitary thing that we can say about what happens is this: that someone responds to a need in front of them and is not distracted from responding to that need by the fear-filled thoughts that less courageous people would be distracted by. Similarly with self-control. If the mark of the virtue is anything single, it is apparently not so much the presence of a disposition to do the temperate thing, as the absence of any number of dispositions to do any number of intemperate things: the mark, again, is an ability to avoid distractions. Wisdom or intelligence, too, seem to be marked out by focus and concentration; and that too seems to be a matter more of elimination than of acquisition, of subtraction rather than of addition: a matter, not of having any particular disposition, but of not having the dispositions that would make me go after those various distractions. But then, for any disposition, there seem to be indefinitely many more ways of not having it than of having it; so with these virtues, it is not clear how they are to be understood as unities. And they can easily come to look like virtues not just of vision, as Iris Murdoch likes to say, but also of blindness: what is distinctive about these virtues is not only an ability to see 'what is important', but also an ability not to see what is not important. Troy Jollimore (2012) has made this point about what we don't see a key theme of his wonderful book Love's Vision (which I sometimes think he might almost as well have called Love Is Blind). The mountaineer Mark Twight once said that 'In every endeavour those who concentrate and refuse to quit become the elite': it's all about what you don't think about (like the downward view from halfway up El Capitan). This phenomenon of knowing what to ignore is insufficiently attended to in ethics.
With justice things are, perhaps, rather different; here too, justice shows its nature as what David Hume (Treatise 3.2.1.1) called an 'artificial virtue', a virtue of human beings as citizens rather than of human beings 'as such' (whatever 'as such' might mean). With justice we do seem to have a place for a direct deliberative focus on the question 'What is the just thing to do here?', in a variety of ways that it would never make sense to ask oneself deliberatively'What is the temperate (or courageous or wise/intelligent) thing to do here?' But even if justice has this deliberative unity, it poses other problems because of its diversity of subject matter: if justice is rectificatory, and distributive, and redistributive, and retributive, and desert-responsive, and rights-responsive, and other things as well, then it will be hard to unify not because there is no prospect of identifying a specific deliberative 'module' that pursues the question 'What would be just?' but rather because the possible right answers to that question are so very various in form.
So there is at least this much diversity and lack of homogeneity in our talk about the virtues; and probably more, given that so far I am only talking about the cardinal virtues. I do not want to say that, in the light of all this diversity, it is simply impossible for us to make virtue ascriptions-to say, for example, 'This was an act of courage' or 'He is a temperate man' or 'She is a wise woman'. But I do want to say that virtue ascriptions are much more problematic, much more complicated, and much less uniform than we generally reckon.
Does this point make more trouble for Constantine, as (in his way) a generalist about virtue and particularist about actions? Or I don't know who it makes more trouble for; but I do know it makes trouble. At any rate, if you think complication is trouble. Though of course, one might also think that complexity is the mark of truth: that (as someone once put it) the truth is obscure, too profound and too pure, and that living it may well prove to be a somewhat detonatory business. 8

Department of Philosophy Faculty of Arts and Social Sciences
The Open University