Ethics and the Vision of Value Part Two: Normative Ethics Chapter 5: The Good-to-Right Function: Some Unsuccessful Accounts The good is enemy of the best. (Undeservedly popular saying) The best is enemy of the good. (Undeservedly unpopular saying) 5.1. Introduction Chapter 4 argued for a distinction between two kinds of normativity, the evaluative and the prescriptive, and set it as the first question for a normative ethics to say something about how these two relate to each other. How do we get from the evaluative to the prescriptive? In this chapter, as I turn from metaethics to normative ethics, I shall aim to answer that question. More specifically, I shall address the question of how, exactly, the evaluative is an input for the prescriptive: the question how we get from the good to the right, or what the good-to-right function is. There is a prior methodological question about the question of the good-toright function: why should this be the question that we start with in normative ethics? I will criticise some familiar answers to the question of the good-to-right function (5.4-5.5), and offer my own answer to it in chapter 6. Before I do any of that, I’ll say a bit more about this methodological question (5.2). Discussing it will lead me into a closer discussion of one topic in particular, that of alienation (5.3). I begin with a comparison of a number of overall approaches to normative ethics—among them my own, which I call value ethics.1 5.2. Locating value ethics: the focus on values When one looks around at the currently-available options in ethical theory, it is easy to get the feeling that professionalised academic philosophy obeys something like a principle of plenitude. Given enough time, enough money, and enough professional philosophers all seeking to stake out their own distinctive “elevator pitch”, every conceptual possibility gets explored eventually. (Descartes has the same feeling, but puts it less charitably (Discourse on the Method Part II, CSM I: 118): “Nothing can be imagined which is too strange or incredible to have been said by some philosopher”.) The point of the name “value ethics” is to stress the contrast with virtue ethics: as virtue ethics focuses on the virtues, so value ethics focuses on the values. The name is far from perfect: but not much hangs on a name, and anyway I can’t think of a better one. (“Non-consequentialist teleological pluralism”, for instance, is accurate, but hardly elegant.) So “value ethics” it is, at least till I come up with something better. Suggestions welcomed. 1 1 Despite the increasingly apparent operation of this principle of plenitude, it is still true that not all obvious and interesting ideas have been properly explored yet. One obvious and interesting possibility in normative ethics, which deserves more exploration than it has had so far in mainstream contemporary ethical debate, is the possibility of founding ethical theory directly on value, and on our awareness of value: on the contemplation of the good or the goods. The name I shall use for this sort of theory is value ethics. What is value ethics? We may start by looking at its contrast-class: at the other main genera of normative ethical theory most commonly discussed today. Which are these? Very roughly and readily, we have the familiar trio of consequentialism, deontology/ rights theory, and virtue ethics. For a more precise statement of the situation, all sorts of caveats, overlaps and exceptions need to be mentioned. For one thing, perhaps (as some philosophers suspect) “virtue ethics” is not the name of any clear theory. (So, for instance, Nussbaum 1999 thinks that “virtue ethics” is a misleading category-name, better replaced by “neo-Aristotelianism” for views like Rosalind Hursthouse’s and Christine Swanton’s, and “neo-Humeanism” for views like Michael Slote’s and Annette Baier’s.) For another thing, perhaps contractualism and particularism should be in the list. For a third, perhaps we should engage with the scholarly enthusiasm that has already generated something approaching a plenitude of further possible positions by hybridisation and fusion from two or more of these different positions. In the converse, reductive, direction, the same enthusiasm has also shown us how we may treat particularism as a special case of virtue ethics, and contractualism as a special case either of deontology or of rule-consequentialism, depending on the exact details of the contractualism in question. (Thus Scanlon’s contractualism can be seen fairly straightforwardly as a deontological theory, Gauthier’s contractarianism as an indirect consequentialist one.) Accommodating these complications a little, we may say that the main genera to be compared are the following six possible normative ethical theories. Each is located by its own framework principle (as I shall call it).2 Value ethics, the theory I shall argue for, is located by the last, (G0). (Cs0) The3 key explanatory notion in normative ethics is that of a good consequence. (Ct0) The key explanatory notion in normative ethics is that of reasonable rejectability. (D0) The key explanatory notion in normative ethics is that of a correct moral rule. 2 I focus on framework principles to avoid a problem about triviality noted at Hursthouse 1999:30. The problem is that it is hard to see why a consequentialist or deontologist need disagree with what Hursthouse calls the first thesis of virtue ethics: “An action is right if and only if it is what a virtuous agent would, characteristically, do in the circumstances”. Hursthouse’s own response to this problem is a tu quoque—the first theses of consequentialism and deontology are, she contends, equally trivial. I doubt that’s true. As Hursthouse points out ad loc., the deontologist rejects the consequentialist’s appeal to consequences in principle, no matter what the consequentialist’s account of what makes a consequence a good one; and the consequentialist responds likewise to the deontologist’s appeal to rules. But anyway it is a simpler and more constructive answer to the triviality problem to characterise theories by their framework principles. 3 Of course it is an over-simplification to suppose that, for each ethical theory, there is just one thing that we may legitimately call “The key explanatory notion”. At this stage, however, it is a useful oversimplification. 2 (P0) The key explanatory notion in normative ethics is that of situational appreciation. (V0) The key explanatory notion in normative ethics is that of a virtuous agent. (G0) The key explanatory notion in normative ethics is that of response to the goods or values. (G0) tells the reader that the theory of normative ethics to be developed here is a theory about the goods or4 values that exist, and about what rational agents will do about them: a theory, in other words, about what reasons the goods give us. Since the goods in question are—as we shall see—goods “out there in the world”, constellations of patterns or properties in external reality, the theory thus directs our attention not towards possible codes of rules or principles (like (D0)); nor towards the question whether such principles can or can’t reasonably be rejected (like (Ct0)); nor towards the agent’s moral discernment or other dispositions of character (like (V0) and (P0)); but externally, towards the world, which is taken on this view as the natural source of nearly all our first-order reasons. If anyone wants to call this an externalist turn in ethics, that’s fine by me: “The direction of attention is, contrary to nature, outward, away from self which reduces all to a false unity, towards the great surprising variety of the world” (Murdoch 1970: 66). This point already marks the position framed by (G0) as clearly distinct from those framed by (Ct0), (D0), (P0), and (V0). (More on (Cs0), and consequentialism, in the latter half of this chapter.) It also, surely, makes the position framed by (G0) a strikingly obvious possibility for normative ethics. How could we not take seriously the option of basing normative ethics—our theory of the right—simply and directly on facts about the values in the world—our theory of the good? To begin at the evaluative level, by identifying the values, and move from there to the prescriptive level, to ask how we should respond to those values: what alternative in ethics could be more obvious than that? Without yet presenting full-dress arguments against the other alternatives, or suggesting that only one of the alternatives I have presented must be exclusively right, a unique royal road to truth, we can suggest that there’s something oddly indirect about most of them, and that (G0) looks like an under-explored option. For one thing, there are grounds for thinking that all these other approaches to ethical theory really need to begin with the goods or values anyway, even if they don’t admit it. As just noted, many, perhaps most, contemporary normative ethicists prefer to start not with the values, but instead with the virtues, or with rules or rights, or with what an agent can reasonably reject, or with some artificial amalgam of the goods called utility, welfare, or what you will. But the virtues, surely, are either (as I would see them) dispositions of responsiveness to the goods, or at least (as Philippa Foot and others see them) the dispositions that we need to have in order to “flourish”, i.e. enjoy the goods. In either case, it is hard to see how we are to think clearly about the virtues unless we have already done some clear thinking about the goods. 5 And the rules and I use the terms “goods” and “values” interchangeably. Notice that both terms are plurals. For more about value pluralism see 6.1. 5 Darwall (2005: 589) sees the same problem for virtue ethics: “Traits of character consist in dispositions of regard, response, and choice that have objects, and whether a trait is virtuous or not would seem to depend on the value of these objects and how the trait reflects and relates to them.” Cp. Crisp 1992: 155: “The value of the virtues is derivative from the value of well-being.” 4 3 rights in question in rule-theories and rights-theories are just those which we need to protect the goods. So it is hard to see how we are to think clearly about rules and rights unless we have already done some clear thinking about the goods. And what an agent can reasonably reject depends on how the proposal to be rejected would interfere with his pursuit of the goods, or (more fundamentally) how it would interfere with him himself as a good. So it is hard to see how we are to think clearly about reasonable rejectability unless we have already done some clear thinking about the goods. And insofar as there is sense to be made of the notion of utility or welfare, which is not always very far, that sense surely comes from the relation of this theoretically-constructed notion to a much more natural or “folk” theory of the goods of human life. So it is hard to see how we are to think clearly about utility unless we have already done some clear thinking about the goods. This clear thinking happens a lot less than might be expected. It is a striking feature of contemporary consequentialism how little interest its advocates often show in the theory of the good. Again and again one hears that consequentialism is the rational approach to the values, no matter what the values turn out to be (thus Pettit 1993b, for example). As I shall aim to show later in my discussion, this simply is not true. The nature of the goods makes all the difference to what sort of moral theory we should adopt. The goods, then, are the true foundation even of those theories that do not acknowledge them as such. This indirectness in much contemporary moral theory—this unwillingness to begin normative ethics with the goods (among non-consequentialists), or to think clearly about the nature of the goods (among consequentialists)—hampers our inquiries for a number of reasons; one is, because it tends to induce alienation. It is a significant advantage of value ethics that it does not have this tendency. The subject of alienation is interesting and complex enough to need a section of its own; I discuss it in 5.3. 5.3. Alienating ethical theories A now classic example from Michael Stocker (1976: 453) introduces the notion of alienation in ethics: [S]uppose you are in a hospital, recovering from a long illness… when Smith comes in… You are so effusive with your thanks and praise that he protests that he always tries to do what he thinks is his duty, what he thinks will be best. You at first think he is engaging in a polite form of self-deprecation… But the more you two speak, the more clear it becomes that he was telling the literal truth: that it is not essentially because of you that he came to see you, not because you are friends, but because he thought it his duty, perhaps as a fellow Christian or Communist or whatever, or simply because he knows no one more in need of cheering up and no one easier to cheer up. What is alienation, according to Stocker in this passage? And what is wrong with being alienated? Let us agree to begin with that at any rate there is something wrong with Smith. For Smith to say sincerely what he does to the hospital patient 4 (whom I shall call “Jones”) evinces a problem, a mistake, in Smith’s way of thinking about ethics. This could just be denied, of course. Someone might think that Smith is right to care about duty, or the best, in a way that he evidently does not care about Jones. Such a position could be asserted. But the assertion seems far too doctrinaire; it seems to be a sort of moral fanaticism: There is something odd about the idea of morality itself, or moral goodness, serving as the object of a dominant passion in the way that a more concrete and specific vision of a goal (even a concrete moral goal) might be imagined to serve… when one reflects, for example, on the… Saint… giving up his fishing trip or his stereo or his hot fudge sundae at the drop of the moral hat, one is apt wonder not at how much he loves morality, but at how little he loves these other things. (Wolf 1997: 43) Assuming we reject this sort of austere moral fanaticism, it remains to get clear exactly what alienation is, and why it is a problem. One suggestion is that alienation is a problem about inhumanity, or as we might also put it, about foregrounding. It seems inhuman to be motivated, as Smith apparently is, by some abstract ideal when what ought to motivate him, what ought to be in the foreground of his practical reasoning, is the plight of Jones. Thus the problem of alienation arises when a moral theory, or a moral agent such as Smith, ought to appeal to warm and living particularities, such as Jones, and instead appeals to cold and bloodless abstractions, such as moral ideals. “Morality is made for humans—not humans for morality.”6 Smith is alienated because he spends too much deliberative time thinking about utility (or the moral law, or the Marxist utopia, or the Kingdom of Heaven, or whatever), and not enough thinking about Jones. Again, it is not hard to see how a theory like Kant’s can tempt us inhumanly to foreground the wrong sort of factors in our practical reasoning. To quote a passage that has become notorious (Groundwork, G398): …there are many minds so sympathetically constituted that, without any other motive of vanity or self-interest, they find a pleasure in spreading joy around them and can take delight in the satisfaction of others so far as it is their own work... in such a case an action of this kind, however proper, however amiable it may be, has nevertheless no true moral worth, but is on a level with other inclinations... It is just in this that the moral worth of the character is brought out which is incomparably the highest of all, namely, that he is beneficent, not from inclination, but from duty. To judge by this quotation, it looks as if Kantianism is vulnerable to a foregrounding form of the problem of alienation, insofar as Kantianism—as actually presented by Kant, and ingenious rereadings such as Herman 1993 aside—requires the agent to have morally worthy motivations, and restricts the possible content of these to “duty” alone, excluding “inclination”. (“Insofar as”: here a standard response for Cp. Railton’s quotation from Frankena 1973: 116 at Railton 1984: 139. The echo of Christ’s dictum at Mark 2.27, and hence the parallel with religious concerns, is obvious. 6 5 Kantians is that their theory does not in fact require that the agent should act on morally worthy motivations all the time, or even much of it.) The same verdict seems to apply to the problem that Bernard Williams raises in his famous commentary on the familiar example7 of a person who is caught in some disaster in such a way that he can only rescue from it either one of two persons, and who chooses to rescue the person who is his wife. About the agent in such cases, Williams comments, moral theorists characteristically have the idea that moral principle can legitimate his preference, yielding the conclusion that in situations of this kind it is at least all right (morally permissible) to save one’s wife… But this construction provides the agent with one thought too many: it might have been hoped by some (for instance, by his wife) that his motivating thought, fully spelt out, would be the thought that it was his wife, not that it was his wife and that in situations of this kind it is permissible to save one’s wife. (Williams 1981: 18) To have Williams’ famous “one thought too many” (not just “This is my wife”, but also “This is my wife, and in this sort of situation morality allows me to rescue my wife”) is, in my terms, precisely a matter of foregrounding. A theory is likelier to produce the distorted or over-moralised thinking that I call the foregrounding problem if its theory of the good says that something highly general is supremely important: “He who would do good to another must do it in Minute Particulars. General Good is the plea of the scoundrel, hypocrite, and flatterer” (William Blake, Jerusalem, Chapter 3). This point has obvious applications where Blake no doubt meant to apply it—to Bentham and Godwin on the one hand, and Robespierre8 on the other. Less obviously, and in a different way, the point also applies to Kantianism: “If utilitarianism abstracts in moral thought from the separateness of persons, Kantianism strikingly abstracts from their identity”. 9 Any theory that bases everything on a highly generalised and abstract axiology is likely to 7 William Godwin, Enquiry concerning Political Justice, Ch.2: “In a loose and general view I and my neighbour are both of us men; and of consequence entitled to equal attention. But in reality it is probable that one of us is a being of more worth and importance than the other. A man is of more worth than a beast; because, being possessed of higher faculties, he is capable of a more refined and genuine happiness. In the same manner the illustrious archbishop of Cambray was of more worth than his chambermaid, and there are few of us that would hesitate to pronounce, if his palace were in flames, and the life of only one of them could be preserved, which of the two ought to be preferred…that life ought to be preferred which will be most conducive to the general good... Supposing I had been myself the chambermaid, I ought to have chosen to die, rather than that Fénelon should have died... Supposing the chambermaid had been my wife, my mother or my benefactor[, t]his would not alter the truth of the proposition...What magic is there in the pronoun 'my,' to overturn the decisions of everlasting truth?” “If the spring of popular government in time of peace is virtue, the springs of popular government in revolution are at once virtue and terror: virtue, without which terror is fatal; terror, without which virtue is powerless. Terror is nothing other than justice, prompt, severe, inflexible; it is therefore an emanation of virtue; it is not so much a special principle as it is a consequence of the general principle of democracy applied to our country's most urgent needs.” See http://www.fordham.edu/halsall/mod/robespierre-terror.html. 8 9 Here I invert the word-order of the contrast drawn at Williams 1981: 3. 6 be especially prone in this way to the foregrounding problem. The best way to avoid it, therefore, must be to get better values: to develop a more particularised axiology— one which takes there to be a large variety of different, incommensurable, and highly concrete goods. This is the sort of axiology that value ethics gives us—as I shall argue in more depth in chapter 6. There is another version of the alienation problem. On this account, alienation involves an objectionable instrumentalisation of people. Smith does think about Jones in his practical reasoning—as he might not do, much or even at all, on the foregrounding account. But he thinks about him in the wrong way, because he does not think of Jones as an end, or more grandly as “an end in himself”. Smith sees Jones merely as an opportunity or a means for bringing about whatever Smith’s ideological goal may be. It is plausible that moral theories can be alienating in either the foregrounding way or the instrumentalisation way, or indeed in both ways, and that this danger is a serious one for some important moral theories. Bentham’s moral theory, for instance, is based upon the idea that the maximisation of pleasure is the sole moral end; so it seems pretty clearly to face the instrumentalisation charge. Likewise Godwin’s utilitarian moral theory, to go by the quotation in Footnote 7, makes people valuable only as receptacles for happiness (“A man is of more worth than a beast; because, being possessed of higher faculties, he is capable of a more refined and genuine happiness”); I shall have more to say against the “receptacle view” of the value of persons in 6.3. Maybe Aristotle’s ethics too faces this charge of alienation by foregrounding the wrong factors or thoughts. Williams 1995 emphasises a phrase of Aristotle’s that is troubling in this respect, prohairoumenos di’ auta (NE 1105a32). The virtuous man, Aristotle tells us here, is to choose the virtuous act “for its own sake”: to do it because it is virtuous. Both Williams 1995 and Hursthouse 1995 observe that this phrase can be read to mean only that you don’t count as a virtuous agent if your motives are always ulterior. However, their reading does rather leave hanging the question what non-ulterior motives are like for Aristotle. Similarly, Sarah Broadie’s conclusion that Aristotle “appeals to our rough (but firm) sense of a difference between doing what in fact is right and doing it in the right spirit” (Broadie 1991: 88) leaves us asking what exactly counts as “doing it in the right spirit”. Doesn’t the requirement that the virtuous man choose the virtuous act “for its own sake” suggest an alienating focus on one’s own virtues, where a focus on something else would seem more natural, less self-consciously moralistic and priggish? That was a rhetorical question; but in fact, different virtues suggest different answers. As Williams remarks, it is not problematic to suppose that the just man does the just action on “non-ulterior motives” when he does the just action explicitly and self-consciously because it is the just act. A judge who frees a prisoner because it benefits the prisoner, rather than because it is what justice requires, is clearly acting inappropriately. With other virtues, however, such an explicit and self-conscious appeal to the virtue itself in deliberation would seem alienating. So for instance with the virtue of friendship. Clearly, one way for Smith the hospital visitor to act less than fully virtuously is for him to visit Jones “because that is what friendship requires”. Such a motivation is not vicious, as it would be, for instance, to visit Jones in the hope 7 of getting a legacy should he die. But it is obviously less virtuous than visiting Jones simply as a spontaneous expression of care for him, because it is too self-conscious: it foregrounds the wrong considerations, or instrumentalises Jones as a means to Smith’s virtue, or something of this kind. Here too, no doubt, we should be sensitive to a variety of possibilities as to how exactly Smith might visit Jones “simply because Smith thinks the virtue of friendship requires him to”. It is one thing for that description to be true of Smith in the case where it is difficult for Smith to visit Jones, and he has to grit his teeth and fall back on the motivational resource of his knowledge of what the virtue requires, there being nothing else to make him go. It is quite another thing where Smith is just a cold moralistic automaton, and woodenly does what he takes the theory of the virtues to require of him in circumstances where other agents would visit Jones out of warmer and more spontaneous motivations. In the former case, what Smith displays is resolution; it is only in the latter case that he displays alienation. A decent account of alienation ought to capture this distinction, which I suggest rests on the different attitudes that Smith might have to theoretical and to “naturally” (non-theoretically) good motivations. If Smith acknowledges the force of naturally good motivations even when he does not feel them, as is likely to be true in the former case, there is no reason as yet to call him alienated. But if those warmer motivations are alien to him or rejected by him, as they seem to be in the latter case, then he does seem alienated. This helps us to see that coherence between an agent’s theoretical and non-theoretical attitudes is a key test for alienation; more about that below. We should also be aware that the foregrounding and instrumentalisation forms of alienation are not so very far apart. The instrumentalisation version of the problem does not, despite its name, always have to be a problem about means-ends relations. Some utilitarians—perhaps Mill is one of them—meet the objection that their theory breeds the instrumentality form of alienation by insisting that (for example) any individual person’s welfare is not a means to the general good, but an instance or part of the general good. The problem with this response is that treating an individual as an instance of the general good can be just as much a way of instrumentalising him as treating him as a means to the overall good. If my interest in someone’s welfare is motivated only by the thought that his welfare fits the description “instance of the overall good”, my attitude to him seems no less instrumental than it would be if my interest in his welfare was motivated only by the thought that it fits the description “instance of the overall good”. In either case, we might also say that I seem to be too interested in these descriptions, and not enough interested in other possible descriptions of him that treat him less as a place-holder and more as an individual. At this point the foregrounding and the instrumentality formulations of the problem of alienation are very close together. The problem is not exactly that the utilitarian values these aspects of his life only as a means to an end, for the enjoyment that he and others get from these aspects are not a means to, but only a part of, the general happiness. Nonetheless, he values these things only because and insofar as they are a part of the general happiness. He values them, as it were, under the description “a contribution to the general happiness”. (Wolf 1997: 88) 8 Of course, there is a familiar line of response to the objection that a given moral theory produces alienation in either of these first two senses. When (the response runs) we reflect on what the foregrounding and the instrumentalisation accounts of alienation suggest goes wrong in the alienated agent, we may well accept that it’s a mistake for Smith to tell Jones that he’s visited him in hospital because it’s Smith’s aim to maximise overall utility (or follow the categorical imperative) in all he does. However—the response continues—just because Smith shouldn’t say this, it doesn’t follow that it isn’t true. It would be unwise for me to make an offer of marriage by pointing out to my beloved that she and I alike are both instinctively inclined to propagate our own DNA. Yet that, apparently, would be true, however shatteringly inapposite it might be. Likewise, perhaps, it is a general truth that there are claims that belong in the discourse of moral theory, yet are quite out of place in actual moral practice, and conversely that many things that are typically said in actual moral practice, can’t be said in moral theory. Maybe this is how things stand, for instance, with Williams’ one-thought-too-many example. If Brown has the thought “This is my wife, and that in situations of this kind it is permissible to save one’s wife”—as opposed to the thought “This is my wife”—then perhaps what is wrong with Brown is not that his thought is false, but rather that it is not assertible in this context. This response to the alienation objection deploys the use/ mention distinction to defend “sophisticated” moral theories (Railton’s word) with a certain familiar sort of two-level structure. There is a moral language, a first order, and a moral metalanguage, a second order. The meta-language tells us what sort of talk and thought should go on at the first-order level, but it does not (and cannot) itself use those firstorder sorts of talk and thought: it can only mention them. Peter Railton’s well-known distinction (Railton 1984) between an agent’s criterion of rightness (CR)—what makes her action right—and her decision procedure (DP)—the deliberative process whereby she actually gets to the right action—exemplifies this idea of a distinction of orders of discourse. As Railton argues, an agent’s DP can be benevolent thoughts of the familiar commonsensical kind even though his action’s CR is some austerely theoretical thought about obedience to principle or maximisation of utility. This latter is a thought which the good agent will not actually have at all, so long as she is engaged in action as opposed to reflection. Where the criterion of rightness says “Promote utility”, the thoughts that should appear in the agent’s decision procedure are those thoughts which will most promote utility by appearing there. But it can easily be that these utility-promoting thoughts are not themselves thoughts about promoting utility. Where the utility-promoting thoughts in the DP are instead such thoughts as “He’s my friend”, “She’s my wife”, “Our relationship is of great value”, or whatever, the criterion of rightness will itself dictate that these thoughts should appear in the DP; the criterion of rightness will not use those thoughts, but it will quote them, as things that should be in the decision procedure. Furthermore, where CR-level thoughts about promoting utility would actually decrease utility if they appeared in the DP, the criterion of rightness itself will prohibit their appearing in the DP. This is the sense in which second-order thoughts about utility can be selfeffacing: they can counsel the agent to avoid them while she is operating at the first order. Does this deployment of the use/ mention distinction succeed as a way of blocking the alienation objection? That depends on why any given theory finds it 9 necessary to invoke this distinction. If we can’t say at the meta-language level what we can say at the object-language level, the question is always: Why not? Where the answer to this question is “Because what is said at the object-language level contradicts what is said at the meta-language level”, we have a form of alienation that still presents a threat to Railton’s sort of two-level theory. To see why, return to Smith the hospital visitor. Smith would find himself in logical conflict if he held the following two claims:10 1. Nothing matters in itself except the impartial maximisation of utility. 2. Jones matters in himself. Smith would also find himself in logical conflict if he held both of the following two claims: 3. Nothing matters in itself except doing one’s universalisable duty. 2. Jones matters in himself. And Smith would find himself in logical conflict if he held both of the following two claims: 4. Nothing matters in itself except acting in the way that the virtuous agent would act. 2. Jones matters in himself. In each case, Smith would face a logical conflict, because the former claim contradicts the latter: if 1, 3, or 4 is true, then 2 cannot be true, and vice versa. 1, 3, and 4 are alienating claims, in the sense that accepting any of them alienates Smith from 2—logically prevents Smith from accepting 2. If Smith accepts 1 or 3 or 4, then regarding 2, his choices are (a) to deny it, or else (b) to affirm a contradiction. Either way Smith will be alienated, because either way Smith will be denying 2. He will deny 2 outright if he takes option (a); but he will also deny 2 if he takes option (b), since that commits him to affirming both 2 and the contradictory of 2. The upshot of this argument is that any theory that denies certain claims that are central to ordinary moral thinking will be wide open to the alienation objection in a form that is not easily dismissed. If a normative ethical theory actually implies the denial of claims such as (2) in the example just sketched, then it does not merely face the (arguably negotiable) foregrounding, instrumentality, or use/ mention forms of the alienation objection. It also faces the (unquestionably fatal) contradiction form of that objection. Such a theory faces a choice between rejecting claims such as (2)—and giving up the whole pretence that what goes on in the object-language of ethics can be accepted by the serious moral theorist at all—and accepting contradiction as part of moral life. This choice is an impossibly unattractive one for any moral theory to have to face. But more contemporary theories face this choice than might appear from 10 Claims of the form of (2) are, I believe, true, and will get a full defence in 6.3. I am not here defending, or presupposing the truth of such claims; here I use (2) for illustration purposes only. 10 Railton’s attempt to turn the flank of the alienation objection. For example, it is not unusual for contemporary consequentialists, even “sophisticated” ones, to say both that nothing matters except net overall welfare, and also that net overall welfare is not best achieved if people believe that nothing matters except net overall welfare. (Railton asserts this thesis; Parfit describes it.11) The dilemma facing this thesis is simply the question: what should the good consequentialist agent believe? His alternatives, apparently, are to believe that “Nothing matters except net overall welfare”, or to believe the contradictory of that. But which? (It is no answer to say that he has an epistemic reason to believe the one thing, and a moral reason to believe the other. For first, saying that he has these conflicting reasons to believe does not answer the question “What should he believe?”. And second, the idea that there are moral reasons to believe things is dubious enough in general, but especially dubious when a moral reason to believe something is opposed to an epistemic reason not to believe it. These same arguments are decisive against a second suggestion: that we say that what the good consequentialist should believe is simply that he has these reasons. Nor is a third suggestion viable, namely, to say that the good consequentialist ought to believe the one thing qua moral agent, and its contradictory qua moral theorist. I cannot set myself to have two contradictory beliefs under two different descriptions of myself; for it is still myself who is to have these beliefs, under whatever description.) Thus contemporary sophisticated consequentialism faces serious trouble from the contradiction form of the alienation objection—perhaps more serious than is generally recognised in the current state of the debate. But it is not only consequentialism that faces this trouble. The same problem arises for any theory that sharply distinguishes what is good from what it is good to believe is good. For example, some versions of virtue ethics clearly face just this trouble.12 Hursthouse (1995; 1999: Chs.6-7) and Williams (1995) seek a way of preserving the claim that virtue is both criterion of rightness (CR) and decision procedure (DP) for the good agent’s actions. According to them, being “motivated by virtue” is not simply a matter of having the crass thought “x is the virtuous thing to do, so I’ll do x”. Rather, an agent A is motivated by the virtue V when A has the relevant “X-reasons” (Williams 1995: 18; Hursthouse 1999: 131), i.e. the motivating thoughts that an agent fully possessed of virtue V would have. So if Jones is ill in hospital, Smith’s decision to bring him flowers is virtuous—instantiates the virtue of kindness—if it is motivated by thoughts like “He loves begonias, and those wards are so bare”, or “This is just the thing to cheer him up”. This is the thought that belongs in the DP; the thought that belongs in the CR is a quite different thought: “In this kind of situation, it is right to be motivated by thoughts like ‘He loves begonias, and those wards are so bare’”. But this, once more, looks very like a distinction between what is good, and what it is good to believe is good. If so, it will be just as prone as the consequentialist’s distinction to lead the virtuous agent into the severe form of alienation trouble that arises from contradiction. 11 12 A thoroughgoingly self-effacing theory of ethics is sketched, but not endorsed, in Parfit 1984: 40-43. Some “sophisticated” versions of Kantian theory may face it too: see e.g. O’Neill 1997. 11 The same is true of the “pure” virtue ethical theories defended by Slote 1997 and Zagzebski 1996. A “pure” theory “understands rightness in terms of good motivations and wrongness in terms of the having of bad motives” (Slote 1997: 241); for “pure” virtue ethics, “the property of rightness… emerges from the inner traits of persons” (Zagzebski 1996: 79). In the end, then, a “pure” virtue ethics will have to answer “Why is A the right thing to do?” by “Because A is the action that issues from motives XYZ”. So a “pure” virtue ethics of this sort will direct agents to act on a decision procedure that uses motives like XYZ. But its criterion of rightness will not use motives like XYZ; it will merely mention them, as examples of right-making motives. This theory too apparently leads us to the thought that there is a difference between what is good, and what it is good to believe is good; and so as close to the dangers of the contradiction form of alienation as sophisticated consequentialism. A position in virtue ethics that perhaps avoids this trouble is Christine Swanton’s. Swanton claims that the virtuous agent is motivated by what (Swanton 2003) she calls “the targets of the virtues”, where these are “aretaically specified”13 values to which a variety of different forms of response—not only promotion—are appropriate. Hence, for her, “there is no dissonance between the values which are the focus of a virtue—i.e. the values to which the agent is responsive when exercising the virtue—and those values which provide the rationale of a trait as a virtue… the rationale of the virtue gives the agent’s proper motivation” (Swanton 1997: 174-5). For Swanton, what is good and what it is good to believe is good are one. In this respect, Swanton’s view and value ethics are very close. For value ethics too is well fitted to deal with the alienation problem.14 Because its account of the goods, like Swanton’s, is fundamentally a generous, inclusive, and pluralistic one which is happy to take the data of (virtuous) common sense on trust wherever possible, value ethics does not tend to generate a conflict between what is good and what it is good to believe is good. As moral theorist, the value ethicist says that Smith is worth visiting in hospital “because visiting him is a way of responding to the individual human good that Smith instantiates”; as moral agent, the value ethicist says that Smith is worth visiting in hospital “because Smith matters”. The difference between these two remarks is not the difference between two diverse, and potentially contradictory, accounts of what we should think of as valuable. Rather, it is simply a difference of register: between the tone and vocabulary of moral theory, and that of ordinary life. And there is nothing wrong or alienating about this difference— provided it remains one that can always, in principle, be eliminated. Thus we reach the conclusion about alienation that is my main thesis in this section. One reason why value ethics is a good normative ethical theory is that it leaves this possibility open, and so, unlike a significant variety of other ethical theories, need not be prone to any serious form of the alienation objection. Value ethics’ success in avoiding alienation also shows something about the methodological question that I considered above: the question whether the good-toright function is the thing to look for in normative ethics. I say some more about that issue in 5.4. Meaning that it takes virtue to specify them accurately; there is no “view from nowhere”, outside the virtuous person’s outlook, from which the targets of the virtues can be specified. 14 I am grateful to Fred Miller for pressing me on this issue. 13 12 5.4. The teleological intuition: for and against Having considered alienation, let us come back to this chapter’s opening question. How do we get from what’s good to what’s right? By what operation do we work out, on the basis of our beliefs about what actions or dispositions or characters or states of affairs (…etc.) are good, which of those actions (etc.) are right: which of these actions we ought to do or pursue, or have reason to do or pursue? As I’ve suggested we put it: What is the good-to-right function? The “teleological intuition”, as I shall call it, is the intuition that this is the right question to be asking: that if we want to know the right, what we have reason to do or ought to do,15 then we need to start with the good (or goods). If rightness is (generally, though not exclusively) some function of the good, it is crucial to establish what function of the good it is. Teleological ethics, the project of trying to do this, is my main project in the latter half of this book. A quick word in passing here about “the moral” and “the practical”. I take these terms, and their co-ordinates such as “moral reason” and “practical reason”, to be close to synonymous. In my view (as also in Aristotle’s, Hobbes’, Kant’s, and Bentham’s), no useful or plausible analysis of moral reasons makes them out to be anything other than practical reasons of the most general sort. The point clearly has wide-ranging implications. For instance it means that, pace Joyce 2005: 288, “to hold a moral error theory” such as Mackie’s is “to hold an error theory for practical normativity in general”. A second point in passing: pace many authors, e.g. Pettit 1993, “teleological” and “consequentialist” should not be considered synonyms. Aristotle, Plato and Aquinas are all teleological ethicists. None of them is a consequentialist, and Aquinas is a paradigm anti-consequentialist. Teleological theories are theories that specify a good or end (τέλος), or a plurality of such goods or ends (τέλη), and arrive at an account of ethics by asking about how we should respond to that good or goods—in my terminology, by looking for a good-to-right function. Consequentialism is only one possible answer to this question, because consequentialism involves a very particular form of response to the good: a maximising or promoting response. In 5.5 I shall explain why I reject the consequentialists’ particular account of the good-to-right function. Before that, it is necessary to defend the teleological intuition a little further. For there are reasons for thinking that at least some part of what is right is not dependent on any good-to-right function. It seems that we can have reasons that have nothing to do with the nature of the good. If we can, then maybe the search for a good-to-right function is either unnecessary, or—more strongly—downright misguided. Maybe we should dispense with it, and look in some entirely different direction for the basis of our normative ethics. The chief source of this worry is, of course, Kant’s practical philosophy. Without the least pretence of presenting a scholarly exegesis of Kant, I may just say Or “are permitted to do”. At this stage I am not yet distinguishing the right (= the obligatory-to-do) from the permissible (= the not-obligatory-not-to-do), as species of what we have reason to do. 15 13 this: Kant argued (and to my mind, argued successfully; for more on this see Chappell 2005b, chapter 3) that practical rationality’s own logic determines some substantive tests on rightness and wrongness (in a broad, practical sense: “what is to be done”). To be an agent is to value some ends or other, and to pursue them in action. It follows that, no matter what else an agent values, he has reason to value his own agency; for his agency is his own capacity to pursue whatever else he values. It also follows that, on any conception of agency (including the Humean and Hobbesian models that Kant probably has in mind as his most important targets), it would be practically irrational to undermine or to destroy one’s own practical rationality/ agency. Crucially, this argument does not show that my own practical rationality/ agency is a value; it shows only that I am bound to treat it as if it were a value. Kant has in place a very general scepticism about the possibility of direct knowledge of the external world, including knowledge of value in the external world. His point is not— as mine is: see 6.3—that we ought to add each agent’s own practical rationality or agency to our list of values “out there in the world”. His point is that, no matter what we think about what values are “out there in the world”, a certain sort of positive and protecting attitude to our own practical rationality/ agency is rationally mandatory. For Kant, then, our reasons to act are not based on the good or goods at all; they are based on the internal logic of practical reason itself. Hence, for Kant, the right is not a function of the good, but of reason. Now we can accept Kant’s point about the internal logic of practical reason without accepting either Kant’s scepticism, or his transcendental solution to that scepticism. If we do so, we will be free to agree with Kant that some of our practical reasons can be derived from the nature of practical reason itself. Those having to do broadly with self-preservation and self-development can be so derived, for instance. (If, as I shall argue (6.3), each individual human is a good, these reasons can also be derived from those goods; in which case the existence of these reasons is overdetermined.) Whether reasons having to do with the preservation and development of other agents can also be derived from the nature of practical reason itself is a harder question. Kant thinks they can; elsewhere (Chappell 2005b, chapter 3) I have raised some doubts about his derivation. Obviously the issue is crucial for Kant, since he hopes to derive an entire normative ethical system from the nature of practical reason alone, and to show that this ethical system is the only possible one: to show, as we might say, that ought implies Kant. If the step from self to others cannot be made, then this implication won’t go through. Kant will have succeeded in deriving only a self-regarding ethics—an ethics that has each person recognise his duties to himself, but not to anyone else. But such an ethics is a (possibly quite high-minded) form of egoism; and whatever else Kant intended, it surely wasn’t egoism, high-minded or not. Be that as it may, we can accept that some of our practical reasons derive from the nature of practical reason itself without having to accept—as Kant’s scepticism leads him to accept—that there can be no other source for our practical reasons. Our conclusion will then be that not all our practical reasons are derived from our values: that is, that there is some rightness that is not determined by any good-to-right function. But, of course, to show that rightness is not exclusively a function of the good is not to show that it is not a function of the good at all. The basic teleological intuition is still in play. 5.5 will explore it some more. 14 5.5. Against maximising: why the good-to-right function cannot go via the best Our question is: by what function do we get from the good to the right? The most obvious answer to the question seems to be that we get from the good to the right by looking for the best. So—apparently obviously—the good-to-right function is maximisation: “Practical rationality in the most general sense is identified with maximisation” (Gauthier 1986: 22). However, this seemingly obvious answer runs into insuperable difficulties. What is at first sight the best-looking candidate for the role of the good-to-right function turns out, on closer examination, to be quite incapable of playing this role. This closer examination brings us to a consideration of act consequentialism. Everybody who has done any philosophy knows that standard-issue act consequentialism is precisely that teleological theory that computes the right from the good via the best: it tells us that in every choice what we ought to do is the best thing available. Unfortunately for contemporary ethics, the stark implausibility of this injunction, if we take it literally, is as little noted as the serious danger of loose and self-serving moral reasoning, if we take it loosely and non-literally. As a purported justification of an action, nothing can sound more authoritative than the claim that that action is the best one available. But if this claim is not meant literally, then its apparent authority is a sham, a temptation to muddled thinking, special pleading, and specious reasoning. If, on the other hand, the claim that a given action is “the best one available” is meant literally, it turns out that no one can ever be warranted in asserting it. No one can ever be warranted in asserting it, because the best is always either both unknowable and indeterminate, or else just unknowable. The best is both unknowable and indeterminate in cases like the following. An agent A has to choose between two courses of action or ways of life X and Y, each of which brings with it its own distinctive moral outlook: that is to say, it is part of the choice that if A chooses X, then A will come to hold a perspective within which A justifiably believes that X is a better course of action or way of life than Y, whereas if A chooses Y, then A will come to hold a perspective within which A justifiably believes that Y is a better course of action or way of life than X. These beliefs are both justified, because both are well-grounded in the different perspectives within which they arise. A’s Ybased grounds for believing Y better-within-one-perspective than X, and conversely his X-based grounds for believing X better within-another-perspective than Y, are such that they make good sense even to competent observers who do not share either of A’s possible perspectives. When things are like this, the goodness of X and the goodness of Y are incommensurable with each other, in the sense that it is false that X is better than Y, false that Y is better than X, and false that Y and X are of equal value. None of these three is the case, because outside the Y-based perspective and the X-based perspective, the value-relations of X and Y to each other16 are “To each other”: these words must be noted well. Incommensurability in one place in a theory of value is often assumed to lead to a fatally anarchic indeterminacy everywhere in that theory. If this assumption depends on an inference from X’s indeterminacy-relative-to-Y to X’s indeterminacyrelative-to-everything, then it depends on a non sequitur. To anticipate my examples, it can be that there is no fact of the matter as to whether marrying Zoe or marrying Chloe would be better, even when 16 15 indeterminate. In such cases there is, therefore, no extra-perspectival fact of the matter even about which of X and Y is better, never mind about which of them if either is best. In such cases, therefore, maximising accounts cannot give us a credible goodness-to-rightness function.17 “But are there any such cases?” There surely are lots of cases like this, including some examples that are trivial and small-scale, and others that are much more momentous. To take a trivial example, there need be no antecedently determinate fact of the matter about whether it would be better to eat Italian or Chinese this evening, even though it is (happily) the case that, if we go for an Italian, the evening will go in such a way that Italian will come by the end of the evening to seem18 to have been the better choice all along. Just likewise, though on a much larger scale of importance, there need be no antecedently determinate fact of the matter about whether it would be better to marry Zoe or Chloe, even though it is (happily) the case that, if I marry Zoe, my life will develop in such a way that marrying Zoe will come after a few years to seem to have been the better choice all along—and likewise if I marry Chloe. (“Happily”: it is, of course, notorious that the converse is also possible: it can be that whichever I choose of X or Y, I will later come to think I made the wrong choice. Beyond acknowledging its existence, I won’t discuss that unfortunate possibility here.) I say “come to seem”, but let me stress that there need be no illusion involved in such changes of perspective. It’s not that, before I decided to marry Zoe rather than Chloe, it was false that it was better to marry Zoe than Chloe (so that I was wrong to think this). It’s rather that, before my decision, there was no definite fact of the matter as to which was better; my decision, we might even say, created the fact of the matter. My choice, and my living out of that choice under conditions which (if I get lucky) will more or less approximate eudaimonia, precisified the values at stake; it gave them particular and definite values for me, in the context and from the perspective of the narrative pattern of my life, which they did not have outside that context or perspective. In this sort of way, it seems to be a perfectly familiar part of human experience that choice is not determined by exact pre-existing rankings of values, but is itself what determines them. In a slogan, choice is anterior to value; less catchily but more exactly, choice can be anterior to fully-determinate value even though it is posterior to rough value. What a mistake, then, it would be to take my behaviour in such cases as revealing something already precisely fixed and given that we may call my “preferences”, or my “utility-function”;19 or to think that it must be the case that I chose as I did because I measured the options against each other by reference to a fully determinate scale of values that, implicitly or explicitly, I already accepted, according to which my choice came out as best. there is a fact of the matter as to whether, say, marrying Zoe or marrying Dewi (or marrying Zoe or murdering Chloe) would be better. 17 Notice that here, in contrast to chapter 2, the introduction of a perspective is not a relativisation of ethical truths, but a precisification of them. 18 I do not here mean “seem” to be understood as contrastive with “be”. Cp. my remarks above about seeing things from within a perspective. 19 And it would be mistake upon mistake first to read off my supposed preferences from whatever I do in fact prefer, and then to claim that my adherence to this canon of preferences shows that I am a maximiser, when on the contrary it shows only that I prefer whatever I prefer. 16 If there are cases of choice which are like this, then there are cases where, anterior to choice, the best is both unknowable and indeterminate. But of course, even if some cases of choice are like this, not every case is. In some cases, the best may for all we can tell be determinate. But we can’t tell, because in all these cases the best is unknowable.20 The reason why it is unknowable is not just because of the familiar point that the future is open, so that the assessment of consequences is an impossible and uncompletable task. That point is true, and (as Jimmy Lenman has recently argued, in criticism of Shelly Kagan and others) it is a much more serious objection to maximising consequentialism than recent consequentialists’ nonchalance about it tends to suggest.21 But there is another point as well, which does not depend on the feasibility of assessing the overall goodness of indefinitely many different possible futures. This is the point that any agent, at any node of choice, has an indefinitely large option range (range of alternatives open to him); therefore, no attempt to enumerate any of his option ranges can ever be counted complete; therefore, it can never be known that the best option in any of his option ranges has been identified. An insuperable epistemic barrier stands in the way of knowing, for any agent at any time, which is the best action available to him. The same barrier therefore stands in the way of knowing that the agent has done that best action. Why should we think that “any agent, at any node of choice, has an indefinitely large option range”? The answer lies in the denseness of possibilia: in the infinite fine-grainedness of possible action-descriptions, and the impossibility of knowing, for each possible however-tiny adjustment of an action-recipe, that it won’t produce a hugely different action. (Moving your arm just so may be saluting the Queen a little perfunctorily; moving it a bit more may be saluting the Queen with deep reverence; moving it a tiny bit more than that may be parodying saluting the Queen; and so on.) Of course we could stipulate that we were going to ignore all these finer-grained differences, and just concentrate on one particular way of individuating options. But that would be a decision to look away from the problem, not a decision that solved the problem. For we could never know, for any such stipulation, that it would pick up all and only the differences between action-descriptions that matter, and do so more effectively than any alternative possible stipulation. Hence we get a second-order version of the original difficulty: attempting to avoid the problem that there are indefinitely many options in any option-range, we find ourselves having to 20 The argument of the remainder of this section is, among other things, a reworking of the argument of Chappell 2001. My thanks to John Skorupski, Adam Morton, Bernard Williams, Earl Conee, Jonathan Lowe, Cora Diamond, Onora O’Neill, and other commentators on the original argument. Since “Option Ranges” appeared, a rather similar case has been forcefully argued by Wiland 2005. 21 “[Kagan] does not begin to do justice to the worry. The worry is not that our certainty is imperfect, but that we do not have a clue about the overall consequences of many of our actions. Or rather—for let us be precise—a clue is precisely what we do have, but it is a clue of bewildering insignificance bordering on uselessness—like a detective’s discovery of a fragment of evidence pointing inconclusively to the murderer’s having been under seven feet tall. We may not be strictly without a clue, but we are virtually without a clue” (Lenman 2000: 349-350). Lenman’s argument is criticised by Mason 2004, who argues that we have no practically rational option but to count unknown consequences as cancelling each other out. Even where that is true, it does not seem to undermine Lenman’s conclusion. Suppose I decide to do A on the basis that 0.0015% of the evidence is favourable, 0.0014% of the evidence unfavourable, and the remaining 99.9971% of the evidence is unknown. Agreeing with Mason that, if what we should do is weigh consequences, then my choice is practically rational, is no bar to agreeing with Lenman that my choice is clueless. If cluelessness is what we get when we do what the method of weighing consequences rationally requires, this fact surely tells against that method. 17 decide between indefinitely many ways of individuating option-ranges so as to limit them. “Are there no agents of whom it’s true that they only have a few, and a definite number of, options available? What about a disabled person, whose paralysis is so severe that his options are only (1) to twitch his finger, or (2) not to twitch it?” But it’s a mistake to say that even this person has only two options. The mistake rests on a confusion between actions and bodily movements. Certainly this person has only two bodily movements available to him at any given instant, but—as the experience of severely disabled people has repeatedly shown22—that by no means implies that he only has two actions available to him at any given instant: he might use those bodily movements to tap out the Iliad in morse code, to order porridge for breakfast, to propose marriage to his nurse, or indefinitely many other things. If even this agent, in his unhappily extreme circumstances, still has indefinitely many options available to him, then a fortiori so do all other agents. “Your argument presupposes that we always need to be able to survey every option in a given option-range to be able to know which option in that range is best. And sometimes we don’t: for example, if there is a bomb in this room, then we know that the best thing for us to do is to leave immediately.” No: what we know is that the right thing for us to do is leave immediately, for reasons that have nothing to do with the delusory search for the best option. (One evident advantage of the present argument is that it enables us to respect the intuitive gap23 between “right” and “best”—a gap that philosophers try too often to close. Since “right” is most plausibly understood as equivalent in meaning “wrong not to do”, whereas “best” is most plausibly understood as equivalent in meaning to “better than all the other alternatives”, it is far from obvious why we should want this gap closed.) “Your argument confuses decision procedure and criterion of rightness. Act consequentialists don’t at all say that we can know the best option in practice, at the moment of deliberation or decision. They say only that it is by reference to the best option that each choice within each option-range is justified or criticised.” As a matter of fact, this is false: act consequentialists (and other sorts of consequentialists too) are constantly saying or implying that access to the best option is a regular part of ordinary agents’ decision procedures. Here are two prominent examples: [A] consequentialist theory... tells us that we ought to do whatever has the best consequences... [More formally,] the consequentialist holds that... the proper way for an agent to respond to any values recognised is... in every choice to select the option with prognoses that mean it is the best [i.e. the most probably successful/ highest scoring] gamble with those values. (Pettit 1993b: xii and 232-3; italics added) Is Pettit talking here only about a criterion of rightness? On the most natural reading of this quotation, it doesn’t look like it. Consider the further evidence of this quotation from the same context: “An agent promotes certain values in his choices, “After suffering a massive stroke, Jean-Dominique Bauby, editor-in-chief of French Elle and the father of two young children, found himself completely paralysed, speechless and only able to move one eyelid. With his eyelid he 'dictated' this remarkable book” (from the blurb to Bauby 1998). 23 A gap which is an open-question gap, among other things: 4.4-4.5. 22 18 we can now say, if—and indeed only if—the agent ranks the prognoses of options in terms of these values... and ranks the options—where the ranking determines his choice—in terms of their prognoses” (italics added). Second, Peter Singer: Suppose I then begin to think ethically... I now have to take into account the interests of all those affected by my decision. This requires me to weigh up all those interests and adopt the course of action most likely to maximise the interests [sic; I suppose Singer means “maximise the satisfaction of the interests”, though the meaningfulness of this phrase too is questionable] of those affected. Thus at least at some level in my moral reasoning I must choose the course of action that has the best consequences, on balance, for all affected. (Singer 1993: 13; italics added) Is Singer talking here only about a criterion of rightness? On any credible reading of this quotation, far from it. This is meant to be a description of what actually happens, in real agents’ psychologies, when they begin to do what Singer is pleased to call “ethical thinking”. (The obvious point that, in fact, nothing much like what Singer describes does happen in real agents’ psychologies when they are doing real ethical thinking is one that has already emerged from my discussion of alienation in 5.2.We shall return to it in chapter 6.) Still, let us be charitable, and allow that consequentialists could have been more careful than these quotations show they are about distinguishing what is knowable at the decision-procedure level from what is knowable at the criterion-ofrightness level. Even then, the distinction does not help them against the present objection, because there is nothing in the structure of my argument to give them any hope of knowing which option is the best even at the criterion-of-rightness level. If the problem is, as I say, a problem about the indefinite fine-grainedness of actiondescriptions, then this problem will apply as much at the criterion-of-rightness level as at the decision-procedure level. This means that act consequentialism, the theory that what we should do is always whatever available act will have the best consequences, can deliver no determinate conclusions at all about how we should act. Contrary to what I have heard act consequentialists claim at this point—“All right, so our theory faces severe epistemic difficulties; that doesn’t mean it’s false”—this suffices to refute act consequentialism. For act consequentialism is a normative ethical theory, a theory the point of which is that it purports to tell us what is right, what we ought to do.24 If the argument just given is correct, then act consequentialism tells us nothing about what we ought to do beyond the bare information that we ought to do what’s best. That is simply not enough on its own to give us a worthwhile theory of normative ethics. I have also heard consequentialists respond to the present argument by divorcing “what is right” or “what is best” from “what we ought to do”; the same divorce also seems to be the upshot of positions such as Alastair Norcross’s “scalar consequentialism”. This divorce seems to me a pretty clear sign of desperation. “It is fine for a theory in physics to tell us about its central notions in a way which leaves it obscure how it can move from these notions to action, for that passage can be left to something that is not physics; but the passage to action is the very business of ethics” (Frank Jackson (1991: 467) as quoted by Lenman 2000: 361). 24 19 Compare the Aztec Theory of normative ethics, which tells us that what we should do at any time is the action that is commanded then by the god Huitzilopochtli. However, no one ever knows what Huitzilopochtli commands; indeed no one ever can know that, given that, happily, Huitzilopochtli does not exist. Since the Aztec Theory of normative ethics is supposed to tell us what to do, and since the Aztec Theory tells us nothing about what to do, our ignorance of Huitzilopochtli’s wishes not only renders the Aztec Theory completely useless; it thereby refutes it. Pari passu, the impossibility of anyone’s ever knowing what the best action is not only renders act consequentialism completely useless; it thereby refutes it. However, there is another way of defending a view rather like standard-issue act consequentialism; one which is still fairly definitely a form of consequentialism. Suppose we spell out maximising as trying to bring about the best, and promoting as trying to bring about the better. (Since the best is the better-than-all-others, maximising thus turns out to be a species of the genus promoting.) This way of distinguishing promoting and maximising gives a useful determinacy to two often rather loosely-deployed terms, though it is not, so far as I can see, exactly standard. (It is not for instance the usage suggested by Pettit as quoted above: he seems to treat them as synonyms.) Suppose, anyway, that we adopt it. Then what would be wrong with going for a promoting but non-maximising view, and saying that we get rightness out of goodness by looking, not for the best, but for the better? Not much would be wrong with it, so far as it goes. But it goes almost no distance at all without a specification of better than what. As soon as we ask this question, the pressure is on to move from promoting to maximising. The promoting function tells us to seek the better. But whenever we find the better, by asking the question of any two a and b “Which is better?”, there is nearly always some c readily available, such that we can ask whether c is better than either a or b. Then if we choose c on the grounds that that is better, there is nearly always some d available… and so on, until promoting collapses into maximising. Iterating the promoter’s search for the better pushes him quickly towards the maximiser’s search for the best. Refusing to iterate would itself be a decision that could be challenged by asking whether this refusal was better. We can reach the same conclusion by going up an order, and comparing the promoting and the maximising procedures by asking the promoter’s question about them—“Which is better, the promoting function or the maximising function?”. Here the promoter might respond that actually his function is better, precisely because it avoids the objection that I have developed about the unknowability of the best: since he is content to look for the better, and does not even try to seek the best, he can ignore this objection about unknowability. However, the promoter’s idea must be that a promoter can avoid the maximiser’s problem by stopping asking the question “Which is better?” at some particular point. Then for any given point, we can raise the question whether it is better to stop asking the question “What is better?” at that point rather than at some other; just as, on the maximising model, the attempt to place any limit on otherwise interminable searches for the best option always faces the question whether that particular limit is the best one to impose. The only way out of this is to accept either that it is arbitrary at what point we stop looking for the better; or that there is more to the good-to-right function than betterness. To admit the former alternative is to admit that betterness cannot furnish us with a good-to-right 20 function—since the search for a good-to-right function was all along a search for a non-arbitrary good-to-right function. To admit the latter is to reduce consequentialism to the claim that it should be a necessary condition upon choice that it should promote in the following exceedingly minimal sense: all morally and rationally acceptable choice involves the choice of something that is better than something else. Since it is hard to think of anything whatever that is not better than something else, and harder still to envisage circumstances in which anyone might be tempted to choose something that is not better than anything else, this claim is almost completely vacuous. No better than this is a possible promoting view that we might call contrastive consequentialism, which says that the thing to do is a iff a is better than some set of salient alternatives b, c, d. Just as before we faced the question “Better than what?”, so here we face the question “What is salience?”. We need to know, at the first order, what makes it the case that b, c, d are the right alternatives to a to consider, and, at the second order, whether the account of salience that is on offer—if any—looks better than other accounts (or, indeed, other salient accounts) of salience. Without clear answers to these questions, the practical reasoning that runs “a is better than b, c, d; so a is the thing to do” is no better than the theoretical reasoning that runs “A mouse is cleverer than a cockroach, a tortoise, or a house-fly; so a mouse is clever”. A third possibility is satisficing, the idea that the good-to-right function should be defined by good enough. Some versions of satisficing theories are merely tactical: they claim, in brief, that a satisficing decision-procedure is to be adopted because it is the best way of meeting the demands of a maximising criterion of rightness. These theories are not really satisficing theories at all. As is clear from their explanation of why we should satisfice, they are maximising theories, and as such are dealt with by the objections to maximising that I have already offered. Other versions of satisficing theory are genuinely satisficing, in that the satisficing they advocate is not merely tactical, but strategic: these theories are serious, “all the way down” as it were, about the idea that the good-to-right function should be defined by good enough. Let us turn to these. One standard objection to strategic satisficing consequentialism (see Mulgan in Ratio) is that it would, e.g., allow someone who could press button A and save a million lives, or button B and save two million lives, to press button A, thus knowingly permitting a million needless deaths. But strategic satisficing consequentialism is not committed to allowing this choice. The satisficer is not claiming that in every situation, something clearly less good than some other available option is good enough. He is only claiming that in many situations, something clearly less good than some other available option is good enough. It is consistent with that claim to refuse to allow the agent in this example to press button A. Another objection (Bradley 2006) is that, in many easily conceivable situations, the strategic satisficing consequentialist’s principles will commit him not only to allowing the best or the better not to happen, but to acting to prevent it from happening. But, it is said, acting to prevent the best or the better from happening is unacceptable for a consequentialist. Since consequentialists don’t typically acept the 21 doing/ allowing distinction25, allowing the best or the better not to happen must be unacceptable for consequentialists too. However, this objection fails against satisficing because a satisficing view need not be a consequentialism: I myself shall argue for a satisficing non-consequentialism. And it fails against satisficing consequentialism because it begs the question against that view, the whole point of which is that acting to prevent the best or the better need not be unacceptable for a consequentialist. A better objection to satisficing consequentialism is encapsulated in the question “How much is enough?”. When we push this objection, we come—as just hinted—to see that although satisficing looks like it offers a promising way of defining the good-to-right function, satisficing consequentialism does not. It will take some time to explain in full what I mean by this claim. To explain it turns me from the negative task of criticising others’ account of the good-to-right function, to the positive task of presenting my own account. That will be my business in chapter 6. 25 On this see chapter 8. 22