OBJECTIVIST COMPATIBILIST UTILITARIANISMbyNICHOLAS CAMPBELL SLEIGHB.A., The University of British ColumbiaM.A., Simon Fraser UniversityA THESIS SUBMITTED IN PARTIAL FULFILLMENT OFTHE REQUIREMENTS FOR THE DEGREE OFDOCTOR OF PHILOSOPHYinTHE FACULTY OF GRADUATE STUDIES(Department of Philosophy)We accept this thesis as conformingto the required standardTHE UNIVERSITY OF BRITISH COLUMBIADecember 1993@ Nicholas Campbell Sleigh, 1993In presenting this thesis in partial fulfilment of the requirements for an advanceddegree at the University of British Columbia, I agree that the Library shall make itfreely available for reference and study. I further agree that permission for extensivecopying of this thesis for scholarly purposes may be granted by the head of mydepartment or by his or her representatives. It is understood that copying orpublication of this thesis for financial gain shall not be allowed without my writtenpermission.(Signature)_____________ _______Department of_____________The University of British ColumbiaVancouver, CanadaDate “O’.1c IDE-6 (2/88)ABSTRACTIn this thesis a version of the ethical theory ofutilitarianism is defended. The version defended is called‘objectivist compatibilist utilitarianism’, or ‘OCtJ’. On thisversion, utilitarian metaethics includes the propositions thatthere is an objective, intrinsic property of goodness entifying,motivating, and grounding ethics, and that act and ruleutilitarianism are compatible since under plausibleinterpretations both true. While this metaethical theory hasperhaps not been stated explicitly before, theories in this veinhave been popular since the mid—l9th century, and have beenexpounded by philosophers such as J.S. Mill and G.E. Moore.OCU will be defended by examination of six influentialobjections to various of its hypotheses. In each case, thoroughconceptual analysis, aided by consultation of relevant scientificfacts about human nature, will reveal that the objection isseriously flawed. In the process of dispatching these negativeconsiderations, a comprehensive positive ethical theory willemerge. Most other currently popular ethical theories are notcomprehensive, but instead take no position (or several, whichamounts to the same thing) on one or more of the major issuesthat any comprehensive ethical theory must deal with. OCU thusemerges as one of only a very few contemporary comprehensiveethical theories — and on balance the most plausible of the lot.iiTABLE OF CONTENTSAbstract iiTable of Contents iiiAcknowledgement vForeword viChapter One Initial Considerations1. Introduction 12. Moral entities 73. Qualons and indexons 12Chapter Two The property of goodness1. Aristotle’s dilemma 202. Goodness and coloredness 253. Skepticism about determinable properties 334. Some puzzles resolved 42Chapter Three The concept of goodness1. The metaphysics of concepts of determinables 472. Entertaining concepts of determinables 603. The feel of goodness and the sense of right 86and wrong4. The open question argument 97J-’1Chapter Four Components of utilitarianism1. Coinpatibilist utilitarianism 1102. Utilitarian rules 123Chapter Five Supererogation1. Small beneficial sacrifices 1482. Valuable but not obligatory 1533. The sacrifice/gain ratio 180Chapter Six Two metaphysical background issues1. Goodness and pluralism 1892. The Synthetic A Priori 2083. Conclusion 226Bibliography 227ivACKNOWLEDGEMENTMy thanks to Randy Reiffer, John Stewart, Dale Beyerstein,Kurt Preinsperg, Leo Paquin, Paul Russell, Richard Sikora, MarkDickson, Earl Winkler, J.L. Wisenthal, Thomas Hurka, and othersfor helpful discussions and useful comments on earlier drafts.VFOREWORDThe bulk of Chapter Two has been previously published; thebibliographic information is as follows.Sleigh, Nicholas (1992), ‘Objective Goodness and Aristotle’sDilemma’, Journal of Value Inquiry 26, 341—51.viChapter OneINITIAL CONSIDERATIONS1. IntroductionIn this thesis I shall defend a version of utilitarianismcalled ‘objectivist compatibilist utilitarianism’, or ‘OCU’. Onthis version, utilitarian metaethics includes the propositionsthat there is an objective, intrinsic property of goodnessentifying, motivating, and grounding ethics, and that act andrule utilitarianism are compatible since under plausibleinterpretations bothtrue. While this metaethical theory hasperhaps not been stated explicitly before, theories in this veinhave been popular since the mid—l9th century with Mill (1861).This branch of ethics reached the height (so far) of itsinfluence in the early 20th century with the work of Moore (1903,1912, 1922, 1959). I shall not defend this historical claim ofantecedency, though I shall refer to some points made by thesemoral philosophers.Current skepticism (among metaethicists, at least) about OCUis due largely to six objections, which this thesis will examine.The first objection is that the proposition that anobjective property of goodness exists entails a lethal dilemma:1if (a) whatever (intrinsic) goodness there isin the world exists as a property whichsupervenes on the natures of certain things,e.g., pleasure or honor, then the things’natures seem deprived of any direct role inexplaining the goodness of things, whichseems wrong. Yet if (b) whatever goodnessexists does not supervene on the natures butexists wholly within them, then since some ofthe natures have, qua good things, nothing incommon, a fortiori the natures have noproperty of goodness in common. If goodnessis a property then either (a) or (b), henceno property of goodness exists.Call this ‘Aristotle’s dilemma’ after Aristotle’s expression ofit (1985, p. 11). Aristotle’s dilemma grows from the feelingthat we should be able to say more about the property of goodnessthan merely that it exists; we should be able to say what kind ofproperty it is, and explain its basic features in standardmetaphysical terms. Yet, it is sometimes claimed, when we try tosay more about the property of goodness we run into insuperablemetaphysical difficulties, as Aristotle’s dilemma illustrates.The second objection to OCU is that even if a satisfactorymetaphysical account of the property of goodness can be supplied,the phenomenological difficulty remains that most people areunable to detect such a property; the concept of goodness eludes2them despite their following whatever procedures the moralobjectivist recommends for the inducement of this mental state.Indeed, some have argued that no such concept is possible.The third objection to OCU (and utilitarianism in general)is that the most plausible and popular utilitarian doctrines areforms of either act or rule utilitarianism, but that each ofthese doctrines suffers a repellent objection. The doctrine thatan act is right if it will result in at least as much utility asany other available act is Act Utilitarianism (AU); the doctrinethat an act is right if endorsed by a rule from a set whoseadoption by society will result in at least as much utility asthe adoption of any other set of rules is Rule Utilitarianism(RU). AU is repellent for its apparent implication that theintuitively justified rules of commonsense morality (e.g., thatpromises should be kept, that truth should be told, and thatinnocent persons should not be punished) deserve no special favorand on any given occasion should only be followed if doing soseems likely to produce more utility than not doing so. Cynicismand social disorder seem more likely than harmonious happiness toresult from taking these rules so lightly. Call this thecounterintuitivity objection. RU, on the other hand, isrepellent for its apparent implication that rules should befollowed even on occasions when doing so would not gain anyutility. A moral code that makes demands unsupported by the onlyintrinsic value it (overtly) posits is unconvincing. This is therule worship objection. Thus both the chief forms of3utilitarianism are seriously flawed, hence utilitarianismincluding OCU should be abandoned altogether.The fourth objection to OCU is the objection fromsupererogation. A supererogatory act is, to define it in as muchdetail as space permits, an act, beyond the call of duty, thatbenefits someone other than the agent, and is altruisticallymotivated. Saints and martyrs are extreme examples of the kindsof people to whom supererogatory acts have traditionally beenattributed, though at the other end of the spectrum, very smallaltruisms may be supererogatory, e.g., taking a stray cat to theSPCA. In the context of OCU, a supererogatory act is analtruistically motivated act, uncalled for by any rule in theshort moral code of the agent’s society (i.e., beyond the call ofduty, hence, in a sense I shall define, not obligatory), thatboth increases overall utility and increases utility for othersentients. The agent typically sacrifices some of its ownutility to make these increases, but self—sacrifice is notessential to supererogation. The fourth objection to OCU is thatit has the unworkable feature of deeming some actssupererogatory. OCU’s form of supererogation can be criticisedin three ways, some ways applying to all forms of supererogation,other ways applying only to some forms, including OCU’s.The first criticism is that under this definition of‘supererogation’, any personal sacrifice, no matter how small,for the sake of any gain for others, no matter how great, wouldcount as supererogatory as long as it increased utility but4wasn’t called for in the short moral code of the agent’s society;by the intuitions of some, that doesn’t seem right (R.I. Sikora,personal conununication).The second criticism of OCU’s form of supererogation is thatthe practice of supererogation is unjustified because the conceptof supererogation is paradoxical, in that a supererogatory actwould make things better all things considered, but is notobligatory. This nonobligatoriness seems inconsistent withthe resilient intuition that if an action isgood, this gives us a reason to do it, and ifit is the best available, we have more reasonto do it than to do anything else and soought to do it. If this intuition is sound,the best available action will always be theright one, the one that one ought to do.(Dancy 1988, p. 176)This general criticism of any morality’s form of supererogationis starkly applicable to OCU’s, for according to OCU all and onlyutilities intrinsically ground reasons for action.The third criticism of OCU’s form of supererogation is thatit is impossible that the set of rules forming OCU’s short moralcode could ensure that the correct sacrifice/gain ratio is onaverage achieved by everybody, simply because there is no suchthing as a sacrifice/gain ratio that is correct for everybody.These three criticisms jointly form the fourth objection toOCU, the objection from supererogation.5The fifth objection to OCU is that moral pluralism might betrue, but that OCU is a form of moral monism, hence isimplausible to the degree that moral pluralism is a live option.The sixth objection to OCU is that OCU assumes the existenceof some synthetic a priori propositions, but that no suchpropositions exist. OCU does make this assumption, so mustdefend the synthetic a priori. OCU makes this assumption byassuming that determinable properties exist; it will be seen thatdeterminable properties, whether construed as supervening ontheir determinates or as existing within them as parts, stand inmetaphysically necessary relations with their determinates, orwith those parts of their determinates that are ontologicallyindependent, and that the propositions describing these states ofaffairs are synthetic a priori. OCU also assumes the existenceof analytic a priori propositions, but these are much lesscontroversial entities, so will not be defended in this thesis.In this thesis I shall refute these six objections to OCU,hence show that OCU is a much stronger comprehensive moral theorythan is often supposed. The counterarguments I make will reveala substantial positive moral theory. The first objection to OCUwill be dealt with in ch. 2,the second objection in ch. 3, thethird in ch. 4, the fourth in ch. 5, the fifth in the firstsubsection of ch. 6, and the sixth in the second subsection ofch. 6.Many other objections that have at times been influentialagainst OCUish utilitarianism have already been refuted, e.g.,6the lack of calculation time objection (refuted by Mill 1861, p.30—32), the cold, calculating mentality objection (refuted byJ.J.C. Smart 1973, p. 44—45), the objection of obsession withpleasure (refuted by Blake 1967, p. 432—33), the objection thatpleasure cannot be quantified (refuted by Blake 1967, p. 434—35;see also §6.1), the objection that utilitarianism should butcannot attribute intrinsic value to the distribution of goodsaccording to desert (refuted by Sikora, forthcoming), and theobjection that utilitarianism should but cannot promote, for atleast extrinsic reasons, distributions of goods according toegalitarian considerations (refuted by Brandt 1979, p. 219—20,ch. 16). Similar refutations appear elsewhere in works toonumerous to mention. Some of these refutations point outproximate errors in the relevant objections, others consist inshowing that the objection applies to any form ofconsequentialism, and appealing to the extreme implausibility ofany ethical theory on which the consequences of an act are of notthe slightest importance in deciding whether one should do theact. Many of these objections are susceptible to both sorts ofrefutation.In the remainder of this chapter I shall explain a fewdetails of the nature of OCU, and a metaphysical distinctionrelevant to my defense of OCU.2. Moral entities7Moral language includes many kinds of terms, some referringto ethically fundamental entities, others functioning otherwise.Among the latter, and resembling the former, are terms referringto ethically derivative entities. For instance, the terms‘right’, ‘ought to’, ‘obligatory’, ‘permissible’, and ‘duty’ allhave ethical content but are all derived, according to OCU, fromthe term ‘good’, which in the relevant sense attributes theirreducibly ethical property of intrinsic goodness. Goodness,and its negative twin (or negative half), badness, together makepossible definitions of the other ethical terms listed: e.g., insimplified form, for ‘the right act’, read ‘the act that would dothe most good’; for ‘ought to do X’ read ‘would do the most goodby doing X’; for ‘is obligatory under any circumstances’ read‘would uniquely do the most good’; for ‘is obligatory all elseequal’ read ‘all else equal will do the most good’; for ‘ispermissible’ read ‘would not all else equal do the most good notto’; for ‘has a duty to do X’ read ‘all else equal would do themost good by doing X’. These definitions in some cases reflectactual usages from the natural language English, and in othersare reforming definitions designed to pick out aspects of moralreality more closely than previous definitions. These sampledefinitions are simplified for brevity, and in full would takeaccount of many complications, such as the right act beingstrictly speaking not the act that does the most good, but theone that brings about the optimal balance of good over bad,taking into account both the absolute quantities of good and bad,8and their mutual proportions (the literature on suchcomplications is extensive; e.g., Moore 1912, ch. 2, 3).However, it is clear in what sense these full definitions wouldbe derived from ethically fundamental entities. Since all othermoral definitions are derived from goodness and badness, andsince the metaphysics of badness is in every respect either thesame as or the complement of the metaphysics of goodness, I shallexplain and defend the entities and language of OCU almostentirely in terms of goodness. Exceptions occur where a relevantclaim by another writer has been couched in terms of derivativemoral entities, and it is simpler to discuss the claim in thesesame derivative terms than to reverse the derivations.It might be objected to my free use of existing moralterminology, including the claim that at least some of theseterms have existing senses that pick out real moral entities,that existing moral terms are hopelessly vague and ambiguous,hence that reliance on the linguistic intuitions they embody canyield no definite results in moral inquiry (e.g., Brandt 1979, p.4—6). It must be admitted that few or no existing moral terms ofordinary language have single, clear, language—wide meanings.However, when I claim to use existing moral terminology, I meanmerely that the terms in question have, among their varioussenses, well—established senses that pick out real entities.These well—established senses can be used to yield definiteresults in moral inquiry. These senses are well—established inthat for many years they have been known to and used by many9people, not all of whom are extreme specialists in the relevantsubject.Terms throughout language have senses well—established inthis way; for instance, a nautical term might have a sense knownto almost no one but sailors, but which was a common term amongthis group, with a clear working sense known to almost all ofsuch people. This is not to say that sailors could give apolished technical definition of the sense, merely that theysuccessfully use it in their trade and easily explain it tonewcomers. As Sikora remarks,the best test I know of for whether teachingthe meaning of a given sort of expression ispossible is whether people can be taught bythe procedure in question to use the newword with substantial agreement in regard toan open class of cases. (1981, p. 448)The moral terms I have in mind pass this test. If words thatmeet this criterion, but not a stronger one, are hopelessly vagueand ambiguous, making them unfit for philosophical use, then agreat deal of what was hitherto taken to be excellent philosophymust be scrapped. But it is silly to force impossibly highstandards upon language. When I claim that ‘good’ has anexisting sense that I wish to use, I think this claim verified bythe fact that ‘good’ has for centuries been used by philosophers,theologians, and many others in a way quite as practical andprecise as the way in which sailors use ‘marlinspike’, and which10received expert clarification and endorsation almost a centuryago with Moore (1903, ch. 1). Such use, clarification, andendorsation is no guarantee that the entity the sense picks outexists. I rely on the linguistic intuitions this sense embodiesmostly to help suggest to the reader the sort of entity I have inmind, not to prove that such an entity exists; it is the task ofthis thesis to help provide evidence that the sense picks out anexisting entity. The existence of the relevant sense is someevidence, but far from conclusive. But as Brandt agrees (1979,p. 7), nothing much hangs on such linguistic questions. If nosuitable sense of the term ‘good’ had existed in English, I wouldhave invented one.As to my use of existing moral terms other than ‘good’ tohelp introduce topics or frame questions, even though I freelyadmit that such terms are in some respects vague or inaccurate, Ihold that such terms are nonetheless useful in directing thereader’s attention at least to the general area of inquiry I wishto pursue. Brandt is in no position to criticize this tactic,for he uses it himself: he opens his book thus:Moral philosophy has traditionally been asystematic attempt to answer some questionsof apparently universal interest about: whatis worth wanting or working for; what is thebest thing for an agent to do from his ownpoint of view; what is morally right; andwhat is morally just. (1979, p. 1)11The tactic is endorsed by Russell:I do not pretend to start with precisequestions. I do not think you can start withanything precise. You have to achieve suchprecision as you can, as you go along.(1986, p. 170)One of the trials of conducting philosophical inquiry in anatural language is the scarcity of accurate terminology withwhich to begin the inquiry. However, the only alternatives areto use an artificial perfect language, which does not exist, andif it did would render much inquiry superfluous, or to conductinquiry without using language, an activity which cannot becoherently imagined. So let’s be practical and get on with thejob.3. Qualons and indexonsElucidating certain relevant metaphysical points willrequire us to distinguish between two kinds of property; callthem ‘qualitative properties’ and ‘indexical properties’.Qualitative properties ontically express what a thing is, whatkind of concrete stuff it is made of; indexical propertiesontically express where a thing is. A qualitative property of myfridge is that it has mass; two indexical properties of my fridgeare that it is to the left of the stove and that it is twice as12tall as the stove. A qualitative property of my visualsense—datum of my fridge is that it is green; two indexicalproperties of my visual sense—datum of my fridge are that it isto the left of my visual sense—datum of the stove and that it istwice as tall as my visual sense—datum of the stove. Thus agiven qualitative property cannot be realized both mentally andphysically, whereas at least some indexical properties can berealized both mentally or physically (see McDowell 1985, p.115—16).I call the first sort of property ‘qualitative’ because ofthe view that for things to have concrete natures is forsubstances to be qualified. I take ‘substance’ in a broadlySpinozan sense (Spinoza 1967; Bennett 1984) in which substance isthe space—time continuum augmented to incorporate any otherirreducible locational field. Qualitative properties are allthose concrete properties possibly inhering in substance soconstrued. Qualitative properties may be either mental orphysical (or, counterfactually, of some other sort of concretestuff), and are necessary though not always sufficient for causalevents.I call the second sort of property ‘indexical’ for twoetymological reasons. First, the locational fields they navigateserve as indexes to individuate whatever qualitativelyindistinguishable entities they contain (hence may be termed‘indexical fields’). Two identical golden balls in an otherwiseempty universe could only be distinguished by their different13spatiotemporal locations. The index reads ‘ball A is atsuch—and—such a location, ball B is at such—and—such a location’;since these locations differ, balls A and B are different balls.Second, many such properties make essential reference toindexicals such as ‘this’, ‘I’, ‘my’, ‘here’, and ‘now’ (Perry1979). For example, to have the indexical property of being tothe left of the stove is to be to y left of this stove. To havethe indexical property of being a meter long is to be the samelength as this stick. Perhaps all indexicals reduce to somesubset, perhaps not. The parts of an indexical field areidentical, but can be picked out by ostension or qualitativecontents. The possibility of an indexical field entails thatsufficient indexical properties exist to ontically express thefield, whether or not such properties are actually instantiated.Some properties which might seem qualitative, such as theproperty of being circular, are indexical; in the present case,to say that a doughnut is circular is to say that to say that itsdoughnut parts are spatially located in a certain way relative toeach other, and space is an indexical field. The notion thatsuch properties are qualitative makes sense only if one takes thesubstances they inhere in to be bare particulars, which aremerely conventional entities, rather than parts of a Spinozanfield, which is an objective existent. To hold that a baseballis a set of properties clothing a single particular, as the bareparticular theory suggests, is implausible. If I cut thebaseball in half, so that there are now two objects hence two14bare particulars, at the severance of which atomic connection didthe bare particular bifurcate? All the problems of identity overtime in Greek ships (e.g., Chisholm 1969) attach to the bareparticular theory, whereas the Spinozan substance theory isimpervious to such objections because Spinozan substance isextended.The metaphysics of qualitative and indexical properties iscomplex and difficult; a full defence of my interpretations ofthem must for brevity be omitted from this work. However, thegeneral distinction involved is almost universally made in someguise or other. Locke, for instance, distinguishes ‘primaryqualities’ and ‘secondary qualities’ in roughly such a way as topick out, respectively, indexical and qualitative properties(1894, p. 170); Williams terms qualitative properties ‘qualities’and reduces indexical properties to information “shown” insuitable statements in a logically improved language lackingpredicative terms signifying the sizes, shapes, or movements ofthings (1969, p. 312); Husserl distinguishes pure essence, havingcontent, from form, which has no content (1970, p. 455). Theexact way in which this distinction is made is unimportant toOCTJ; it is important only that the distinction be made. Itherefore invite those who disagree with my interpretation of thedistinction to read on forbearingly, and to later mentallyrearrange the relevant details to suit themselves. Indeed, Ileave it open that some or all indexical properties reduce awayas Williams suggests, for ontic parsimony is as important as15realism in giving an account of the phenomena underlying thedistinction. Such a reduction would be consistent with realismabout indexical phenomena, for indexical facts would be preservedin the states of affairs to which indexical properties were inthis way reduced. Facts are truths, and by the correspondencetheory of truth any truth must correspond to some real state ofaffairs. Not all reductions would be thus acceptable. Areduction of mental states to physical states would be in thisway disfavored, for something of the conceptual flavor orcharacter of mental states is utterly lacking in physical states,whereas the suggested reduction of indexical properties toinformation “shown” in suitable statements in a logicallyimproved language lacking predicative terms signifying the sizes,shapes, or movements of things would preserve the conceptualflavor involved; the disagreement would be about a theoreticalentity. The reduction of molecules to atoms is a successfulexample of the latter sort.Note that the ‘reduction’, if the term is appropriate, ofgoodness to a property would not, as some people seem to think,be disfavored on the ground that the conceptual flavor orcharacter of ethical thinking would be utterly lacking in atheoretical entity from technical metaphysics such as a property.On the contrary, to assert that a property of goodness existswould be to assert that the conceptual flavor or characterinvolved is a real, tangible thing in the world, just as thepositing of mental properties asserts the existence of the flavor16or character of the mental as a real, tangible thing in theworld, not to be explained away as having its existence solely insome other real, tangible thing of a different conceptual flavoror character.Analogously, it is sometimes objected that free will cannothave a mechanistic explanation, since the mechanistic displacesthe purposive, yet free will involves purpose (see Dennett 1982).If events of free choice are reduced to mechanistic sequences,says the objection, then the responsible character of freedomwill be lost. The vague intuition underlying the objection seemsto be that if freedom is precipitated out as concrete objectslinked causally, then it won’t be ethereal or fluid enough tomatch our feelings when we do things freely. To this objectionthe modern response is that freedom and causation are compatible,perhaps even necessarily linked (see Hobart 1934). Ethereal,fluid phenomena are nonetheless concrete objects capable ofstanding in causal relations with other objects. As withfreedom, so with goodness; ethically ethereal goodness can beentified as a qualitative hence concrete property without losinganything of its character.Define an object composed wholly of qualitative propertiesas a ‘qualon’; the color set green and white is a qualon. Definean object composed wholly of indexical properties as an‘indexon’; a rectangle is an indexon. ‘Quale’ and ‘qualia’ wouldbe just as good as ‘qualon’ and ‘qualons’ for this purpose, butthe former terms already have established usages according to17which the subjective, conscious, experiential parts of thinkingbrains are termed ‘qualia’ (e.g., Shoemaker 1990, p. 110; Angel1989, p. 62). If one used the term ‘qualia’ in this sense topick out the items I call ‘qualons’, then one would reject thepossibility of being conscious of indexical properties; however,we are conscious of such properties. Also, some non—indexicalproperties are physical, which ‘qualon’ allows for, whereasqualia are always mental. Rather than confuse matters by coininga new sense for ‘qualia’ taking care of these objections, itseems simpler to coin ‘qualon’.Part of the aim of distinguishing qualons and indexons is tohelp dispel the widespread confusion between scientific inquiryand inquiry about the physical nature of reality. Scientificinquiry is empirical inquiry about instantiation patterns amongproperties, chiefly about causal property successions. Thecomparative ease of scientific inquiry about purely physicalproperty instantiation patterns, and difficulty of scientificinquiry about sets of properties some of whose members aremental, has caused many to believe that science is concerned onlywith physical properties (e.g., R.N. Smart 1964, p. 106—07), orworse, that mental properties are somehow less than fully real,and are only intellectually respectable when reduced to physicalproperties. Such confusion has sometimes led to unjustifiedmoral skepticism (see Taylor 1989, p. 235—36).It is unclear whether my interpretation of mental andphysical entities is a form of monism or dualism. As will be18discussed in §3.1, the monism/dualism distinction is difficult toconstrue in the traditional way, i.e., exclusively. Myinterpretation of mental and physical entities holds that mentaland physical concrete entities each constitute quite real,tangible kinds of thing in the world, having distinct conceptualflavors or characters, but that both mental and physical concreteentities are qualons. It is unclear why one should pay moreattention to their difference than their similarity, or viceversa, and that is what one needs to do in order to opt forinonism or dualism construed exclusively.19Chapter TwoTHE PROPERTY OF GOODNESS1. Aristotle’s dilemmaThe first objection to OCU is that one of its premises, theproposition that a property of goodness exists, in a robust,objective, intrinsic sense in the spirit of Moore (1903, ch. 1;1922, ch. 8; 1959), is subject to a lethal dilemma:if (a) whatever (intrinsic) goodness there isin the world exists as a property whichsupervenes on the natures of certain things,e.g., pleasure or honor, then the things’natures seem deprived of any direct role inexplaining the goodness of things, whichseems wrong. Yet if (b) whatever goodnessexists does not supervene on the natures butexists wholly within them, then since some ofthe natures have, qua good things, nothing incommon, a fortiori the natures have noproperty of goodness in common. If goodnessis a property then either (a) or (b), henceno property of goodness exists.20Call this ‘Aristotle’s dilemma’ after Aristotle’s expression ofit (1985, p. 11; see also Strawson 1949, p. 26—27). Aristotle’sdilemma grows from the feeling that we should be able to say moreabout the property of goodness than merely that it exists; weshould be able to say what kind of property it is, and explainits basic features in standard metaphysical terms. Yet, it issometimes claimed, when we try to say more about the property ofgoodness we run into insuperable metaphysical difficulties, asAristotle’s dilemma illustrates. Some writers are pessimisticabout the prospects of answering such difficulties; for instance,J.L. Mackie says thatif there were objective values, then theywould be entities or qualities or relationsof a very strange sort, utterly differentfrom anything else in the universe. .What is the connection between the naturalfact that an action is a piece of deliberatecruelty — say, causing pain just for fun —and the moral fact that it is wrong? Itcannot be an entailment, a logical orsemantic necessity. Yet it is not merelythat the two features occur together.(1977, p. 38)A Moorean moral objectivist (as, for brevity, I call theproponent of a Moorean property of intrinsic goodness, though asMoore notes, there is also a sense in which one can be a moral21objectivist without positing intrinsic goodness or any otherintrinsically valuable ethical entity — 1922, p. 255; variousversions of moral objectivism will be formally distinguished in§3.2) must answer such questions, both by giving a positivegeneral account of moral entities and by showing how this accountresolves standing particular objections. I think that Mooreanmoral objectivism can supply such answers, and as evidence showhow it, or at least the version of it developed in this thesis,can deal with one of the most influential metaphysicalobjections, Aristotle’s dilemma.I argue that the Moorean moral objectivist can escapeAristotle’s dilemma through either fork (a) or fork (b),depending on how a metaphysical issue is resolved. The solutionis that the relevant relation between the property of goodnessand the properties comprising the natures of good things is aspecies of entailment. To say that property A entails property Bis short for saying that the proposition that x has A entails theproposition that x has B. This conditional is itself true invirtue of the natures of A and B in a way I shall discuss.I interpret supervenience to obtain only betweenontologically independent entities, so that a complex property isnot supervened on by its parts. I use ‘entailment’ not in thenarrow sense of strict logical implication, but in the broadersense of conceptual necessitation. While entailment in thisbroader sense is a relation whose instances are knowable apriori, propositions asserting these instances may be either22analytic or synthetic. I take a proposition to be analytic ifits logical form, as revealed by the meanings or meaningful usesof its terms, reveals its truth—value, and synthetic otherwise; Itake a priori knowledge of a proposition’s truth—value to restessentially on a correct understanding of the meanings ormeaningful uses of the proposition’s terms.Thus there may be different kinds of entailment. To escapeAristotle’s dilemma, the Moorean moral objectivist specifies thatmoral entailment is of the same general kind as the entailment bywhich necessarily a red thing is colored. Perhaps theresemblance between color entailment and moral entailment (as Icall these two kinds of entailment) is not perfect, but theMoorean moral objectivist need only hold that the two kinds ofentailment are alike in whatever common aspect of theirstructures allows them to escape dilemmas of the form ofAristotle’s. Such an aspect exists (at least in colorentailment) because, given that coloredness exists, Aristotle’sdilemma cannot be used to prove the nonexistence of the propertyof coloredness.(Using the term ‘common aspect’ is to some extentquestion—begging, but like Neurath’s boat—carpenter, I can dolittle about it beyond noting that all rational inquiries suffersome form of this defect, yet many are successful — see Neurath1937, p. 276. We must frame our questions from thepresupposition—laden linguistic material we find at hand, hopingthat ensuing inquiry will, as often happens, reveal or clarify23the presuppositions and shed light on their truth—values. Thesepresuppositions must eventually be analysed and perhapsdiscarded; their preliminary use for pragmatic reasons is not anexcuse for intellectual mush, such as Rorty 1989).Thus I follow the time—honored strategy of seeking outrespectable siblings for a shady proposed entity. Goodnessresembles other entities in various ways. Goodness and itsnegative twin (or negative half), badness, together make up thewhole of an intensity continuum, just as do the properties ofheat and coldness. Goodness and humans, like pleasure andhumans, are such that necessarily humans who grasp the concept ofgoodness (or pleasure) have some positive disposition towardgoodness (or pleasure), though not necessarily to the exclusionof a simultaneous negative disposition toward goodness (orpleasure). And goodness and coloredness are alike in that theyare determinable properties, by which I mean (at least) that theycannot be instantiated in the absence of the instantiation of oneof a range of other, determinate properties, such as pleaure andredness, in virtue of whose instantiation the determinableproperty in question is also instantiated.My defense of this solution to Aristotle’s dilemma has twoparts: first, an explanation of how I interpret coloredness, andhow the property thus interpreted, hence also goodness, escapesAristotle’s dilemma; and second, a discussion of somemetaphysical objections to the existence of properties such asgoodness and coloredness which ontically express the unity of24ranges of less controversial properties.2. Goodness and colorednessI interpret coloredness in such a way as to avoid somerelevant disanalogies between goodness and coloredness as thelatter is sometimes interpreted. The first disanalogy stems fromthe view that determinate colors are dyadic properties, so thatwhat is, say, red to me may be green or transparent to you.By a dyadic property I mean a two—place relation. Thisinterpretation views relations and properties as notfundamentally different, but merely as universals obtaining ofdifferent numbers of particulars. Armstrong attributes this viewto Plato and possibly Aristotle (1989, p. 15—17), and Russell(1912, p. 94—96) and Oddie (1991, p. 21) subscribe to this view.To illustrate, the dyadic property of heaviness obtains of theordered set {a 5 year old child, a sack of potatoes}, and thedyadic property of lightness obtains of the ordered set {theworld’s most muscular human, a sack of potatoes}, for the sack ofpotatoes is light to the one human trying to lift it, and heavyto the other. The sack of potatoes thus participates in bothlightness and heaviness, objectively but relatively. A monadicproperty is a ‘one—place relation’, i.e., a property in thenon—Russellian sense in which the word ‘property’ is often used.Colors such as redness and greenness are sometimes interpreted as25objective but dyadic properties, e.g., by Dolby 1973, so that thesame physical object might be red to me but green to you. Onthis interpretation an object might to you be green, hencecolored, and to me be transparent, hence not colored.Plainly, the Moorean moral objectivist cannot allow theanalogue for goods and goodness that one person might correctlyjudge a state of affairs to be pleasant, hence morally good, andanother person correctly judge the identical state of affairs notto be pleasant, hence not morally good. Such relativism woulddefeat the whole spirit of Moorean metaethics (also the letter;e.g., Moore 1959, p. 96—97).Thus life would be easier for the Moorean moral objectivistif colors were non—relative, monadic properties. If there wereno disanalogy between goods and colors, hence also, mutatismutandis, between goodness and coloredness, in this respect, afortiori no disanalogy would be relevant to Aristotle’s dilemma.As it happens, color terms such as ‘red’ and ‘green’ are used inseveral senses in ordinary English, and in one sense they aremonadic. I intend this latter sense by ‘red’ and ‘green’, and somutatis mutandis for other color terms such as ‘coloredness’.Clearly colors in this sense exist.Consider a person who has normal human vision and sees aripe tomato. The existence of the vivid phenomenal qualityredness such a one experiences does not entail the existence ofany ripe tomato or light—beam. The entity in which the vividphenomenal quality is instantiated, namely the sense—datum, or26seeing, or perception, or piece of visual field, or appearance,or whatever you want to call it, could exist non—veridically inthe absence of any ripe tomato or light—beam, as is shown by theillusions we all experience from time to time, and by the colorpatches we sometimes experience which aren’t of any representedthing outside our minds. If two qualitative properties are everontologically independent then they are always ontologically (ifnot causally) independent. Thus, if a tomato and an instance ofmy having a sense—datum (or whatever) depicting such a tomatohappen to coincide, it is an accident of history. The vividphenomenal quality redness is ontologically independent of thenon—mental objects the color relativist must claim itontologically depends on. Perhaps relative, dyadic colors existalso; if so they seem dispositional and reducible at least inpart to non—mental properties of physical objects.A second disanalogy between coloredness and goodness is thatwhile a clear distinction exists between intrinsic and extrinsicgoodness, to speak of something’s being intrinsically rather thanextrinsically colored is senseless. However, construed as anobjection this claim trades upon the ambiguity of the term‘senseless’. Certainly to speak of something’s beingextrinsically rather than intrinsically colored is senseless ifthis is taken to express the view that the term ‘extrinsicallycolored’ is assigned no sense in the natural language English.The term has no meaning in ordinary speech. However, the claimis implausible if taken to express the view that to apply the27intrinsic—extrinsic distinction to coloredness would be futilebecause the result would be nonsense. On the contrary, we couldeasily coin a sense for ‘extrinsic coloredness’ analogous to thesense of ‘extrinsic goodness’. We could hold that just as anextrinsically good thing, such as a cold beer on a hot day, is bydefinition an important cause of an intrinsically good thing,such as a pleasant, exhilarating taste sensation, anextrinsically colored thing, such as the firing of an opticnerve, is by definition an important cause of an intrinsicallycolored thing, such as a visual sense—datum. Few would find ituseful or interesting to talk about such extrinsically coloredthings, which explains the absence of ‘extrinsically colored’from ordinary English, but such talk would not be incoherent.Thus the intrinsic—extrinsic distinction grounds no relevantdisanalogy between coloredness and goodness.Coloredness thus interpreted escapes Aristotle’s dilemmathrough either fork (a) or fork (b). Consider these forks inrelation to coloredness:if (a) whatever (intrinsic) colorednessthere is in the world exists as a propertywhich supervenes on the natures of certainthings, e.g., red things or yellow things,then the things’ natures seem deprived ofany direct role in explaining thecoloredness of things, which seems wrong.Yet if (b) whatever coloredness exists28does not supervene on the natures butexists wholly within them, then since someof the natures have, qua colored things,nothing in common, a fortiori the natureshave no property of coloredness in common.If coloredness is a property then either(a) or (b), hence no property ofcoloredness exists.Which fork coloredness escapes through depends on whether colorentailment statements are analytic or synthetic. If color (andby hypothesis hence moral) entailment statements are analytic,then fork (b) is indicated; if synthetic, fork (a). Let us seehow this works.If color entailment propositions are analytic, then theproposition ‘All red things are colored’ is of the formx((Fx.Cx)-Cx) where ‘C’ attributes coloredness and ‘F’ attributessome property F which combined with coloredness makes redness.One might object that a property such as F is inconceivable. AMoorean moral objectivist might attribute this inconceivabilityto human cognitive deficiency, or, alternatively, admit theepistemic possibility that necessarily F cannot be conceived, butdeny that the truth of this possibility entails that no suchproperty exists. Instead, the Moorean moral objectivist mightappeal to Bennett’s notion of unabstractable differentiae:There are in fact unabstractabledifferentiae, or so it seems. If we ask,29‘What is the value of F such that redness iscolouredness—and—F?’ there seems to be noanswer (except the trivializing one ‘It isthe property of being either not coloured orelse red’). . . . we must drop the idea thatthere is a concept corresponding to everyproperty. (1984, p. 144—45)Husserl seems to agree (1970, p. 454). If a range of propertiessuch as F exists, then color entailment statements are analytic,and coloredness is a part of redness, yellowness, etc., thussatisfying the requirement of fork (b) of Aristotle’s dilemmathat coloredness not supervene on the natures of red things,yellow things, etc., but exist wholly within them.Simultaneously falsified is the dilemma’s assumption that some ofthe natures of red things, yellow things, etc. have, qua coloredthings, nothing in common. These things have in their naturescoloredness in common. Thus fork (b) is satisfied yet itstroublesome alleged consequence is avoided.Likewise, the Moorean moral objectivist might hold that arange of unabstractable differentiae exists so that each memberof the range combined with goodness forms a good, such aspleasure. Thus the impossibility of factoring out the concept ofa good such as pleasure into two concepts, one being the conceptof goodness, is accounted for without giving up the view that theproperty of pleasure (or whatever) factors into two properties,one being goodness. Fork (b) is satisfied, since goodness exists30wholly within the natures of good things, and the assumption isavoided that these things have, qua good things, nothing incommon. They have goodness itself in common.One might object that if no concepts correspond tounabstractable differentiae such as F, then a statement of theform x((Fx.Cx)-Cx) cannot express an entailment, since entailmentis conceptual necessitation. However, while F cannot beconceived, the form x((Fx.Cx)-’Cx) can be conceived, and this formrenders the statements in question conceptually necessary, henceentailments.Nor does it help to object that coloredness might turn outto be not a new sui eneris property, but a familiar property,such as being a light—emission of some minimal wavelength. It isnot crucial to Aristotle’s dilemma that the determinable bear nointernal relation to anything save (each of) its determinates.What is crucial is that the determinable bear some suitableinternal relation to its determinates. The fact that theproperty of being a light—emission of some minimal wave—lengthapparently lacks a suitable internal relation to redness andyellowness suggests that it is not the common part sought.Likewise, Aristotle’s dilemma would still be satisfied ifobjective goodness turned out to be cheddar cheese, so long asthe relevant internal relations obtained (though since theydon’t, it isn’t).If, on the other hand, color entailment statements aresynthetic, then the statement ‘All red things are colored’ is of31the form x(Rx-’Cx) where ‘R’ attributes redness, ‘C’ attributescoloredness, and ‘Rx’ is not further analyzable as ‘Bx.Cx’. Thissatisfies the requirement of fork (a) that coloredness exist notwithin the natures of red things, yellow things, etc., butseparately supervening on these natures. Since the truth of ‘Allred things are colored’ is conceptually necessary, a thing’sredness does play a direct role in explaining that thing’scoloredness. If the thing were not red but otherwise exactly asit is, we should have no explanation for its being colored. Asignificantly useful explanation must be of one thing as due toanother, and conceptual necessitation is an extremely strong,ontologically rather than linguistically based relation; hence,the redness of a thing’s explaining its coloredness is as aboutas strong and direct as an explanation can get. Thus theunwanted alleged consequence of the requirement of fork (a) isavoided.Similarly, the Moorean moral objectivist might view goodnessand any particular good as separate entities, thus satisfying therequirement of fork (a) that goodness exist not within thenatures of particular goods but as a property supervening uponthem. This view is consistent with the Moorean moralobjectivist’s further possible view that moral entailmentstatements such as ‘All pleasures are good’ are, as synthetic apriori truths, conceptually necessary, so that the pleasantnessof a thing explains its goodness about as strongly and directlyas anything can be explained. Thus the unwanted alleged32consequence of the requirement of fork (a) is avoided.The Moorean moral objectivist can escape Aristotle’s dilemmawithout being committed to either fork (a) or fork (b) alone.Moorean moral objectivism is secure so long as goods and goodnessare regarded as metaphysically analogous to colors andcoloredness, leaving for another day the question whether colorentailment statements and moral statements are analytic orsynthetic.3. Skepticism about determinable propertiesA critic of this solution to Aristotle’s dilemma forobjective goodness might allow that goodness and coloredness aremetaphysically analogous in the required way but hold that I haveremoved little of the mystery they are cloaked in and have merelyprovided a reason for doubting the existence of a property ofcoloredness as well as a property of goodness. I reply thatwhile determinable properties are somewhat mysterious, theirabsence would be more mysterious. Determinables account forresemblances between determinates. If determinables don’t exist,then either determinates don’t resemble each other, or theirresemblances exist but have other accounts.The first disjunct is false. Determinate properties doresemble each other in various ways. In particular, every colorproperty resembles every other color property in being a color.33Armstrong is right that “these recalcitrant resemblances seemobjective phenomena, demanding an ontological analysis” (1978,vol. 2, p. 116; on some other points regarding resemblance Iparaphrase Armstrong).The second disjunct, that these resemblances exist but haveother accounts, is dubious because both the chief rival accountsare dubious. To show this I review the five chief ontologicalaccounts of resemblance. Three turn out to posit determinables(including the two fairly plausible accounts already examined) —hence they are benign competitors for purposes of the moralobjectivist — and the other two are flawed.We have examined the first account of resemblance, thatresembling determinates do so in virtue of having common parts —call this the ‘common part theory’. We may not be directly awareof the complexity and partial identity of, say, the colors, butwe can hold that they are ontologically complex thoughepistemologically simple. Given that the colors contain a commonpart ontically expressing their color resemblance, clearly thecommon part is itself a property (a definite way of being, aone—place universal), and ‘coloredness’ a useful name for it. Ifthe colors are epistemologically simple, then their unsharedparts are unabstractable differentiae, perhaps complex.Coloredness thus construed is a determinable property, escapingAristotle’s dilemma via fork (b); likewise, says the Mooreanmoral objectivist, with goodness.We have also examined the second account of resemblance,34that two objects resembling each other instantiate the samedeterminable, each in virtue of its also instantiating a memberof a range of ontologically independent determinates. Thedeterminable is not part of but supervenes on the determinates,and statements asserting this supervenience are synthetic apriori. This account allows determinables to escape Aristotle’sdilemma via fork (a), and benefits from whatever plausibilityattaches to the notion of synthetic a priori propositions — whichis considerable (see §6.2, and, e.g., Sikora 1981). Thus whilethis account of resemblance competes with the common part theory,it is a benign competitor for purposes of the Moorean moralobjectivist.The third account of resemblance is that determinablesunderwrite resemblance but are second—order properties: commonproperties of the universals concerned. Bertrand Russell favorsthis view: “I should regard ‘red is a color’ as a genuinesubject—predicate proposition, assigning to the substance red thequality color” (1959, p. 171). We need not debate the merit ofthis account, since if false it poses no threat to the previoustwo benign and plausible accounts of resemblance, and if true itlets the second—order property of coloredness escape Aristotle’sdilemma via fork (a). Let’s see how this latter claim works.The two chief conditions for fork (a) are that: first,redness and coloredness (for example) be ontologicallyindependent entities, so that the latter can supervene on theformer, and; second, redness and coloredness (for example) be35such that the nature of the former explains the former’sinstantiating the latter.The first condition is satisfied by the third account ofresemblance, since the main point of positing substances is togive properties something ontologically distinct from themselvesin which to inhere. If determinate properties are taken assubstances, then their determinables must be ontologicallydistinct. This line requires that if supervenience is definedwith respect to a set of objects which if they share allproperties in the supervenience base will also share allproperties in the supervening set (e.g., Kim 1978, p. 152), thenan object must be deemed to possess albeit indirectly theproperties of its properties.The second condition is that redness and coloredness (forexample) must be such that the nature of the former explains theformer’s instantiating the latter. If, as it must, this thirdaccount of resemblance allows that the relevant propositions,such as that the substance red has coloredness, are knowable apriori, and since the ontological distinctness of, say, thesubstance red and the property coloredness makes the propositionssynthetic, it follows that, say, the nature of redness doesexplain red’s coloredness. Propositions knowable a priori may beknown to be true simply from the meanings or meaningful uses ofthe terms involved, and since the relevant propositions cannot beknown to be true from their logical forms as revealed by themeanings or meaningful uses of their terms, nothing is left to36explain the a prioricity of knowledge of their truth but thenatures of the referents of their property terms — i.e., qualonsrather than indexons — as revealed by the meanings or meaningfuluses of these terms. These natures must therefore stand in anappropriate ontologically necessary relation, this relationgrounding the explanation that, for instance, the nature ofredness yields of red’s coloredness. Thus the third account ofresemblance posits determinable properties which satisfy bothchief conditions of fork (a) and hence is benign for purposes ofthe Moorean moral objectivist.The fourth account of resemblance, discussed by Johnson(1921, ch. 11), is that the unity of a class of determinates suchas the shapes or colors is constituted and exhausted by the factthat co—determinates cannot simultaneously qualify the sameparticular. These classes are simply class—of—incompatibles.This account is inconsistent with Moorean moral objectivism,which intends much more by ‘Pleasure is good’ than that pleasureis incompatible with each of a set of other things, all mutuallyincompatible, and perhaps does not even intend theseincompatibilities. However, this account is gravely flawed.First, it is unclear that all classes of determinates whichshow by resemblance an intrinsic unity form aclass—of—incompatibles. Perhaps a thing can be both sweet andsour. Second, this account leaves unanswered the question whatabout the colors, and so on, makes them incompatible. Third, theview that co—determinates have nothing in common except their37incompatibility is phenomenologically implausible. The class ofshapes or the class of colors appear to have much more in commonthan that. They appear to display fully and equally a common,definite way of being. Fourth, determinates that don’t resembleeach other at all might end up under the same determinable simplybecause they could not inhere in the same particular. Thus thefourth account of resemblance is so flawed that it would be verymysterious if it were true and there were no determinables.The fourth account of resemblance might be improved,respecting the fourth objection to it, by specifying thatdeterminates could not fall under the same determinable unlessthey could conceivably individually inhere in the sameparticular. The plausibility of this amendment depends on whatcounts as a particular. If experiences count as particulars anda painful experience can’t conceivably be square, thenpainfulness and squareness would on the unamended fourth(Johnsonian) account resemble one another. However, it might beheld that an experience can’t conceivably be square whether it ispainful or not, hence that on the amended account painfulness andsquareness would not resemble each other — phenomenologically amore plausible result.However, this account of particulars was denied in §1.3. Onthe Spinozan account of substance therein endorsed, the full,underived ontic account of particularity in an experience (orwhatever) would be the extended region of time, and whateverother indexical fields mental events are located in, that the38experience’s qualitative properties were instantiated in.The amendment suggested above for repairing the fourthaccount of determinables, respecting the fourth objection to it,is on this Spinozan account of particularity dubious because itprevents determinables from forming hierarchical structures. If,e.g., the determinate reds fell under the determinable rednessbecause they could individually but not jointly inhere in therelevant indexical locations, and other determinates such as theyellows were excluded from redness because they could notpossibly inhere in these indexical locations, then the reds andyellows would be barred from falling under a common higher—leveldeterminable such as coloredness, for they could not individuallyinhere in any common indexical locations. Since a hierarchy ofdeterminables seems plausible, the amendment fails.The fifth account of resemblance is in terms of second—orderrelations between the determinates concerned. This account isunhelpful to Moorean moral objectivism since pointing at theresemblance between good things is only a way of demonstratingthe existence of a feature that each good would have had even hadit been the only good, hence lacking relations with any othergood. However, this account of resemblance suffers twoobjections.First, every class of resembling determinates isdistinguished not only by the resemblances between its membersbut also by the lack of this sort of resemblance between eachmember and each determinate not a member of that class.39Analogously to the proposed Moorean moral objectivist, thecoloredness objectivist holds that each color would have been acolor even had it been the only color. The single color wouldstill be distinguished by a respect in which it appeared parallelto a sound or smell, but not one. Positing the determinablecoloredness would account for this appearance, for it would beinstantiated, whereas positing a coloredness relation would not,for it would be uninstantiated. The coloredness relationistmight try to escape this objection by holding that a single colorwould instantiate the coloredness relation reflexively, but ifthat secured the single color as a color, then surely it wouldalso do so for each of the multiplicity of actual colors. Itwould then be redundant to posit non—reflexive instantiations ofthe coloredness relation, hence there would be no reason to viewthat universal as dyadic rather than monadic. Monadic universalsare simpler than dyadic, and parsimony favors the simpler of twoentities competing for and able to do the same explanatory job.Second, parsimony also counts against a coloredness relationinstantiated non—reflexively among many determinates. Supposethat n colors form a resemblance class. Each pair of colorsinstantiates the coloredness relation, for each member of eachpair resembles the other in the relevant way. Thus a dyadicuniversal is instantiated n(n—1)/2 times. Positing thedeterminable property of coloredness, however, yields a monadicuniversal instantiated n times. If n > 3, then n(n—l)/2 > n,hence the determinable property of coloredness is instantiated40fewer times than the coloredness relation, hence is favored byparsimony. If n = 3, then n(n—l)/2 = n, leaving parsimony tofavor a monadic over a dyadic universal for the class. If n = 2,then we must compare a dyadic universal instantiated once with amonadic universal instantiated twice, and parsimony favors typeeconomy over token economy, all else being equal. Since eachuniversal does the sante explanatory work, parsimony tells us ineach possible case of a multiplicity of deterntinates to reject acoloredness relation in favor of a coloredness determinable.A coloredness relation does not avoid the objection to acoloredness determinable that it is phenomenologically unclearthat anything about redness is identical to anything aboutblueness; it is phenomenologically no clearer that anything aboutthe red—green resemblance is identical to anything about theblue—orange resemblance. If coloredness relations are not takenas identical, then the unity of a class of first—order universalsis being analyzed in terms of the unity of a class ofsecond—order universals; a vicious regress follows.Thus of the five chief accounts of determinate resemblance,three posit determinables and the remaining two are seriouslyflawed. Therefore, when the determinable property skepticasserts that all determinable properties are mysterious, theMoorean moral objectivist can admit that this is to some extentso but maintain that since determinate class resemblances areobjective phenomena, demanding an ontological analysis, yet allproposed accounts are partially mysterious, mystery while41unwelcome is not fatal to these accounts. All three determinableproperty accounts escape Aristotle’s dilemma, including the leastmysterious account, the common part theory. While Moorean moralobjectivism in the long run owes philosophy an explanation ofthese remaining mysterious aspects, it is under no obligation toproduce this explanation at once; no more than set theory isobliged to produce at once a transparently clear answer toRussell’s paradox. Most substantive philosophical theoriesinvolve metaphysical presuppositions of lingering mystery.4. Some puzzles resolvedThe five rival accounts discussed are not epistemicallymutually exclusive competitors in the competition toontologically analyze color resemblance (hence also, if myanalogy works, moral resemblance) and the other determinate classresemblances. While Occam’s Razor urges us to accept only oneaccount, it is conceivable that investigation of variousdeterminate class resemblances might force us to accept more thanone, though no more than one for any given determinate classresemblance.A final possible disanalogy between goodness andcoloredness: one might object that while colored thingsmanifestly do lie on a continuum, good things could not lie on acontinuum as is suggested (though not entailed) by the analogy42between goodness and determinables such as coloredness, becausegood things such as pleasure, honor, and truth are so disparate.The Moorean moral objectivist can meet this objection byaccepting the constraint on moral theories that their sets ofintrinsically good things not have such disparate memberships.This constraint appears to damage Moorean moral objectivism,since things as disparate as pleasure, honor, and truth aremorally good yet are too disparate to lie on a continuum.However, some versions of Moorean moral objectivism accommodateboth suppositions with minimal strain on our intuitions. Ahedonist Moorean moral objectivist might hold that intrinsicallygood things are all and only pleasures, hence not too disparate,and that things such as pleasure, honor, and truth whiledisparate are all morally good, though some merely extrinsicallymorally good.Even if intrinsically good things do not form an unbrokencontinuum, this disparity would not entail that they are toodisparate to form a resemblance class of the required sort. Theymight still be fully and equally goods, just as red, yellow, andblue would be fully and equally colors even if they were the onlycolors. Some forms of moral pluralism are consistent with moralobjectivism (see §6.1).When Mackie asks what sort of property goodness is, and howit is connected with good—making properties, the moralobjectivist has a straightforward answer: goodness is adeterminable property, connected with good—making properties by43entailment.This interpretation of goodness perhaps explains somethingthat puzzled Moore, who says:many . . predicates of value . . . areintrinsic kinds of value, . . . yet none ofthem are intrinsic properties, in the sensein which such properties as “yellow” or theproperty of “being a state of pleasure” areintrinsic properties. (1922, p. 272)Nonetheless, such values are intrinsic in that they “share withintrinsic properties this characteristic of depending only on theintrinsic nature of what possesses them” (1922, p. 273). Thedifference between intrinsic predicates of value and of suchproperties seems to be “that intrinsic properties seem todescribe the intrinsic nature of what possesses them in a sensein which predicates of value never do” (1922, p. 274). Thisdifference is identified by the Moorean moral objectivist of OCUas the difference between determinate and determinableproperties. An object may be completely specified by itsdeterminate properties alone, in the sense that these propertiesplus any other properties entailed by these properties are allthe properties the object possesses; yet the object’sdeterminable properties are nonetheless intrinsic to it in thatthey depend on nothing else but their bearer object’s intrinsicnature. Determinable properties are reactive in that they are afunction of their bearer object’s determinate properties;44intrinsic values are also reactive in this way, hence it is inthis respect metaphysically convenient to interpret intrinsicvalues as determinables. Intrinsic values are thus distinguishedfrom determinate properties, which are not reactive. Thischaracteristic of non—reactivity would be what Moore sought whenhe said:it seems to me quite obvious that . . . theremust be some characteristic belonging tointrinsic properties which predicates ofvalue never have . . . only I can’t see whatit is. (1922, p. 274)Moore would not necessarily agree with this interpretation of histhought. Also, if determinates and determinables formhierarchies, so that a set of same—level deterininables can serveas a range of determinates for a higher—level determinable, thenthis account of non—reactivity must be correspondingly amended;e.g., by indexing reactivity and non—reactivity to appropriatelevels.Interpreting intrinsic goodness as a determinable alsoallows the Moorean moral objectivist to reply to the objection,mentioned by Moore, that “the goodness of God . . . cannot be aquality, which he might get or lose” (1959, p. 97). Taking‘quality’ to mean a monadic property, as opposed to a relation,the objection’s presupposition that each of an object’sproperties may be freely detached is false; a determinableproperty can’t be detached without pulling away its deterxninates45with it. If goodness is a determinable, then it couldn’t bedetached from God without pulling away the good things in God,which would be impossible given standard intellectual accounts ofGod’s nature; e.g., Brentano: “do we not think of God as thoughhe were the epitome of everything that is good, but raised to aninfinite degree?” (1969, p. 41). To remove such a god’s goodcharacteristics would be like removing the wool from a piece ofknitting.To conclude, the Moorean moral objectivist can escapeAristotle’s dilemma for a property of intrinsic goodness bycomparing goodness with coloredness and interpreting the relationbetween goodness and good things as a species of entailment,ontologically founded on a species of determinable—determinateproperty relation. A tougher job for the Moorean moralobjectivist is to explain why moral entailments are less easilygrasped than color entailments or simple mathematical entailments(this job can now be done — see Sleigh 1993, ch. 10). The mereavailability of an adequate metaphysical structure for a propertydoes not guarantee that the property exists; otherwise a theistmight assert the existence of a property of intrinsic holiness,to the utter frustration of the atheist.The first objection to OCtJ is thus refuted.46Chapter ThreeTHE CONCEPT OF GOODNESS1. The metaphysics of concepts of determinablesThe second objection to OCU is that even if a satisfactorypositive metaphysical account of the property of goodness can besupplied, as I claim to have done in ch. 2, the phenomenologicaldifficulty remains that most people are unable to directly detectsuch a property; the concept of goodness eludes them despitetheir following whatever procedures the Moorean moral objectivistrecommends for the inducement of this mental state. Indeed, somehave argued that no such concept is possible. In this thirdchapter a positive account of the concept of goodness will begiven, for this entity too is important to Moorean moralobj ectivism.It is epistemically possible for a property to exist withouta second entity, a concept of the property, also existing;Bennett (1984, p. 144—45) has suggested that some properties,unabstractable differentiae, may in fact have no correspondingconcepts. Therefore the nonexistence of a concept of goodnesswould not entail the nonexistence of a property of goodness.However, the absence of such a concept would be evidence against47the existence of the posited unconceived and inconceivableobject, just as the unconceivedness and inconceivability of aninteger between 2 and 3 are part of our evidence against theexistence of such an integer. A sufficient accumulation ofnegative evidence by ontic parsimony refutes the positedexistence of an object. Skeptics of Moorean moral objectivismare thus warranted in probing the possible nonexistence of theconcept of goodness.Brandt, for instance, is a skeptic about concepts ofobjective moral properties:although admitting that there are ethicalwords, we ask the nonnaturalist: In whatsense are there concepts of .unobservable, simple ethical properties .corresponding to these words? . . . We do nothave concepts of these properties . . . inthe sense of ability to conjure up an imageof them, as we can in the case of colors orshapes. (1959, p. 189—90)In reply, the Moorean moral objectivist might allow that noconcepts of moral properties exist as images, but hold that moralconcepts are other, non-imaging sorts of concept, for not allconcepts are images. This line must either identify the mode ofmoral concepts as some familiar non—imaging sort, or posit a suigeneris non—imaging conceptual mode unique to ethics. The lattervariant is assumed by ‘ethical intuitionisxn’, as the view has48been called that awareness of ethical facts requiresa different kind of awareness . . . neithersensory nor introspectual . . . , which thespecialists call “intuition”. . . . Thefundamental cognitive situation in morals isthat in which we intuit the rightness of aparticular action or the goodness of aparticular state of affairs. (Strawson 1949,p. 24)As with the property of goodness itself, the Moorean moralobjectivist must give a positive account of the disputed entity,for any plausible metaethic must supply fundamental entitiescohering satisfactorily with our human moral phenomenology.Though humans have much to learn in ethics, it is surely falsethat after their thousands of years of intellectual effort eventhe most elementary moral considerations still elude them.My account of moral concepts is derived from my account ofmoral properties. Goodness is a determinable property, so onewould expect the concept of goodness to in some ways resemble theconcepts of other determinable properties. So when Brandt askshow we can have concepts of moral properties, when we lack theability to conjure up images of them, as we can in the case ofcolors or shapes, the Moorean moral objectivist can reply that weconceive moral properties as we conceive many familiardeterminable properties, abstractly rather than by conjuring upimages. The properties of coloredness and triangularity yield49concepts graspable though abstracted from any images, e.g., ofparticular colors and triangles; likewise with goodness andgoods. This solution identifies the mode of moral concepts as afamiliar non—imaging sort, the abstract mode, rather than as asui generis conceptual mode unique to ethics (i.e., a mode uniqueto ‘non—natural’ qualities — as ethical properties have beentermed by Moore 1903, P. 13—14, to little advantage, I think, foras he suggests, the important questions for Moorean moralobjectivism are unaffected by the natural/non—naturaldistinction).The distinctness of such concepts of determinables fromimages, e.g. of determinates, does not entail that the former canbe entertained without the latter. Perhaps just as one cannotconceive of a colored object without a shape, even though theconcepts of the color and shape are distinct, one also cannotconceive of a determinable without a determinate, hence cannotconceive of goodness without a good, such as pleasure. However,Moorean moral objectivism is not committed to this possibility.The concept of goodness is distinct from any positiveemotional reaction to the concept; as Brentano puts it,the appearance of yirtue is more agreeablethan that of moral perversity. But theessential superiority of what is moral doesnot consist in this fact . .. ; it is,rather, a certain intrinsic correctnesswhich, however, may also have a certain50superiority in appearance. (1969, p. 10—11)One might hold that necessarily humans who grasp the concept ofgoodness have some positive disposition toward goodness, but sucha one would be wise to allow also that such humans might have asimultaneous negative disposition toward goodness. The exact wayin which goodness generates this positive disposition, whichunder some circumstances is sufficient to motivate action, is toocomplex to discuss here, but agrees almost completely withNagel’s plausible account of necessarily motivating objectivemoral truths (1970). Suffice it to say that dispositions, asreactions to items, are ontologically independent from them,though some dispositions also seem metaphysically necessarilyconnected to the items they are reactions to, analogously toeffects being metaphysically necessarily connected to theircauses (see also Goldstein 1989, p. 260). Besides this directmoral motivation, an agent may have other, indirect motivationsfor doing right acts, e.g., a desire to be thought helpful.As with the property of goodness, a comparatively clearpositive account of the concept of goodness suggests solutions tosome standing objections to the existence of this moralconceptual entity. For instance, Strawson objects thatyou will scarcely say that ethical intuitionsare infallible; for ethical disagreements maysurvive the resolution of factualdisagreements. . . . So your use of thelanguage of “unanalysable predicates ascribed51in moral judgment to particular actions andstates of affairs” leads to contradiction.For to call such a judgment “non—infallible”would be meaningless unless there were someway of checking it; of confirming orconfuting it, by producing evidence for oragainst it. But I have just shown that youraccount of these judgments is incompatiblewith the possibility of producing evidencefor or against them. So, if your account istrue, these judgments are both corrigible andincorrigible; and this is absurd. (1949, p.27)OCU’s version of Moorean moral objectivism does not assume thatgoodness is a simple, unanalyzable property, though this isconsistent with available evidence; however, this version ofmoral objectivism does assume that either goodness or one of itsparts is a simple, unanalyzable, ethical property. OCTJ’s moralobjectivism is therefore subject to a correspondingly modifiedvariant of Strawson’s objection.This objection, even thus modified, is subject to a damagingtrilernma. If the premise that ethical disagreements may survivethe resolution of factual disagreements is taken to mean thatethical claims are not truth—functional, then the question isbegged whether goodness is an objective property, claims ofinstantiation of which would either correspond with the world or52not, hence by the correspondence theory of truth be either trueor false, respectively. Strawson’s objection would thus bedisqualified from use against OCU’s moral objectivism.If, on the other hand, the premise that ethicaldisagreements may survive the resolution of factual disagreementsis taken to mean that ethical claims are not a posterioripropositions, then the moral objectivist of OCU can still holdthat ethical claims are a priori propositions, disagreement aboutwhich is possible. As explained in ch. 2, ethical claims of thefoundational sort in question though a priori may turn out to beeither analytic or synthetic, depending on the metaphysicalnature of determinables. While a priori propositions arenoncontingent, humans are not universally able to discern theirtruth—values. Disagreement among reasonable people is common inthis regard, particularly concerning the more complex ofnoncontingent propositions. Moreover, such disagreement can bereduced by consideration of evidence; conceptual analysis is apowerful evidence—gathering tool, as reflected in human progressin a priori fields such as mathematics and logic. There is noobvious reason why such disagreement should not arise regardingdeterminables. One may wonder whether a given entity (e.g., atransparent object) has the determinable property of coloredness,and one may wonder whether a given entity (e.g., a truth—telling)has the determinable property of goodness. Thus one could inthis way call an objective moral judgment corrigible andnon—infallible while agreeing that ethical claims are not factual53claims (a reply overlooked also by Hare — 1981, P. 69—70).If, on the third hand, the premise that ethicaldisagreements may survive the resolution of factual disagreementsis taken to mean that ethical claims are not entailed bynon—ethical truths, then a subdilemrna is generated.If the premise that ethical claims are not entailed bynon—ethical truths is taken in the sense merely that ethics andother subjects are separate subjects, then the moral objectivistof OCU can assent. No one expects truths about what is coloredor smelly to directly resolve questions about what is noisy; norshould one expect truths about these properties to directlyresolve questions about what is good. The various propertieseach ontically express a unique shared character aspect of therelevant range of objects. While propositions about eachproperty have bearing on propositions about each other propertyto at least some extent via the vast web of belief, thisinfluence may be subtle and attenuated; no direct relation ofentailment need connect a given pair of properties. Evengranting Strawson the dubious premise (see Bradley and Swartz1979, p. 167) that there is evidence available for or against anynon—infallible claim, one can simply point to the publishedevidence, for or against ethical claims, available within ethicsvia conceptual analysis. Part, though not necessarily all oreven most, of this evidence consists of the coherence of a givenethical intuition with other related ethical intuitions, such asthose of other people on the same topic. Strawson objects to54this line:it is no good saying that, after all, theethical judgments of other people (or yourown at other times) may corroborate your ownpresent judgment. They may agree with it:but their agreement strengthens theprobability of your judgment only on theassumption that their moral intuitions tendon the whole to be correct. But the onlypossible evidence for the existence of atendency to have correct intuitions is thecorrectness of actual intuitions. And it isprecisely the correctness of actualintuitions for which we are seeking evidence,and failing to find it. (1949, p. 27)Thus Strawson claims that the case for the correctness of anethical intuition is no stronger than the evidence supplied fromoutside ethics. However, this claim is mere Cartesian skepticismapplied to a particular human faculty, that by which we intuitethical truths (and possibly other things, for we have foundwanting the assumption that ethical intuitions require a suigeneris conceptual mode). Cartesian skepticism is the view thatwe should doubt everything that can be doubted, i.e., that hasnot been proven from certain knowledge. But since no foundationof certain knowledge is available, everything must be doubted,including sensory or other epistemic intuitions, hence we have no55clue as to the nature of the world. The falsity of thisconclusion suggests that we should take the argument as areductio ad absurdum of its premise that we should doubteverything that can be doubted. Occam’s Razor suggests thepremise’s contrary, that we should try to believe everything thatour intuitions tell us to believe (see Rescher 1973, p. 54—64).The simplest explanation that fits the facts about why Chen seemsto see a tree is that there is a tree and Chen is seeing it. Onreflection we cannot accept all these potential beliefs, for someare mutually inconsistent; hence epistemology must winnow, amend,and supplement. This applies equally to the faculties by whichwe smell things, see things, feel things, and intuit things’moral qualities. Occam’s Razor presumes against the existence ofmoral and all other intuitions, but once granted their existenceit presumes in favor of their veridicality. This presumption isnot forceless, either, for the existence of intuitions regardingthe various non—ethical states of affairs generally is admitted(though with reservations, some warranted, e.g., Hare 1981, p.9). Thus any given moral intuition is evidentially supported byitself, and perhaps also by cohering other moral intuitions andnon—ethical evidence. The first two of these classes ofintuition need not be forceless, as Strawson suggests they must;in fact, the first class is never forceless (see Baylis 1967, p.447).If, however, the premise that ethical claims are notentailed by non—ethical truths is taken in the sense (probably56the sense Strawson intended) that claims about goodness, e.g.,that x is good, are not entailed by truths about goods, e.g.,that x is a pleasant experience, then again the question isbegged against OCU’s moral objectivism, disqualifying Strawson’sobjection from use against it. While different properties neednot stand in relations of entailment, they sometimes do; e.g., ifx is square then x is rectangular, or, if x is a sound then x hasa pitch. One should not assume without argument that differentproperties exactly one of which is ethically fundamental areunrelated by entailment (aside from the weak but everpresentpresumption by ontic parsimony against entities in general,including entailments).Since the premise expressed by the locution that ethicaldisagreements may survive factual disagreements has to be takensome way or other, yet none of the various ways discussed, each arendering, from a representative set, of a philosophicallypopular way of taking such phrases, yields an interpretation ofthe objection that damages moral objectivism, the objectionfails.Strawson’s objection, developed in the somewhat positivist,linguistically—oriented climate of mid—twentieth centuryphilosophy, is complex and could be analyzed at length. However,it suffices for the moral objectivist of OCU to show as abovethat if ethical intuitions are taken to use as theirintrinsically ethical parts mental items in a conceptual mode ofa familiar non—imaging sort, the abstract mode, rather than as in57a sui generis conceptual mode unique to ethics, then Strawson’spuzzles are seen to apply equally to goodness and to manyfamiliar determinable properties, for many familiar determinablesare conceivable only abstractly (e.g., coloredness andtriangularity). While the existence of even non—ethicaldeterminables is uncertain, their metaphysics and phenomenologyare better—understood than those of ethical properties, allowingmany people to get a satisfactory intuitive idea how Strawson’sobjection might be misplaced against determinables, and how itwould hence be misplaced against OCU’s moral objectivism. Aswith the property of goodness, the existence of the concept ofgoodness is on metaphysical grounds no less probable than theexistence of many familiar determinable properties and theirconcepts. Since no one has provided a better alternativeexplanation of resemblance classes of determinate properties, noone has shown that a world without determinables is lessmysterious than a world with determinables. OCU’s moralobjectivism, including the positing of a concept of goodness, istherefore a currently viable metaethic, available for use byutilitarianism and other comprehensive ethical theories.It might be wondered why, if OCU does not hold that moraljudgments are entertained in a sui generis conceptual mode, asintuitionism holds, I have bothered to address Strawson’sobjection to intuitionism. The reason is that I regard thedistinction between intuitions of familiar mode and sui generisintuitions as a distinction of little practical use, just as I58regard the natural/non—natural distinction (and for that matter,the monism/dualism distinction — see §1.3). The preciseconceptual mode by which any concept is entertained is in somerespects unique and in other respects like other conceptualmodes. Occam’s Razor counts against modes or other items positedto be unique in too many respects, so as not to multiplydeterminables unnecessarily. However, vague impressions of justhow unique an item is are almost as likely to reflect unconsciousprejudices as to reflect anything of much epistemic weight. Forinstance, early estimates of humans’ biological uniqueness ledmany 18th century naturalists to classify humans as the soleoccupants of a taxonomic kingdom parallel to the kingdoms ofanimals and plants (see Gould 1993, p. 29), whereas modernopinion is that humans and other animals are not only members ofthe same kingdom, but also, in the case of, e.g., humans andchimpanzees, members of the same phylum, order, and family,differing only at the generic level (Szalay and Delson 1979).While vague impressions are better than nothing as evidence(i.e., are ‘data’ sensu coherentist epistemology — Rescher 1973,p. 57), in the present case we are not so desperate, for we havemany analytical tools and much detailed empirical evidence at ourdisposal. We are thus better off spending less intellectualenergy examining these vague impressions, and more energyexamining the conceptual mode or modes actually or potentiallyused in ethical thinking, and using these particular results toascertain the uniqueness of such modes, and whether they are59morally objective. This is not to suggest that we should beepistemic slaves to our traditional moral concepts or language,hence do other than “frame our questions clearly and then go outto find answers, letting the chips fall where they may”, asBrandt suggests intuitionists mistakenly do other than (1979, p.3); rather, it is to suggest that when answering theclearly—framed question how closely moral concepts, includingobjectivist concepts if any, resemble other concepts, we arebetter off to examine such concepts than to examine (with as muchenergy, for all data have some weight) our prior intuitions abouthow close this resemblance might be. Brandt’s objection succeedsnot against modern intuitionists of the epistemologicalcoherentist persuasion (sleigh 1984), but against traditionalintuitionists such as Ross (1967, p. 269), who place unwarrantedweight on human moral intuitions, largely due to their mistakenlythinking that intuitions from outside ethics can have no bearingon ethical matters (e.g., Ross 1967, p. 277). Positions midwaybetween these extremes, e.g., Rawls (1971, p. 34—49), aredeficient to the extent that they embrace this mistake. Many whoare not intuitionists nonetheless treat traditional intuitionismas if it were the best or only version of intuitionism, hencejoin with Brandt in overlooking the strongest variant of thetheory they dismiss (e.g., Hare 1981, p. 12, 40).2. Entertaining concepts of determinables60A skeptic of OCU’s moral objectivism might allow thatdeterminables, whether any exist or not, have beensupplied with a fairly comprehensible metaphysical structure,hence allow that a property of goodness construed as adeterminable cannot be ruled out on metaphysical grounds.However, the skeptic might ask why, if this property exists andhas, as do so many properties, a concept available for sentiententertainment, more humans have not claimed to have entertainedsuch a concept:If our inability to identify ethicalqualities is the result of our looking forthe wrong kind of thing, once this error isexposed and we stop making it, thedifficulty should be overcome. But it isnot. . . . Whether or not we are aware ofnon—natural qualities, we are certainlynever cognizant of our awareness, and thismust also be explained. (Sikora 1962, p.63)Nowadays few people, even, as will be seen, among the best moralphilosophers, do report entertaining the concept of goodness. Aswith reports of UFO abductions, if a striking event is reportedseldom enough, one may reasonably wonder whether these fewreports are mistaken, allowing ontic parsimony to snip awayanother unnecessary entity.Regarding the contemporary scarcity of reports of61entertainings of a concept of goodness, it should be noted thatwhile nowadays Moorean moral objectivism is philosophicallyunpopular, only a few decades ago Moore’s influence was great,and Moorean moral objectivism prospered, with reports of conceptsof objective moral entities being common. Since humans are soeasily influenced by popular theory on this matter, we should notpay undue attention to the concept’s contemporary reportage ratewhen judging whether the concept exists. The exact epistemicsignificance of fluctuating popularity in an entity’s concept mayin principle be scientifically ascertained (Sleigh 1993, ch. 10);however, such research has not yet been completed respecting theconcept of goodness.Philosophers who aren’t aware of the concept of such aproperty aren’t expected to take it on faith that others areaware of it, they are expected to take it on evidence. For them,goodness and its concept are purely theoretical entities, such asare commonly posited in many disciplines. For no obvious goodreason, purely theoretical entities have been posited frequentlyin, e.g., physics, with scarcely a murmur of dissent fromphilosophers, whereas extraordinarily heavy weather has been madeof the same move in metaethics. This double standard should beabandoned, presumably by being to some degree less receptive toentity—positing in physics and to some degree more receptive toentity—positing in metaethics. OCU’s moral objectivism is highlyparsimonious, positing only goodness and its negative twin (ornegative half), badness, and the usual attendant entities, such62as the concepts of goodness and badness. It is difficult to seehow a metaethic could posit fewer entities yet still get theexplanatory job done; a metaethic that posited no entitieswhatever would be as needful of scrutiny as a theory of physicsthat posited no atoms, molecules, or other microparticleswhatever.Regarding the rectification of some error aboutidentification of the concept of goodness; previous purportedrectifications may not have worked, yet if goodness is adeterminable property then this element of a full appropriaterectification has not yet been tried widely. Now that thiselement has entered the philosophical literature (Sleigh 1992),changes are possible.Since the quantity of reports of entertainings of theconcept of goodness fluctuates beyond much current epistemicusefulness, let us turn to their quality. The majority of thebest moral philosophers, from Aristotle to Mackie, do not reportsuch a concept. However, some excellent thinkers do reportentertaining concepts of objective moral properties. Mostprominently, Moore writes thatWhenever (one) thinks of ‘intrinsic value’or ‘intrinsic worth’, or says that a thing‘ought to exist’, he has before his mindthe unique object — the unique property ofthings — which I mean by ‘good’.Everybody is . . . aware of this notion,63although he may never become aware at allthat it is different from other notions ofwhich he is also aware. (1903, P. 17)Kant is less clear in his endorsement of concepts of objectivemoral entities. He speaks as if value were an objective,ontologically independent component of the universe, readilyconceived; he says thata good will . . . would . . . shine like ajewel for its own sake as something whichhas its full value in itself. . . . thisIdea of the absolute value of a mere will(1948, p. 62)and thatit is impossible to conceive anything atall in the world, or even out of it, whichcan be taken as good without qualification,except a good will. (1948, p. 61)and he uses in an existence claim (the German equivalent of) thephrase “the concept of the good” (1948, p. 81). However, Kantalso uses the phrase “the concept of duty” (1948, p. 65), butlater cautions: “we leave it unsettled whether what we call dutymay not be an empty concept” (1948, p. 88). Kant’s position thusseems to be that there appears to be a concept of goodness, butthat appearances can deceive in such matters (a position thatagrees with my personal phenomenology; if it turned out that noconcepts of intrinsic ethical entities exist and that no64conceptless entities exist, then I would retreat to a roughlyNagellian version of non—entifying moral objectivism, sufficientto ground hedonic utilitarianism). However, it is uncertain thatKant meant anything more by ‘the concept of the good’ than ‘thecorrect conceptual analysis of goodness’, which is consistentwith a naturalistic or morally skeptical reduction of the conceptof the good to nonmoral concepts, i.e., those other than ofobjective intrinsic worth.The examples of Moore and his ilk suffice to show thatwhether or not a concept of goodness exists, a concept does existthat is either the concept of goodness or else a concept sosimilar to the concept of goodness as to be able to fool someexperts. If the latter, then the place in our ontology for aconcept of goodness is still vacant, hence the concept ofgoodness is in this sense an ‘empty concept’. A mere handful ofexamples such as Moore suffice to show that a goodness—seemingconcept exists, because if we assume, as seems true, that thesepersons were sincere, articulate, sane, intelligent, benevolent,well—educated reporters of their mental states, who had givensustained hard thought to moral matters including moralphenomenology, then only some exceedingly remotely causallypossible error could neutralize the tendency of suchintrospective reports to be veridical. It is as likely thatJulius Caesar came and conquered but did not see as that Moorewrote the words quoted above but did not entertain concepts whichhe examined carefully and took to be concepts of fundamental,65objective, intrinsic moral entities. Examples of a few suchpeople multiply the probability of truth beyond reasonabledoubt — these people may or may not actually have entertainedobjective moral concepts, but they definitely thought that theyhad.It might be objected to this line that evidence of similarstrength is available for the existence of concepts of, orconcepts seeming to be concepts of, all sorts of dubiousproperty, from intrinsic holiness to intrinsic vital force. Themoral objectivist of OCU may respond by accepting the existencesof all these other concepts. Accepting such concepts is far fromaccepting any associated properties or their genuine concepts.Most such concepts presumably will turn out not to be of thepurported properties, so that the moral objectivist of OCU needaccept only a few of the dubious properties — the few that turnout to exist.Against this sort of response Strawson says that if somepeople don’t associate any concepts with words such as ‘good’,but others do, then such words “have quite a different meaningfor one set of people from the meaning which they have foranother set. But neither of us believes this” (1949, p. 25).However, Strawson gives no reason for not believing that suchwords are being used ambiguously. In fact, contra Strawson thisexplanation is very plausible, for ambiguities of this sort arevery common. Etymology shows that word meanings change greatlyover time, and in many cases a mechanism of change is that66concepts are gained or lost from word meanings. Such linguisticchange occurs gradually, often with more than one meaning inpopular use at the same time. As words slowly bifurcate,competing meanings diverge, instantiating serially many degreesof similarity, until separation occurs, producing two words thatmay or may not be spelt the same or pronounced the same, butdefinitely don’t mean the same.. At a certain point in thisprocess, a human linguistic group may find that some of itsmembers are using one meaning for a word, some are using another,and neither set wants to relinquish the claim of uniqueappropriateness for its favored meaning for some key contexts.Such a position has been reached in English for the word ‘line’,by which some mean a straight edge and some mean a straight orcurved edge; the truth or falsity of the parallel line postulate(that for a given point and a line not containing the point, oneand only one line containing the point is parallel to the other,given line) depends on which meaning is used. In a disagreementinvolving a folk—theoretic word, often one party will take theword to mean the concept of a substantive folk—theoretic entity,and the other party will take it to mean nothing, for they thinknot only that no such entity exists or could exist, but also thatno suitable coherent concept is available, not even aself—contradictory concept. For instance, the concept of asquare circle is, though incoherent because self—contradictory,nonetheless composed of coherent elements; by contrast, theconcept of a soul is, to me, a vacuous opacity — not an67incoherent composition of coherent elements, not even a blurrymess. I can grasp nothing of the alleged concept; the word‘soul’ seems a mere place—holder in a metaphysical theory. Tosome people, e.g. Hare 1981, p. 83—86, it seems that ‘good’ asMoorean moral objectivists use it is just a vacuous place—holderin a metaethical theory. Thus such people and Moorean moralobjectivists mean quite different things by ‘good’, hence ‘good’is ambiguous. A like ambiguity seems to obtain of thefolk—theoretic word ‘mind’, which most of us take to mean theconcept of a familiar substantive entity, but which Ryle seems totake to mean, in this context, no concept, for he thinks mentalstates to be entities which not only don’t and couldn’t exist,but are vacuous “occult” entities (1949, ch. 1; p. 51).Strawson’s thinking it implausible that such ambiguity couldaccount for a moral case such as that of ‘good’ is perhapsattributable to the fact that both parties know how to use moralwords such as ‘good’, and may use them almost identically in mostcontexts. To the extent that meaning is equated with use, as itoften was in the philosophical climate prevailing when Strawsonwrote, it will seem implausible that people could use a wordsimilarly yet mean it quite differently.It should be remembered that theories contain concepts, thatthere are many times more false theories than true theories, andthat in many cases the false theories are false because some oftheir concepts are confused. These confused concepts are realand legion, and it should surprise no one if such a concept turns68up in ethics posing as the concept of goodness. Thus skeptics ofMoorean moral objectivism misplace their efforts if they holdthat Moorean moral objectivists typically don’t really entertainany concept when they think that they are entertaining theconcept of goodness. Instead, any such skeptics of Moorean moral.objectivism should admit that Moorean moral objectivists usuallydo at such times entertain a concept seeming to be a token of theconcept of goodness, and concentrate on proving that suchentertained concepts are not tokens of the concept of goodness.Regarding the question whether we are cognizant of ourawareness of ethical properties: while OCTJ’s moral objectivismentails nothing on this issue, some Moorean moral objectivists infact claim to be cognizant of their entertainings of concepts ofobjective moral entities. However, this cognizance seems farfrom inevitable; it is curious how some avowed skeptics ofMoorean moral objectivism turn out to advocate analyses ofpersonal choices and their grounds that closely parallel thestructures of Moorean moral objectivism, leading one to wonderwhether these skeptics may not have entertained concepts ofobjective moral entities without being cognizant of the fact.Mackie, for instance, is famous for his moral skepticism,claimingthat there are no objective values, . .that no substantive moral conclusions orserious constraints on moral views can bederived from either the meanings of moral69terms or the logic of moral discourse.(1977, p. 105)Yet Mackie also holds that hypothetical imperatives, statementsof the form ‘If you want X you ought to do Y’, are true; e.g.,if someone wants to get to London by twelveo’clock, and the only available means oftransport that will get him there is theten—twenty train, and catching this trainwill not conflict with any equally strongdesires or purposes that he has, then heought to, indeed must, catch the ten—twenty.(1977, p. 65—66)Mackie is well aware of Hume’s dictum (1888, p. 469—70, =Treatise, Book 3, Pt. 1, sec. 1) that one cannot derive an‘ought’ from an ‘is’ because doing so would involve introducingsome new relation joining agent and act, this new relation beingdeduced from others entirely different from it, which seemsinconceivable. However, Mackie claims that in his derivationno ‘new relation’ is involved. ‘Ought’ .says that the agent has a reason for doingsomething, but his desires along with thesecausal relations constitute the reason.(1977, p. 66)If so, then Mackie is a moral objectivist by his own account. Itis objective fact that an agent has the desires it has and thatthe relevant causal relations obtain; since these constitute,70apparently noncontingently, a reason for the agent to act, andsince some but not all of an agent’s available acts receive thisobjective endorsement, a substantive, serious constraint on agentconduct objectively exists. Since agent conduct and constraintsthereon are the bulk of the subject of morality, Mackie hasproposed what he calls an objective morality.Let’s distinguish the strongest general version of moralobjectivism, which posits at least one objective, intrinsicallyethical entity, from the less strong general version of moralobjectivism, which does not, but nonetheless holds that facts canbe objective reasons for action, so that some acts arerecommended by rationality over others; call the former version‘entifying moral objectivism’ and the latter version‘non—entifying moral objectivism’. Entifying and non—entifyingmoral objectivism are jointly exhaustive of ethical theories bywhich facts unequivocally demand certain choices of moral agents;call these ethical theories versions of ‘strong moralobjectivism’. However, there are proponents of weaker theoriesby which moral agents should rationally follow certain objectiveprocedures in coming to moral decisions, though the outcome ofthese procedures is unspecified; call these theories versions of‘weak moral objectivism’. ‘Moorean moral objectivism’ isformally defined as the form of entifying moral objectivism thatposits a Moorean property of objective intrinsic goodness, itsnegative twin (or negative half), badness, perhaps theirconcepts, and no other underived ethical entity. ‘OCU’s moral71objectivism’ continues to refer to the version of Moorean moralobjectivism that includes the metaethical detail argued for inthis thesis. ‘Moral objectivism’ will refer to the union set ofstrong and weak moral objectivisms. Moral skeptics will beidentified particularly, e.g., ‘skeptics of entifying moralobjectivism’.Mackie’s position is a form of non—entifying moralobjectivism, not all that different from OCU’s form of entifyingmoral objectivism; only three major differences separate them.First, while Mackie holds that only a given agent’s presentdesires generate reasons for that agent to act, OCU holds thatany sentient’s desire at any time may generate reasons for thegiven agent to act (even possible future desires, e.g., as whenJones would desire to eat this chocolate if I revealed it andgave it to Jones; on possibility and moral judgments, see Moore1912, p. 27; Sikora 1978, p. 126). Mackie discusses part of thisalternative:Do the desires . . . of other people .constitute a reason for me to do something,if I can, or to try to do something tosatisfy those desires if werecognize this as a further class of reasons,independent of any desire that I now have tohelp these other people, we are againbringing in the requirements of somethinglike an institution: an established way of72thinking, a moral tradition. . . . However, the conclusion that is firmlyestablished will be only of the form: Thisinstitution requires such and such an action.If we move to a prescriptive interpretation,we shall be speaking within the institution.But nothing logically commits us to doingso. . . . (1977, p. 78—79)However, this defense fails since the notion of a metaphysicallyunitary ‘I’ being at once the subject of a present desire, theknower that act Y will bring about state of affairs X, the actorwho might do Y, and the enjoyer who would be satisfied if Xobtained, is equally an institution to which no one is logicallycommitted. What Mackie identifies as a reason of prudence (areason of no benevolence; of the purest self—interest) is in facta relation between desires, cognitions, actions, andsatisfactions that are ontologically independent and notobviously morally connected.First, those elements of a Mackiean reason of prudence thatare cotemporal are not thereby bound to each other morally, nomore than one’s present self is morally bound to one’s futureself. As Mackie says, regarding the latter case,We can indeed say that . . . (someone) nowhas a (prudential) reason for an action whichwill tend to satisfy not any desire which henow has, not even a present desire that his73future desires be fulfilled, but only adesire which he knows he will have later• . . and that (other things being equal) heought to act. . . • But in saying this weare leaning on our concept of the identity ofa person through time and the associatedexpectation that a human being will behave asa fairly coherent purposive unit overtime. . . . Human beings are more likely toflourish if they show such purposivecoherence over time, so that it is notsurprising that we have this useful clusterof concepts and expectations. (1977, p. 78)For similar evolutionary reasons, it is no more surprising thatwe have a useful cluster of concepts and expectations about theidentity of a time—slice of co—accessed mental states. However,it is logically possible for some mental states to regardthemselves as morally dissociated from other, co—accessed mentalstates (e.g., Nagel 1970, p. 75, 126), and apparently causallypossible as well, for our mental hospitals yield apparentexamples of such sets of awarenesses lacking coherentpurposiveness. It is a mere useful social fiction that withoutexception we humans are psychologically cohesive persons. Eventhose minds which are psychologically cohesive thereby manifestonly certain familiar universals such as resemblance betweentheir composing elements; any inference beyond these familiar74universals, e.g., to relations of duty or reasons to act, needsexplanation.Thus Mackie cannot endorse what he identifies as reasons ofprudence yet criticize those who think that the desires of othersor of one’s future self may help form reasons for one’s action;there is no need for the latter views to assume the warrant ofany institution, no more than Mackie need assume the warrant ofany institution when positing objective prudential reasons forone of two co—accessable mental elements to will the fulfilmentof the desire of the other. An institutional warrant would, asMackie says, fail, but objective reasons exist anyway, on other,better warrant, between co—accessed mental elements and betweenmy action and others’ desires.It might be objected that there’s nothing odd about treatingco—accessed mental elements as morally connected, for there’snothing imaginary about the co—accession, or selfhood, of thebundle of mental elements. However, the mental elements bundledtogether in one co—accessed set of awarenesses are oftendisparate and conflicting. Consider one who, having been raisedin a prejudiced society, finds mentally accessable some negativedesire toward a certain minority. This person might repudiatesuch a desire but be unable to eradicate it from consciousness.If Mackie is right, then such a person is stuck with theprejudiced desire, and out of self—interest must act to satisfyit. But why cannot such a person say ‘that desire is no morepart of me than a shrapnel fragment; it’s in me but it’s other,75so its satisfaction is no gain for me’?Second, and less controversially, not all of the elements ofa Mackiean reason of prudence are coteirtporal. The desire andcognition take place before the action, which takes place beforethe satisfaction; hence Mackie’s reasons of prudence also involvemoral relations crossing temporal gaps. This secondconsideration is by itself sufficient to show that Mackie’sreasons for action are the same as altruistic reasons for actionin their status as objective, intrinsic, rational moralrecommendations, and that Mackie’s metaethic is a version ofstrong moral objectivism.A determined skeptic of strong moral objectivism mightaccommodate this objection by renouncing all objective reasonsfor action, even those reasons seemingly generated by desiresentertained here and now (i.e., reasons either justified bylogically antecedent ‘unmotivated’ or ‘basic’ desires, orexplanatorily presupposed by ‘motivated’ or ‘nonbasic’ desires —see Nagel 1970, p. 29; Foley 1978, p. 69; Gewirth 1978, §2.1;Dancy 1991, p. 415—16), thus rejecting even Mackiean reasons foraction. However, strong moral objectivists such as Moore andMackie have a deep contrary intuition that there are such thingsas objective reasons for action, even though such strong moralobjectivists are divided on questions such as whose desires helpform reasons for whose action. The conceptual relations positedin these reasons seem knowable a priori, hence to thoseintellectually equipped, self—evident; to some, perhaps Mackie,76the objectivity of such action—guiding considerations is sotransparent that it is taken without question, withoutexplanation, without cognizance of the awareness. It is easy forhumans to incorrectly interpret their reasons for action (Nagel1970, p. 82).Another moral philosopher who is in practice an endorser ofprudential reasons, hence a strong moral objectivist, is Rawis,who claims that:each person must decide by rationalreflection what constitutes his good, thatis, the system of ends which it is rationalfor him to pursue. . . . (1971, P. 12),that:we have . . . substituted for an ethicaljudgment a judgment of rational prudence.(1971, p. 44),and that:it is an open question whether the principleof utility would be acknowledged. Offhand,it hardly seems likely. . . . Since each(person) desires to protect his interests,• . no one has a reason to acquiesce in anenduring loss for himself in order to bringabout a greater net balance of satisfaction.(1971, p. 14)The second major difference between Mackie’s moral77objectivism and OCU’s is that while Mackie does not stress theimportance of desire—satisfaction produced when the causal meansto the desired state of affairs is effected, OCU’s moralobjectivism does: the reason the agent has for doing Y, whichwill bring about X, which is desired, is that X will (it ishoped) satisfy the desire. The unique importance ofdesire—satisfaction (i.e., pleasure — see §6.1) may be shown bythought—experiments. For instance, consider one who desires tomeet a certain movie star; does a friend who knows that the moviestar is obnoxious in person have reason to introduce the fan tothe star if the opportunity arises? Such a meeting would producethe object of desire but not satisfaction; the desire would notbe quenched appropriately. Mackie’s line is that examining suchvariations is pointless since there are no objective moralrequirements which might or might not be fulfilled (1977, p. 77);however, it is just as easy to hold that since examining suchvariations is not pointless, there are objective moralrequirements which might or might not be fulfilled. Perhaps onehas more reason to have love and lose than never to love at all,and surely one has more reason still to love and win.The third major difference between Mackie’s moralobjectivism and OCU’s is that while OCU holds that the objectsgrounding objective reasons resemble each other so as to form anatural kind, and interprets such resemblance classes asdeterminates whose resemblance finds ontic expression as adeterminable property, Mackie holds his set of objects grounding78objective reasons to be merely a family resemblance set, hencewith no determinable property instantiated by each member of theset. I have already given OCU’s evidence for the existence ofdeterminables, and the rest of this third difference is merelypossible disagreement about whether desire—satisfactions form aunified resemblance class, like the colors, or a mere familyresemblance class, like cats.I have explained this parallel between MacIde’s moral viewand OCU in some detail in the hope that moral skeptics of allsorts may see how entifying moral objectivists see themselves,not as like theists in the age of Darwin and the atom, but asrealists who, like many, see objective reasons for action in theworld, and, like not so many, require theories to assign entitiesto any objective structure posited. It is not at all clear whichmetaethical view is true; what is clear is that the metaphysicalissues on which OCU disagrees with Mackie and many otherself—proclaimed ‘moral skeptics’ are comparatively minor, andphilosophically controversial, not dead and decided; in fact,many such ‘moral skeptics’ are merely non—entifying strong moralobjectivists. Therefore Mackie’s and other purported refutationsof strong moral objectivism fail.These three differences between OCU’s and Mackie’s accountsof objective reasons for action also provide a framework forclassifying such moral philosophies. For instance, Brentano, astaunch moral objectivist, agrees with Moore, regarding the firstdifference, that objective reasons for action may cross temporal79gaps:one must consider not only oneself, but alsoone’s family, the city, the state, everyliving thing upon the earth, and one mustconsider not only the immediate present butalso the distant future. . . . To furtherthe good throughout this great whole as faras possible — this is clearly the correct endin life, and all our actions should becentred around it. (Brentano 1969, p. 32)Brentano also sides with Moore, regarding the second difference,in that he does not think pleasure the only good thing: “some• . . hold, in opposition to what experience makes evident to us,that pleasure is the only thing good in itself” (1969, p. 30).However, Brentano agrees with Mackie and Moore in thinking thatpleasure, broadly construed, is good: “we prefer joy to sadness.Were there beings who preferred things the other wayround, we would take their attitudes to be perverse, and rightlyso” (1969, p. 22). Brentano, like Mackie, parts with Moore andOCU regarding the third difference, the issue whether goodness isa property. Brentano agrees with Moore that all goods are alikein their goodness, that the range of determinate goods sharetheir resemblance equally:There is a common concept for those thingsthat are • . . called “good”. . . . Theconcept of what is good in itself is thus80univocal in the strict sense and not, asAristotle taught, univocal only in ananalogous sense. (1969, P. 75—76)However, unlike Moore and OCU, Brentano shies away from positinga determinable property ontically expressing this resemblance:“the good is that which is worthy of love, that which can beloved with a love that is correct” (1969, p. 18). In thusidentifying the good with good things, Brentano declines thechance, at the natural place in his argument, to assert, henceimplicitly denies, that good things have goodness distinct fromtheir good—making properties. Brentano, like Mackie, istherefore a non—entifying strong moral objectivist. As theidentification of Mackie as a strong moral objectivist shows,these three issues are useful in classifying moral theories, andcan reveal surprising alliances.Such alliances need revealing, for contemporary moralphilosophy is clouded by widespread ambivalence on the issueswhether entifying moral objectivism is true and whether strongmoral objectivism is true. Physics was similarly clouded when itwas found necessary to talk about light as if it were both a wavephenomenon and a particulate phenomenon.The revelation of these alliances amounts to a remodellingof the contemporary orthodox moral landscape. For instance,while Brandt has often been regarded as a moderate moralobjectivist somewhere between the extreme moral objectivism ofMoore and the extreme moral skepticism of Mackie, Mackie is81revealed to occupy a position somewhere between Moore and Brandt.Moore is an entifying strong moral objectivist, Mackie is anon—entifying strong moral objectivist, and Brandt is a weakmoral objectivist. Brandt is a weak moral objectivist (at leastin places) because though he gives objective procedures forprocessing one’s desires before pronouncing the resulting set‘rational desires’ and acting on them, he specifies nothing aboutthe nature of these ‘rational desires’ or the acts orconsequences they bring about (Brandt 1979).While some philosophers who have been aware of concepts ofobjective ethical properties seem not to have been cognizant oftheir awareness, others who have been cognizant of theirawareness seem either not to have fully grasped the nature of theconcept, or not to have correctly identified the set ofobjectively intrinsically valuable things. The disagreementsthus produced among such philosophers may suggest the followingobjection:there has been considerable disagreementamong philosophers who think that they haveperceived objective moral qualities both asto what sorts of things possess them and asto what objective moral qualities are. Forexample, some of them have thought that wehave a moral sense, while others would denythis. Some hold pluralistic views as to thegood, and there is considerable disagreement82among them as to what sorts of things areintrinsically good and to how they could beranked, while others hold monistic views.And there is the difference betweendeontologists like Ross and utilitarians likeMoore. This suggests that most of thesephilosophers thought that they wereperceiving moral qualities when they weren’t.Thus the group of philosophers who may haveperceived objective moral qualities wouldseem to be very small indeed. (R.I. Sikora1992, personal communication)However, these disagreements seem no different in character fromthose among researchers in perfectly respectable areas ofinquiry, where the researchers in question, despite theirdisagreements of the above sorts, nevertheless haveuncontroversially entertained the same concept of the entity inquestion. Consideration of some details of these disagreementsin their ethical forms will be found in various sections of thisthesis (locations indicated); however, to reply to the aboveobjection it suffices merely to provide harmless examples of thedisagreements in non—ethical forms. The harmlessness of theseexamples shows that these sorts of disagreement are capable ofharmlessness, hence that an ethical concept is not provendeficient merely because such disagreements can arise about it.First, for instance, just as futurologists disagree on whether we83entertain certain beliefs about the future by a sense ofprecognition or by reasoning, consciously or subconsciously, fromknowledge of the present, even though futurologists of bothpersuasions are entertaining the same beliefs about the future,entifying moral objectivists disagree on whether we entertainconcepts of objective moral entities by a moral sense (see §3.3)or by some other method, even though entifying moral objectivistsof both persuasions are entertaining the same concepts ofobjective moral entities. Second, just as hedonists disagree asto whether there is one sort of pleasant thing or a plurality ofsorts of pleasant thing, even though hedonists of bothpersuasions are entertaining the same concept of pleasure,entifying moral objectivists disagree as to whether there is onesort of good thing or a plurality of sorts of good things (see§6.1), even though moral objectivists of both persuasions areentertaining the same concept of goodness. Third, just asproponents of the synthetic a priori disagree as to what sorts ofproposition have the property of synthetic a prioricityunderlying the concept of synthetic a prioricity they entertain,even though these proponents of the synthetic a priori areentertaining the same concept of synthetic a prioricity,pluralistic entifying moral objectivists disagree as to whatsorts of things have the property of intrinsic goodnessunderlying the concept of goodness they entertain, even thoughthese pluralistic entifying moral objectivists are entertainingthe same concept of intrinsic goodness. Fourth, just as84pluralistic hedonists may disagree as to how pleasant thingscould be ranked in intensity of pleasantness, even though thesepluralistic hedonists are entertaining the same concept ofpleasure, pluralistic entifying moral objectivists may disagreeas to how intrinsically good things could be ranked in intensityof goodness (see §6.1), even though these pluralistic entifyingmoral objectivists are entertaining the same concept of intrinsicgoodness. Fifth, just as philosophers of mind disagree as towhether intelligence is a functional entity or an entity ofqualia, even though these philosophers of mind, though perhapsnot fully grasping the concept of intelligence, are grapplingwith the same concept of intelligence, entifying moralobjectivists disagree as to whether intrinsic value is intrinsicgoodness (value in things that are not acts) or intrinsicrightness (value in acts), even though these entifying moralobjectivists, though perhaps not fully grasping the concept ofintrinsic value, are grappling with the same concept of intrinsicvalue. This fifth sort of disagreement is perhaps reducible tothe third sort.Thus the view that these disagreements in moral contextssuggest that most of these philosophers thought that they wereperceiving moral qualities when they weren’t turns out to bemistaken. Thus the group of philosophers who may have perceivedobjective moral qualities would not, after all, seem to be verysmall indeed.853. The feel of goodness and the sense of right and wrongFinally, the skeptic of OCU’s moral objectivism might askhow it feels, in a loose sense, to entertain the concept ofgoodness. One cannot answer such a question directly, for thesame reason that one cannot directly explain to a blind personhow red and yellow look, or explain to one who has not tastedthem how chestnuts taste, or explain to one who has never beendrunk how alcoholic intoxication feels. Goodness has acharacteristic experiental ‘flavor’ which is unique and can onlybe hinted at to non—experiencers of the concept. To paraphraseNagel (1974), there is a of being an entertainer of goodnessthat is not identical to any other way of being, and is in someways completely inaccessible to non—entertainers of goodness.However, let me do what I can to explain by analogy what it feelslike to entertain the concept of goodness — or at least whatentertaining the concept I take to be goodness feels like.As is often the case when trying to conjure up, as vividlyand tangibly as possible, a concept of a determinable, one mayusefully start by reviewing the concepts of some of the relevantdeterminate properties. One then fixes on that aspect in whichthese concepts resemble each other, on the way in which theseconcepts have that aspect wholly and completely and equally,unlike family resemblances, which admit of degrees ofresemblance. When I follow this procedure for goodness, itstrikes me that certain determinate states of mind are all86wonderfully fulfilling and worthwhile in themselves, positivelysublime whether cheery or tragic. This common aspect dazzleswith peace and contentment, yet also engages with change andenergy, no matter how little or great the intensity of goodnessinvolved (likewise, hotness is present equally in the sensationsof both warm and boiling water). I realize that to those wholack such a concept of goodness., these analogies may be of nomore use than a winemaker’s claim that their wine tastesconfident yet unassuming, or an advertiser’s assurance that theirproduct smells fresh as springtime, or a human’s claim that redlooks like a trumpet sounds; however, it is the best I can do.Poets, artists, and musicians are often better able thanphilosophers to provoke elusive mental states in audiences.This feel of the concept of goodness helps account for thefact that goodness and humans, like pleasure and humans, are suchthat necessarily humans who grasp the concept of goodness (orpleasure) have some positive disposition toward goodness (orpleasure), though not necessarily to the exclusion of asimultaneous negative disposition toward goodness (or pleasure)(in agreement with Nagel 1970, p. 109—il). The feel of goodnessalso helps account for one’s ability to commend by attributinggoodness despite the coincident descriptive function of suchattributions: this conceptual content is naturally commendatory(Baylis 1967, p. 447, agrees). I use the term ‘naturallycommendatory’ in much the same sense that Brandt, who uses theterm ‘rational’ to denote those acts one would do if one’s87decision process were fully and errorlessly molded by availableinformation, believes that:the term in that sense will naturally becommendatory . . . (1979, p. 1),there is no incompatibility between a statementbeing true or false and with a descriptivemeaning, . . . and its having performative (e.g.recommending) force just on account of thatmeaning. (1979, p. 15)Again, a theoretical structure often treated as prohibitivelymysterious when asserted of objective moral goodness (e.g., Hare1981, p. 71—73) resurfaces asserted of objective prudence as ifperfectly straightforward. In truth, such structures are neitherstraightforward nor prohibitively mysterious.Those curious what it feels like to conceive of goodness andderivative moral entities such as rightness might also wonderwhat the faculty is by which we instantiate these moral conceptsin occurrent thoughts, e.g., in moral judgments, and what itfeels like to have such a faculty. This faculty cannot yet bespecified precisely; OCU assumes only that ethical intuitions useas their intrinsically ethical parts mental items in a generalconceptual mode of a familiar non—imaging sort, the abstractmode, rather than as in a sui generis conceptual mode unique toethics. However, humans as a matter of empirical psychologicalfact typically do have a sense (taken broadly) of moral right and88wrong, subsuming what is often called a sense of justice. Thispsychological fact is routinely assumed by normative ethicists,e.g., Rawls (1971, p. 46). Possibly it is an illusory sense,like a sense of witch—detection which humans in some societiesclaim to have, but which is aimed at no natural (i.e., objective)kind. Nonetheless, humans typically have a moral sense, in thatthey claim to be able to tell, as a ‘basic action’ (sensu Danto1963 or Goldman 1970), whether many states of affairs are rightor wrong, and whether some states of affairs are more or lessright or wrong than others. This sense of right and wrong is notomnieffective, for there are many states of affairs that appearto be matters of ethics, but on which one’s moral sense yields noclear verdict as to whether the state of affairs is right orwrong, merely a set of conflicting intuitions. Likewise, one maybe sure that an object held in one’s hand has a weight, but beunsure whether that weight is above or below a suggested amount.Even if the sense of right and wrong is aimed at ethical naturalkinds, it is logically possible that it is a completelyunreliable sense, like the sense of precognition some humansclaim to have, which is aimed at real (future) events but whoseaccuracy is not substantiated by any significant empiricalevidence. Possibly more than one kind of sense of right andwrong exists; possibly some people make moral judgments byconsulting sensed ethical properties, and others make moraljudgments by other criteria, in which case their sense of rightand wrong may reduce to applications of some other sense or89senses. Though the claim must be qualified in these ways, humansuncontroversially have a sense of right and wrong. The cause oftheir having such a sense is evolutionary (Simpson 1969, p. 134).It is sometimes objected to the moral sense theory that weshould but don’t know of any organ by which moral sensing isdone. According to OCU the primary organ by which humansexercise this moral sense is the brain. Cognitive theory isprogressing rapidly, and it may soon be possible to specify justwhich parts of the brain perform just which functions in theprocess of making moral judgments. However, our current lack ofthe full details of these functions does not entail that no moralsense exists or that we don’t know which organ is the organ ofmoral sense. We don’t yet know the full details of how theorgans of any senses work. For instance, we have a sense ofsight — but all we know of the details are that the eyes feedcertain neural impulses to the brain, and that certain neuronstructures and brain regions seem involved in the processing ofthese impulses and consequent production of visual qualia. Weinfer these details from facts about, e.g., eye structure, brainblood flow and brain—wave production in the various brain partsduring episodes of seeing, arid from details from artificialintelligence theory about how neurons might act as inferentialmatrices. We know even less about our sense of, e.g., addition,whereby we mathematically intuit the sums of numbers. Butdetails similar to those available about our sense of sight arealso available, via the same current technology, about additive90sensing and moral sensing. Supposing these tests were done, andwe knew how the neural impulses fed from the ears to the brain,in one asked a moral question, and how certain patterns of bloodflow and brain—wave production were subsequently provoked — wouldthis count as knowing the identity of the moral sense organ? Ifso, then the moral sense theory is on firm anatomic ground; ifnot, then the senses of addition and sight are also open to theno—organ objection, which hence is not strong enough to be aserious worry to the proponent of a sense for which other goodevidence is available.Those humans whose sense of right and wrong does not allowthem to directly consult objective ethical properties may wonderwhat it feels like to have such a capacity. It feels nodifferent from any other capacity to perform a basic action. Onewills the moral evaluation to occur, and it does. Unless one isa specialist, one has no more idea why the act of will isfollowed by the evaluation than one has why another kind of actof will is followed by one’s leg moving. It just happens, as oneexperiences in so many aspects of life. Virtually all of theexperiental novelty is in the moral concepts, e.g., of goodness,themselves.This sparse theory of sensing right and wrong troubles some;Nowell—Smith objects that:the intuitionist’s reply to the question ‘Howdo I know what I ought to do?’ is . .unenlightening. For it turns out to be: “You9].know what you ought to do by intuiting thenon—natural characteristic of obligatorinessthat inheres in certain actions. But, mydear sir, do not be alarmed by thismysterious phrase. It is only another way ofsaying that you know that you ought to dothose actions.” We know what we ought to doby knowing what we ought to do. Opium sendsus to sleep because it has a virtusdormitiva. (1954, p. 43)However, since, unlike the traditional intuitionist, OCU holdsthat moral intuitions may have much in common with other sorts ofintuition, OCTJ can and does hold that due to such a resemblance,this objection to basic acts of moral judgment applies equally tonumerous common mental basic acts. How does one know whether aword on the page spells ‘bib’ or ‘did’? One just looks, and thejudgment occurs correctly and with so little delay that one mightmistake it for the looking itself were it not for the reports ofthose with cognitive deficiencies preventing them from makingsuch basic acts of visual judgment. They look, they see theletters on the page, but nothing else happens, like a paralyticunsuccessfully willing their inert limb to move. As with anybasic act, mutatis mutandis, one intuits goodness by intuitinggoodness; if one has the appropriate faculty, doing it is thiseasy; if not, little in the way of coaching or explanation islikely to help much. Since Nowell—Smith’s objection fails92against common mental basic acts such as some visual judgments,because such acts exist despite the objection’s applying to them,the objection fails generally, including against the mental basicact of intuiting goodness. What Nowell—Smith interprets asexplanatory vacuity is actually mere ontological parsimony: don’tposit entities where appearances aren’t complicated enough towarrant it.It might be objected to the existence of a sense of rightand wrong directly consulting objective ethical properties that agreat many people report having no such sense, whereas with othersenses such as the senses of sight or hearing, most people havesuch a sense.I reply, first, that if, as argued in §3.2, prudentialreasons for action involve objective moral entities, then sincevery many people think that they can see that prudential reasonsfor action exist, very many people are, whether they arecognizant of it or not, aware of objective moral entities, hencehave a sense giving them such awareness. A second considerationis that even if a large proportion of people lacked such a sense,this would not necessarily be unusual. Some senses are morespecialized than others, and may be lacked frequently. Forinstance, some specialized aural capacities are often lacked byhumans; e.g., the capacity to sense and contemplate certainsubtle, complex musical rhythms. Such a lack may afflict a greatmany people; so might it be with the sense of objective right andwrong. An individual human typically has very many fewer atomic93sensory capacities than the full set ever recorded in one or morehumans. For causal reasons, the sets humanly instantiated tendto form family resemblance clusters in sense—space; however, manyother sets are causally possible. The potentially spottydistribution of human capacities is illustrated by idiot savants,who are infantile or vegetative in most respects, but highlyskilled in isolated capacities, such as the ability to do certainmathematical operations, or play complicated pieces on a musicalinstrument.It might be also objected to the existence of a sense ofright and wrong directly consulting objective ethical propertiesthat almost everybody having a sense such as sight or hearing iscognizant of it, whereas, if the first part of my reply to theabove objection is correct, many people having an ethical senseare not cognizant of it.I reply that while some senses, such as sight and hearing,are broad enough and important enough to human living that theirabsence is immediately noticable, other sensory capacities arenarrow enough and unimportant enough that an unreflective human,as most are, may easily not notice their lack. For instance, ifthe capacity to sense and contemplate certain subtle, complexmusical rhythms is lacked by a human, all that may be noticed isthat the human doesn’t like classical music or jazz (see Moore1903, p. 191—92). A broad sense such as sight or hearing is,strictly speaking, not one sense but the sum of a great manynarrow, specialized senses. If all of such a set are lacked, the94consequences are striking, but if only one or a few are lacking,no one may notice, not even the human having the lack, whether ornot the lack is widespread. A human lacking the ability to sensea certain shade of green may not find out unless professionallytested, for since they can sense most shades of green, the lackleads to few or no mistakes in everyday life. Most people lackthe ability to sense certain shades of extreme violet, but theydo not notice the lack, for it’s news to them that such shadesexist. Depending on how narrow the sense of right and wrong is,and how closely related to other human senses it is, the lack ofthis sense may not be very conspicuous, even to the human withthe lack.Conversely, people having a narrow sense that others lackmay not notice the sense in themselves or its lack in others. Ifpeople are told by enough others, and pleased enough by thesuggestion, that the decisions they made and actions they took onthe basis of information gained by the sense in question wereactually made and taken on some other basis, they often believeit and suppress awareness of the real basis. For example, manypeople who are experienced and skilled at character analysis bymeans of subconscious processing of subtle verbal and behavioralclues are persuaded that they actually came to their beliefsabout the subject’s character via telepathy or tarot cards orsomesuch. As in the case of character analysis, so with ethics:a spectrum of cognizances results. Some people are keenly awareof the moral sense in themselves or others, while other people95are oblivious to the same. Regarding the objection under presentscrutiny (that almost everybody having a sense such as sight orhearing is cognizant of it, whereas arguably many people havingan ethical sense are not cognizant of it), it suffices that thereare some other specialized senses, even socially important ones,that many humans possess but are unaware of possessing.It might also be objected to the existence of a sense ofright and wrong directly consulting objective ethical propertiesthat people reporting having no such sense are able to livenormal, independent lives, whereas with other senses such as thesenses of sight and hearing, people having no such sense arethereby limited in their pursuit of normal, independent living.I reply that the discrepancy, if any, is much smaller thanclaimed. First, people lacking even broad senses such as sightand hearing often learn to compensate considerably by careful useof their other senses, and need help only in minor ways. Second,people lacking a sense of right and wrong live lives constrainedand impoverished in some important ways not generally recognizedas consequences of their moral sensory lack. Such humans maylack benevolence, which in many situations prevents them fromsharing the fruits of cooperation. Human populations tendingtoward a Hobbesian state of nature are noticably more violent andshortage—prone than societies with high levels of kindness andhelpfulness. Or such humans may lack prudence, consequentlyneglecting the means to their own welfare, both immediate anddistant. Because the losses incurred by moral sensory lack are96so often long—term consequences of present error, these lossesare often not recognized as such consequences, and present erroris not recognized as such. Blind humans run the risk of walkinginto walls; morally sense—impaired humans run the risk of beingmugged on the way home from work, or being impoverished in theirold age.4. The open question argumentThose who are able to entertain the concept of goodness canuse their ability to satisfy themselves that the property ofgoodness exists. The argument involved is a variant of what hasbecome known, adapting Moore, as the ‘open question argument’.Moore writes that:ethics aims at discovering what are thoseother properties belonging to all thingswhich are good. But far too manyphilosophers have thought that when theynamed those other properties they wereactually defining good; that theseproperties, in fact, were not simply‘other,’ but absolutely and entirely thesame with goodness (1903, p. 10),that:the attempt to define good is chiefly due97to want of clearness as to the possiblenature of definition. There are, in fact,only two serious alternatives to beconsidered, in order to establish theconclusion that ‘good’ does denote a simpleand indefinable notion. It might denote acomplex, as ‘horse’ does; or it might haveno meaning at all. . . . (1) Thehypothesis that disagreement about themeaning of good is disagreement with regardto the correct analysis of a given whole,may be most plainly seen to be incorrect byconsideration of the fact that, whateverdefinition be offered, it may always beasked, with significance, of the complex sodefined, whether it is itself good. .(1903, p. 15),that:(2) . . . the same consideration issufficient to dismiss the hypothesis that‘good’ has no meaning whatsoever. It isvery natural to make the mistake ofsupposing that what is universally true isof such a nature that its negation would beself—contradictory. . . . And thus it isvery easy to conclude that what seems to be98a universal ethical principle is in fact anidentical proposition; that if, forexample, whatever is called ‘good’ seems tobe pleasant, the proposition ‘Pleasure isthe good’ does not assert a connectionbetween two different notions, but involvesonly one, that of pleasure, which is easilyrecognized as a distinct entity. Butwhosoever will attentively consider withhimself what is actually before his mindwhen he asks the question ‘Is pleasure (orwhatever it may be) after all good?’ caneasily satisfy himself that he is notmerely wondering whether pleasure ispleasant. And if he will try thisexperiment with each suggested definitionin succession, he may become expert enoughto recognize that in every case he hasbefore his mind a unique object, withregard to the connection of which with anyother object, a distinct question may beasked. (1903, p. 16),and that:if we recognize that, so far as themeaning of good goes, anything whatevermay be good, we start with a much more99open mind. . . . when we think we have adefinition, . . . we shall start with theconviction that good must mean so and so,and shall therefore be inclined either tomisunderstand our opponent’s arguments orto cut them short with the reply, ‘This isnot an open question: the very meaning ofthe word decides it; no one can thinkotherwise except through confusion.’(1903, p. 20—21)I view Moore’s expression of the open question argument assusceptible of a plausible interpretation, as follows. Theexperiental feel and flavor of the concept of goodness yield twopremises to anyone properly able to entertain the concept ofgoodness (and to otherwise think intelligently): first, thepremise that a proposition of the form ‘x(Gx—*Nx)’, where Gattributes goodness and N attributes some non—ethical property or(non—trivially) self—consistent complex of properties, cannot beanalytic; and second, the premise that such a propositionconcerns relations between two positive, contentful ways ofbeing. The first premise follows from the novelty and uniquenessof the experiential feel and flavor of the concept of goodness.That such a concept could turn out to be constructed entirely ofbits and pieces of other concepts with other experiental feelsand flavors seems as likely as that the concept of redness shouldturn out to be composed entirely of concepts of light vibrations100or popsicle sticks. From the first premise it follows thatgoodness is not identical with any non—ethical property orcomplex of properties, hence that no such definition of goodnesscan succeed. Thus goodness cannot be reduced to some lesscontroversial entity from elsewhere than the realm of objectiveintrinsic value, as has often been attempted. The second premisefollows from the stubborn tangibility of the concept of goodness;if it’s not there, why won’t it go away, and where is itsexperiental flavor and feel coming from? From the second premiseit follows that goodness is not a conceptual void, a theoreticalplaceholder, but a substantial thing in the world, a property.Since goodness is not nothing and is not something else, it mustbe itself — an independent property adding a new and importantway of being to our ontology. This argument is unfortunatelyunavailable to those unable to entertain the concept of goodness,but is compelling to those who have this ability, hence haveepistemic access to the two key premises.It does not follow from these results that it is notanalytic that goodness stands in the determinable—determinaterelation with determinate good things. As shown in ch. 2, it ispossible that determinates carry within themselves, as a part,the determinable in question. If so, then a proposition of theform ‘x(Ax-Gx)’, where A attributes the determinate good inquestion, could not also be of the form ‘x(Nx-Gx)’, for A hasethical content, its good part, whereas N does not.Brandt criticizes Moore’s version of the open question101argument on the ground that Moore has not attended sufficientlyto the distinction between overt and covert synonymy (as Brandtterms them). Brandt writes that:two property—referring expressions mean thesame if and only if they satisfy thefollowing two conditions: (1) For everyactual thing or situation, the speaker must,if called upon to judge, be willing eitherto apply both expressions, to reject both,or must be in doubt about both. . . . (2)The first test is not quite adequate, for itwould permit us to say that “unicorn” and“centaur” mean the same. In order to avoidthis difficulty, we must consider the speechresponses of a person to things orsituations that are not actual (1959, p.160),that:Some writers appear to feel that matters aremuch simpler, that all we have to do is askourselves about our intentions — what weintended someone else to learn from what wesaid. But how are we supposed to find whatour intention was, except by noticing whatwe would have been content to say instead ofwhat we did say — except, perhaps, in102specially simple cases? (1959, p. 162),that:It is useful to mark a difference betweenthe case of two expressions satisfying ourcriteria and also obviously meaning thesame, and the case of expressionssatisfying our criteria but obviouslymeaning the same. In the former case weshall say they . . . are overtly synonymous;in the latter case we may say they onlyare covertly synonymous (1959, p.163),that:Moore’s most important suggestion was the“open question” test or criterion forsameness of meaning. . . . Suppose you havetwo terms, P and Q, and it has beensuggested that they mean the same. Now,Moore said, a way to test whether theyreally do mean the same is this. Compose aquestion of this form: “Is everything Palso Q?” (or the reverse). Then askyourself whether this question is“intelligible” or “significant” or whetheryou understand what it would mean to doubtthat the answer is affirmative. If the103question is intelligible and significant,Moore said, then the terms do mean thesame, for if they did, the question wouldbe no more intelligible than would be thequestion: “Is everything P also P?”. .Moore thought no naturalistic definitions(of ethical terms) pass this test.Certainly “desirable” and “is desired bysomebody” do not, for “Is everything that isdesired by somebody also desirable?” seemsto be an intelligible and significantquestion (1959, p. 164),and that:What can Moore mean . . . when he says that,in doubtful cases, we are to ask ourselveswhether his question is intelligible, an“open” question? Perhaps what he means issimply this: that P and Q do not mean thesame for a person if, when he asks himself“Is everything P also Q?” he is doubtful ofthe answer, or the answer does not seem tohim obviously affirmative. . . . But ifthis is what Moore meant, then it appears hewas mistaken at least for covert synonymy,for it might well be the case that two termsdo covertly mean the same, as a given person104uses them, so that he could never correctlydeny that everything that is P is also Q,but, if the person were not clearly aware ofhis use of terms, he might remain doubtful,and the answer might seem to him notobviously affirmative. . . . Furthermore,two terms might pass Moore’s test even ifthey didn’t mean the same. (1959, p. 165)Brandt succeeds in showing that so far as the open question testis construed as a quick test of sameness of meaning forproperty—referring terms, it is deficient in the respects hementions. It is unclear whether Moore’s argument is seriouslydamaged by this demonstration, for it might be held that sinceMoore held that one must, to see that goodness is an indefinableproperty, by repetition become “expert” (1903, p. 16) in theapplication of the open question test to goodness and its variouspurported definitions, it is reasonable to explicate Moore’sdemand as including that one must apply Brandt’s clarified testwhen arguing for synonymy, hence would not make mistakes byclaiming synonymy quickly or unwarily. However, Moore is guiltyat least of omitting responses to some reasonable if ultimatelyundamaging objections.The merit of Moore’s position is, however, not my presentconcern; my present concern is whether Brandt’s objections applyto the construal of the open question argument I claimed above toshow that goodness is an independent property adding a new and105important way of being to our ontology. That argument has twoparts.The first part uses the premise that a proposition of theform ‘x(Gx<—Nx)’, where G attributes goodness and N attributessome non—ethical property or self—consistent complex ofnon—ethical properties, cannot be analytic. This premise wasclaimed to be epistemically accessible to any intelligent personable to entertain the concept of goodness, and to yield theconclusion that goodness cannot be reduced by definition to someless controversial non—ethical property, simple or complex.Brandt’s correct observation that detecting synonymy is seldom ornever a simple process, available to the quick or unwary, doesnot damage this first part of my argument, for I argue not forsynonymy, but for lack of synonymy. As Brandt remarks,the proper procedure for showing that adefinition is mistaken is to pose a“counterexample’ — an example, actual orpossible, to which all or most people wouldapply the one term but not the other. Clearcounterexamples definitely prove that adefinition is mistaken. . . . it is mucheasier to disprove a definition than it isto establish that one is correct, for adisproof (except where there are severalsenses) requires only a singlecounterexample. (1959, p. 162)106The lack of synonymy I argue for is between the term ‘good’ andsome non—ethical term such as ‘highly evolved’. I imagine astate of affairs including a highly evolved object such as abirch tree, but excluding any manifestation of the feel andflavor forced upon my consciousness when I contemplate the aspectin which various determinate goods fully and equally resembleeach other. To this state of affairs I would be willing to applythe term ‘highly evolved’, but unwilling to apply the term‘good’. One does not become certain of such lack of synonymyovernight, but the longer I think about it, the surer I become.After consideration and rejection of a number of possiblesynonymies between ‘good’ and various non—ethical concoctions,one becomes aware of a sameness to the gaps of meaning, and awareof a growing conviction that the whole project is misconceived.Like one who considers and in turn rejects various physicalistdefinitions of a mental term such as ‘red’, one comes to see thattwo quite different sorts of concepts are being compared, andthat consideration of further definitions of the sort in questionis a waste of time.The second part of my argument uses the premise that aproposition of the form ‘x(GxE—>Nx)’ concerns relations betweentwo positive, contentful ways of being. This premise was claimedto be epistemically accessible to any intelligent person able toentertain the concept of goodness, and to yield the conclusionthat goodness is not a conceptual void, a theoreticalplaceholder, but a substantial thing in the world, a property.107Brandt’s correct observation that detecting synonymy is seldom ornever a simple process, available to the quick or unwary, doesnot damage this second part of my argument, for I argue not forsynonymy, i.e., that the two terms have the same meaning, butrather that each term has , meaning. Brandt does notspecifically address the question how one is to tell whether aterm has a meaning, but from his recommendation of the strategyof counterexamples for proving lack of synonymy, it may beextrapolated that a term has a meaning if a set of states ofaffairs can be specified to which all or most people wouldexclusively apply the term. While ‘good’ perhaps does notsatisfy this criterion, it does satisfy a criterion similar butrestricted to those who entertain the concept of goodness. Onemight even argue that ‘good’ also satisfies Brandt’s unamendedextrapolated criterion, on the ground that ‘good’ is used inseveral senses, and that the relevant sense is used appropriatelyby most or all of those who use it at all.To this line it might be objected that it is not the casethat most or all of those who entertain the concept of goodnessapply the term ‘good’ exclusively to a certain set of states ofaffairs. The marked disagreement among certain Moorean moralobjectivists about which things are good might be held to showthat such people apply the term to rather different sets ofstates of affairs. However, this objection mistakes disagreementabout which determinate things have a given determinable fordisagreement about which things are the determinable in question.108People can disagree about whether transparency is a color withoutsignificantly disagreeing about what coloredness is; and they candisagree about whether knowledge is good without significantlydisagreeing about what goodness is. When Moorean moralobjectivists entertain the concept of goodness, they recognizethat they have done so — whether or not they have entertainedthis concept in an appropriate context. The enduring popularityof hedonistic utilitarianism among Moorean moral objectivistsevidences that they frequently, perhaps even usually, entertainthe concept of goodness in appropriate contexts, hence are aswarranted in their positing instances of a property of goodnessas others are in positing instances of a property of colorednesswhen they see reds and greens. Their concepts are entertained atthe right time, for the underlying property is also present.Thus whether or not Moore’s version of the open questionargument is sound, a version supporting OCU is sound.Thus in this chapter a satisfactory positive account of theconcept of goodness has been given; hence the second objection toOCU is refuted.109Chapter FourCOMPONENTS OF UTILITARIANISM1. Compatibilist utilitarianismThe third objection to OCU (and utilitarianism in general)is that the most plausible and popular utilitarian doctrines areforms of either act or rule utilitarianism, but that each ofthese doctrines suffers a repellent objection. AU is thedoctrine that an act is right if it will result in at least asmuch utility as any other available act; RU is the doctrine thatan act is right if endorsed by a rule from a set whose adoptionby society will result in at least as much utility as theadoption of any other set of rules.Some ethicists (e.g., Brandt 1963, p. 109—110; Harrod 1936,p. 147) are repelled by AU for its apparent implication that theprima facie intuitively justified rules of coinmonsense morality(e.g., that promises should be kept, that truth should be told,and that innocent persons should not be punished) deserve nospecial favor and on any given occasion should only be followedif doing so is likely to produce more utility than not doing so.Cynicism and social disorder seem more likely than harmonioushappiness to result from taking these rules so lightly. Call110this the counterintuitivity objection. The intuitions violatedcan be construed in either of two ways; they can be construed asscientific intuitions that subscription to the rules in questionmust be beneficial, contra the claims of AU, or as moralintuitions that, beneficial or not, such rules must be obeyed.shall discuss both construals of the relevant intuitions.On the other hand, some ethicists (e.g., J.J.C. Smart 1973,p. 10) are repelled by RU for its apparent implication that rulesshould be followed even on occasions when doing so would not gainany utility. A moral code that makes demands unsupported by theonly intrinsic value it (overtly) posits is unconvincing. Thisis the rule worship objection.Thus both the chief forms of utilitarianism are seriouslyflawed; hence utilitarianism including OCU should be abandonedaltogether.In this chapter I defend a form of utilitarianism, the formassumed by OCU, that combines the main philosophical virtues ofAU and RU while retaining the vices of neither.Many ethicists are attracted to hedonic utilitarianism forits metaphysical parsimony and its positing pleasure, under someconstrual, as intrinsically good, and pain, complementarilyconstrued, as intrinsically bad. For purposes of this chapter,intrinsic values may be interpreted either cognitively, e.g., asin OCU’s moral objectivism, or non—cognitively, e.g., J.J.C.Smart 1973, p. 4. Both AU and RU, the two most well—expoundedgeneral forms of utilitarianism, are susceptible of hedonic111interpretations, non—hedonic interpretations, and ofinterpretations classing both hedonic and non—hedonic entities asutility. However, the interpretations of the third, hybrid sortrightly attract less contemporary support than purely hedonicinterpretations, and interpretations of the second, purelynon—hedonic type rightly attract almost no contemporary support,and hedonism in general is popular and plausible (see Goldstein1989), so I shall confine the scope of my remarks to purelyhedonic interpretations despite these remarks’ mostly alsoapplying to any partly or wholly non—hedonic form ofutilitarianism.To see that there is a form of utilitarianism more plausiblethan either AU or RU as standardly interpreted, we must nowqualify the condensed definitions given above of AU and RU bydistinguishing ontic from epistemic criteria. The onticcriterion for X—hood says what an X . An epistemic criterionfor X—hood says how an X may be known from other objects. Forexample, the ontic criterion for being a platypus is having eachof very many disjunctive anatomic, behavioral, mental, and otherproperties (enough to completely describe platypuses). Oneepistemic criterion for being a platypus is being an egg—laying,web—footed manunal; another is being of the same species as thisanimal here (pointing at a platypus). For a given X there isonly one ontic criterion, but there may be many epistemiccriteria. Different epistemic criteria will most efficientlyserve different observers. An ontic criterion may also be an112epistemic criterion for its object, but this need not be so for agiven observer. For example, a blind human can tell a red tomatofrom a green tomato, but not by color.The distinction between ontic and epistemic criteria hasbeen called, respecting intrinsic goods, the distinction between‘conferring properties’ and ‘identifying properties’ (see Baylis1967, p. 446). I use the wider version of the distinctionbecause it strengthens OCtE metaethically to use premises ofgeneral epistemology and metaphysics rather than premisespeculiar to ethics, for this makes it more obvious that nohidden, question—begging values are being smuggled into OCTJ’spremise set. Ethical value is goodness and badness, and nothingelse.Hare makes a similar distinction between ‘critical moralprinciples’ and ‘prima facie moral principles’, these principlesbeing criteria of right acts expressed in prescriptive form(1981, p. 38—41). Hare defines prima facie moral principles asrelatively simple, general moral principles expressing primafacie duties and tailored to our human reactive attitudes so asto be useful at the intuitive level at which most practical moraldecision—making is conducted. Hare defines critical moralprinciples as moral principles of unlimited specificity chosenunder the constraints imposed by the logical properties of themoral concepts and by the non—moral facts, and by nothing else.While Hare’s distinction works well within his system, andcaptures some interesting differences between the two classes of113moral principles we want distinguished, it is imperfect becauseit mixes moral and non—moral features in a confusing way. Itmixes Hare’s theory of how to choose moral ends with thescientific theory of how to promote ends once chosen. OCU is aform of strong moral objectivisin, whereas Hare is (overtly) atbest a weak moral objectivist, and perhaps not a moralobjectivist at all; OCU and Hare thus advocate choosing moralends in quite different ways. But given ends, OCU and Hare agreealmost completely on the existence of two sorts of criteria forrecognizing these ends, these criteria having separate places anduses in practical thought.This distinction between ontic and epistemic criteria iscrucial to utilitarianism. Any definition of utilitarianismincluding a criterion for a right act is ambiguous between onedefinition interpreting the criterion as ontic and anotherinterpreting it as epistemic. AU and RU as defined above areambiguous in this way. Thus we may distinguish ontic AU fromepistemic AU, and ontic RU from epistemic RU. Epistemic versionsmust be indexed to particular kinds of moral agent to have truthvalues; e.g., ‘epistemic AU, indexed to humans, is true’ = ‘forhumans, epistemic AU is a maximally efficient epistemic criterionof right action’. Ontic AU and epistemic RU are true and theother two theories false; this view is compatibilistutilitarianism (CU). Ontic AU has been expressed in theutilitarian intuition that whether one is considering acts, setsof rules, or whatever, the bottom line in ethics is quantity of114benefit. Epistemic RU has been expressed in the utilitarianintuition that any moral code worth using as an everyday,practical guide must be both learnable and detailed, henceconsisting of neither a huge set of rules nor only a single rule.Ontic AU and epistemic RU are not competitors but the twomutually consistent halves of a full utilitarianism. To bepractical a morality must have efficient epistemic criteria; tobe maximally efficient an epistemic criterion must pick outobjects if f they satisfy the morality’s ontic criterion; hence amorality must also have an ontic criterion (see Hare 1981, p.46). Thus a full utilitarianism must have both epistemiccriteria and an ontic criterion; epistemic RU and ontic AU supplythese criteria. It is ontic RU and epistemic AU which are falseand have done much mischief in ethics.Ontic RU is liable to the rule worship objection. Iffollowing RU rules was itself the goal of ethics, then it wouldnot need to be justified by its effect on utility or anythingelse, and could not be so justified because any justificationwould be via another ontic criterion, and there can be only oneontic criterion per object. Yet moral respect for rule—followingfor its own sake is mere worship. Some hold pleasure orhappiness to be intrinsically valuable, hence the most preciousobjects in the universe; some hold knowledge, or love, or beauty,or honor to share this first rank; very few hold rule—followingto be have anything but extrinsic value. Even Kant onlyconsidered rule—following valuable when it allowed the good will115to choose out of duty; the following of nonmoral rules heconsidered intrinsically worthless.It might be objected to this criticism of ontic RU thatsince the appeal to utility is built right into the descriptionof right acts, appeal is not being made to anything outside ofright acts, such as their effects, hence that no second onticcriterion is implicitly appealed to to justify the first.However, if attribution of intrinsic value is thus built rightinto the description of right acts, then it must apply torule—following as well as utility, for they are equally objectsdescribed as independent components of right acts; hencerule—following is still being valued for its own sake.Epistemic AU is, when indexed to humans, liable to thecounterintuitivity objection, for it leads us to lie, steal,break promises, punish innocent persons, etc. when our primafacie intuitions correctly tell us that we should not. Theintuitions violated can, as mentioned earlier, be construed ineither of two ways; they can be construed as scientificintuitions that subscription to the rules in question must bebeneficial, contra the claims of AU, or as moral intuitions that,beneficial or not, such rules must be obeyed.The bulk of this chapter addresses the first construal; Iargue that AU taken correctly, i.e., ontically, entails that suchrules should be part of our moral code because subscription tothem would be beneficial. Thus AU does not claim that such rulesshould not be subscribed to.116Regarding the second construal of the intuitions inquestion, I argued (chiefly in ch. 3) that moral intuitionsshould be treated like any other sort of epistemic intuition.Epistemic intuitions are fallible, especially prima facieintuitions, but should not be rejected without reason; intuitionsare the only raw materials we have to build a picture of realityfrom, hence carry a presumption of truth which can only benullified by contrary epistemic data (see Rescher 1973, p.55—59). In the present case, other relevant data turn out tocorroborate our moral intuitions on the whole, including theintuitions that the rules of coinrnonsense morality in question arejustified. However, this corroboration comes at the priceof a crucial qualification to the intuitions under discussion.It is false that the rules must be obeyed whether subscription tothem would be beneficial or not. Rather, they must be obeyedbecause it turns out that subscription to them would bebeneficial. The intuitions must be reinterpreted as intuitionsthat subscription to the rules would be beneficial. Prima facieintuitions, moral and otherwise, often need reinterpretation tobe accepted, so this move is not ad hoc. However, some moralintuitionists will find this price too high; with them I mustsimply disagree, and hold that the unqualified intuitions aresufficiently lacking in coherence with our other epistemicintuitions that it is rational to reject them. Adherents ofnon—cognitivist forms of utilitarianism will have, if theysupport the second construal of the counterintuitivity objection,117to explain why moral intuitions count as data at all, but thatdoes not concern cognitivist forms such as OCU.Therefore epistemic AU is, when indexed to humans, liable tothe counterintuitivity objection. The one—rule everyday moralcode of epistemic AU might suit a society of near—omniscientbeings (e.g., Hare’s archangels — 1981, p. 44) who could performthe required calculations, but is ill—suited to beings of humanintelligence. To complete this refutation of thecounterintuitivity objection, let me explain the CU moral code inmore detail to show why subscription to the commonsense moralrules in question would be beneficial.The CU moral code consists of the single high—level rule‘maximize utility’, explaining why the other rules should befollowed, and a short list of lower—level, derivative rules foreveryday practical use, largely consisting of rules ofcommonsense morality, such as ‘keep your promises’, ‘tell thetruth’, and ‘don’t punish innocent persons’. The top rule‘maximize utility’ is not for everyday direct practical use, forthe potential for abuse and misunderstanding is too high.However, this rule is put to everyday indirect practical use, andto occasional direct practical use. The top rule ‘maximizeutility’ is properly used chiefly: first, to help interpretlower—level rules in novel situations falling under theirjurisdiction; second, to help settle new disputes betweenconflicting lower—level rules; third, to help design new rules togovern new situations as culture changes; and fourth, to help118decide when to abandon old rules made obsolete by culturalchange. Uses of the first sort are the least significant andmost common, for they involve no change in the code’s rules, yetare encountered proportionately to the rate of cultural change,which tends to be high in a complex, civilized society. Uses ofthe second sort are more significant, and fairly common sincenovel conflicts between lower—level moral rules also ariseregularly due to cultural evolution. Uses of the third andfourth sorts are the most significant and least common; citizensas democratic self—rulers should be encouraged to think regularlyabout the composition of their moral code, but since the list ofmoral rules for learnability is short, the rules will be based onextremely general features of human psychology, unlikely tochange much in one human’s lifespan, hence there will be littleturnover of rules required. Contemporary examples of issuesinvolving these four sorts of use of the top rule are,respectively: first, the question whether shortening a comatosehuman’s life harms the human; second, the question whether thepsychiatrist’s promise to respect their patient’s desire forconfidentiality outweighs the psychiatrist’s duty to tell thetruth to the police about serious criminal activity learned of;third, the question how to formulate new rules about wastedisposal now that so much waste is toxic to other species (andour own); and fourth, the question whether to drop thestill—popular rule against marijuana use as an irrational taboo(see Boyd 1991).119The mystery how a human’s sincerely trying to maximizeutility at every turn could generate less utility than trying todo something else is explained as follows, paraphrasing Sikora(unpublished). When epistemic RU humans believe that it is wrongto break a rule, they in some choices believe that following therule would maximize utility and in others not. When they believethat it would, they behave exactly as if they were epistemic AUhumans. Epistemic RU humans only behave differently fromepistemic AU humans when they follow a rule even though theybelieve that breaking the rule would have better results. Butepistemic RU humans maintain that their overall pattern ofbehavior in following the rules will have better results than ifthey followed the epistemic AU single—rule moral code. Thissuperiority can’t be explained in terms of the part of theirbehavior that matches that of epistemic AU humans. The onlypossible explanation is that in the choices when they behavedifferently from epistemic AU humans by following a rule eventhough they believe that breaking it would have better results,they are wrong in their belief sufficiently often or their errorsare sufficiently serious to make their policy of following therules yield better results.In practice epistemic RU humans will make numerous decisionsto follow the rules when they believe that it would be better to.break them, and sometimes they will be right that it would bebetter to break them. But if the overall importance of theirerroneous judgments doesn’t outweigh the overall importance of120their correct judgments they would have done better as epistemicAU humans.Sikora calls this explanation the ‘error theory’. Anepistemic RU short moral code is more beneficial than theepistemic AU single—rule moral code if f the error theory is true.But, as Brandt argues, varied life experience of moral eventsstrongly suggests that such a short moral code IS morebeneficial. A human society that subscribed to it would be farhappier than one that subscribed to the epistemic AU single—rulemoral code because the latter leaves too much room forrationalization and because behavior in such a society wouldn’tbe sufficiently predictable. Thus the empirical evidencesuggests that the error theory is not merely logically possiblebut actually true.The phenomenology of refraining from doing what one thinkswould maximize utility because one doubts that doing that actwould maximize utility is odd; it is like the phenomenology ofspearing fish — don’t throw the spear where you see that the fishis, because that’s where you doubt that the fish is (due to thebending of light as it passes through an air—water boundary atother than a right angle). The seeming paradox of thinking boththat p and that -‘p is resolved; one asserts and negates the sameproposition, but not in the same thought even though sometimessimultaneously. One expresses the proposition via two separatemental channels, one asserting it and the other negating it. Itmight be objected that one should not accept apparent121fish—seeings, etc. as beliefs until they have been studied, buthumans often do, treating them as fact for purposes of action,and not as a gamble but with confidence in the beliefs’veridicality. History shows that humans can even believecontradictories.The strategy of not doing x in order to do x crops up oftenin human affairs (Tocqueville 1840, p. 148; Nagel 1970, p. 132).Utilitarians using an epistemic RU short moral code shouldrefrain from breaking the rules even when it strikes them thatdoing so will gain utility for the same strategic reason thatinvestors, whose motive is to gain money, should avoid investingin double—your—money—in—a—week schemes even when it strikes themthat the scheme will work and gain them money. Some types ofappearance are known to be less reliable than others. When twotypes of appearance regularly conflict, one must know which typeto bet on, especially when these appearances are used asepistemic criteria in moral codes.It is inattention to this and other matters of psychologicalstrategy that misleads moralists such as Ross, who writes that:utilitarians say that when a promise ought tobe kept it is because the total good to beproduced by keeping it is greater than the.total good to be produced by breakingit. . . . Now, we may ask whether that isreally the way we think about promises?(1967, p. 275—76)122It seems not to occur to Ross that we may sincerely think aboutpromises and the like in more than one way, according to thesituation, yet without any deep inconsistency. This confusionremains widespread (e.g., Taylor 1989, p. 226), though it hasbeen pointed out by utilitarians for some time (e.g., Sidgwick1874, p. 413; J.J.C. Smart 1956, p. 346; and Hare 1981, ch. 2.Hare, p. 25, attributes the point to Plato and Aristotle).Epistemic RU is logically independent of ontic AU, hencedoes not require CU. Every epistemic criterion must be justifiedby some ontic criterion, but there can be rational disagreementabout which ontic criterion a given epistemic criterion isjustified by. Rule utilitarians such as Brandt typically andplausibly ultimately appeal to utility maximization whenjustifying their short list of everyday rules, which suggeststhat epistemic RU is justified by ontic AU, but perhaps someother ontic criterion is more plausible. However, the existenceof even one plausible candidate ontic criterion to justifyepistemic RU refutes the rule worship objection. Given the lackof any other plausible candidate ontic criterion and the highplausibility of ontic AU (immune to the counterintuitivityobjection to epistemic AU), CU is the best comprehensiveutilitarian theory. OCU is a cognitivist version of CU.2. Utilitarian rules123In this section I explain more precisely how different kindsof rule function in CU; in light of these functions, it will beseen that the currently prevailing forms of AU and RU, Smart’sand Brandt’s versions respectively, differ very little and areessentially the same theory. This demonstration will corroboratemy claim that a form of utilitarianism, the form assumed by OCU,can be defended that combines the main philosophical virtues ofAU and RU while retaining the vices of neither.As suggested in ch. 1, CU has deep intellectual roots. Thedistinction between the one ontic and the various episteiniccriteria for right action, and their compatibility withinutilitarianism, were appreciated by Mill: “Whatever we adopt asthe fundamental principle of morality, we require subordinateprinciples to apply it by” (1861, p. 32); by Sidgwick: “it is notnecessary that the end which gives the criterion of rightnessshould always be the end at which we consciously aim” (1874, p.413); by Moore (1912, p. 30—35); and by Hare (1981, p. 43).Regarding recent utilitarianism, CU’s reconciliation of AU andrules does not differ greatly from J.J.C. Smart’s (1973, p.42—57). CU differs from J.J.C. Smart’s version (aside from thelatter’s overt incompatibilism) chiefly in holding thathigher—level rules should be used more often in situationspermitting careful reflection than J.J.C. Smart suggests, and inholding that higher—level rules should be used as more than mererules of thumb, in that they should be invested with dignity andbroken only in rare circumstances, chiefly emergencies, not124anytime prima facie analysis suggests utility would be gained.J.J.C. Smart holds thatThe act—utilitarian will . .. regard theserules as mere rules of thumb, and will usethem only as rough guides. . . . He acts inaccordance with rules . . . when there is notime to think. . . . When he has time tothink what to do, then there is a questionof deliberation or choice, and it isprecisely for such situations that theutilitarian criterion is intended. (1973,p. 42—43)However, as will be seen, there are sound game—theoretic reasonsfor taking some kinds of rule more weightily.CU’s reconciliation of RU and the ultimacy of utilitymaximization does not differ greatly from Brandt’s (aside fromthe latter’s overt incompatibilism). Brandt agrees that thebottom line in ethics is utility maximization:if it is really true that doing a certainthing will have the very best consequences inthe long run, everything considered, of allthe things I can do, then there is nothingbetter I can do than this. . . . succeedingin producing the best consequences is a kindof success which cannot be improved upon.(1963, p. 121)125Reconciling this bottom line with the need for some rulescompliance with which calls for utility to appear occasionally tobe forgone is difficult; CU differs from Brandtian RU chiefly inholding that the rule ‘maximize utility’ would surely play acentral role in the set of rules optimal for practical use bysentients of human intelligence. This role, herein called thatof ‘top rule’, is essentially the role Brandt calls that ofa consistent and plausible “remainder—rule,”that is, a top—level rule giving adequatedirections for all cases for which thelower—level rules do not prescribedefinitely enough or for which theirprescriptions are conflicting. (1963, p.133)A top rule is an epistemic criterion distinguished by its soleoccupation of the highest hierarchical level in a moral code’shierarchy of rules for practical use, and may or may not also bethe ontic criterion behind the moral code’s epistemic criteria.Brandt thinks it an open question which rule should be the top.rule (remainder—rule) in a human society, whereas CU’s top rulein such circumstances is, for reasons of human psychology, always‘maximize utility’.These differences are minor; CU’s main innovation is showinghow these two theories, J.J.C. Smart’s and Brandt’s, are in allfundamentals consistent, and are essentially the same theory.To see this similarity, let’s consider in more detail the126role according to CU of the rule ‘maximize utility’. Asutilitarianism’s ontic criterion of right action, the rule‘maximize utility’ guarantees success if followed. However, asan epistemic criterion this rule is of limited value, in that inmany contexts right acts can be identified more reliably by theircontingent possession of some other property, e.g., that of beingthe keeping of a promise, which thus specifies a rivalutilitarian epistexnic criterion of right action. This structureis game—theoretically identical to that by which chess’s onticcriterion of right action, the rule ‘trap opponent’s king’,guarantees success if followed, but is of such limited epistemicvalue that the acts it specifies are usually more easilyidentified by means of other epistemic criteria, stated in ruleform as, e.g., ‘castle early in the game’, ‘advance weak piecesbefore strong pieces’, and ‘control the center of the board’(Hare 1981, p. 37—38, agrees).It is contingent whether a given sentient should include therule ‘trap opponent’s king’ in its set of chess—playing rules. Arule—following computer could be built with hardware powerfulenough to learn the formal rules of chess but scanty enough thatalmost no room was left over for rules of strategy; no doubt asentient of like chess capacity is causally possible, so that thesentient could play chess but absorb so few rules of strategythat the rule ‘trap opponent’s king’ would be crowded off theoptimally successful list; e.g., perhaps there would be room fornothing closer than ‘take opponent’s pieces’. Likewise, perhaps127there is a low level of human intelligence at which the humancould absorb so few moral rules that the rule ‘maximize utility’would be crowded off the optimal list. However, it isprohibitively unlikely that a sentient of normal humanintelligence could absorb so few chess—playing rules that therule ‘trap opponent’s king’ (or some logically equivalentexpression) would fail to appear on the optimal list; how couldsuch a human, not even knowing the object of the game, beat ahuman who did know? Likewise, it is prohibitively unlikely thata sentient of normal human intelligence could absorb so few rulesfor living in accordance with utilitarianism that the rule‘maximize utility’ would fail to appear on the optimal list; howcould such a human, not even knowing the object of the lifestyle,live it better than a human who did know?Recall that the appearance of the rule ‘maximize utility’ onone’s list does not preclude the simultaneous lower—level listingof ‘refrain from maximizing utility’ as a means to executing theformer rule. Like the fish—spearer throwing the spear away fromwhere the fish looks to be, the utilitarian may refrain fromdoing what they think would maximize utility because they doubtthat doing that act would maximize utility. Just as theintelligent fish—spearer can use the rule ‘spear away from wherethe fish looks to be’ more efficiently with than without alsohaving ‘spear fish’ on (a higher level of) their list of rules,the intelligent utilitarian needs the rule ‘maximize utility’ ifthey are efficiently on occasion to refrain from doing what they128think would maximize utility.Regarding the question whether and how to use the rule‘maximize utility’ to further the moral ideal of maximizingutility, CU supplies all its human adherents with the onticcriterion rule, ‘maximize utility’, for use as an epistemic toprule, for two reasons.The first reason is that as an empirical fact humans are inthe long run happiest in a fairly egalitarian society with a freeflow of information. In such a society it is difficult andexpensive for CU rules or anything else known to more than a fewpeople to be kept secret. How could one prevent the people of asociety largely accepting and practicing the epistemic RU shortmoral code from finding out that ‘maximize utility’ is the onticcriterion of their moral code? Surely this moral code willpromote education, including university ethics courses, in whichthis information would normally be taught. And if people in thissociety know their ontic criterion of right action, will it notoccur to them to use it from time to time as an epistemiccriterion, for efficiency as the top rule? Surely it will, forpresumably this moral code will encourage its citizens to thinkcreatively and look for new, more efficient ways of gettingthings done. Better, therefore, to accept that the utilitarianontic criterion will be widely known and often usedepistemically, and to plan for this inevitability by promotingcustoms and techniques for properly using ‘maximize utility’ asthe top rule (like dealing with teen lust by promoting condoms129and masturbation rather than trying to suppress sexualknowledge).The second reason CU supplies all its human adherents withthe ontic criterion rule, ‘maximize utility’, for use as anepistemic top rule, is that use of this rule indispensablybenefits human identification of acts that maximize utility, aparticular instance of the benefit of free information flow.Society under CU retains some division of labor, so that somerules are relevant only to some occupations or social positions,so that it is unnecessary for any one person to learn all therules, but anyone may know any rule they wish to.To illustrate the benefits of efficient identification ofright action, consider Brandt’s suggestion of the possible toprule (remainder—rule):‘do act x if f a person who knew the relevantfacts and had them vividly in mind, had beencarefully taught the other rules of thiscode, and was uninfluenced by interestsbeyond those arising from learning the code,would feel obligated to perform that action’(paraphrasing 1963, p. 133)While it is an empirical matter whether utilitarians usingBrandt’s suggested top rule would do better than those using‘maximize utility’ as their highest—level epistemic criterion,human history reveals so many horrendous events being started bypeople extrapolating from their moral intuitions in a sincere but130vague way uninformed by metaethical theory that it is more likelythat utilitarians using Brandt’s proposed top rule would doworse, as game theory predicts via the chess example. The roadto hell is paved with intentions coherent with good ones butvague and untested themselves, and invented by sentients evolvedto have a significant capacity for both selfishness (Hardin 1985,p. 132) and self—deception (Trivers 1985, ch. 16).Consider a sample advantage of the episternic criterion ofutility maximization over other possible top rules. Since thiscriterion is also the relevant ontic criterion, it is uniquelyuseful for quickly identifying objects with high probabilities ofhelping constitute states of affairs falling under thejurisdiction of moral rules. A human chess player aware thattrapping opponent’s king is the object of chess will be well ableto spot the advantage of monitoring the defensive capacity ofopponent’s king and surrounding pieces, whereas an otherwisesimilar player unaware of the importance of trapping opponent’sking would be significantly more likely to spend resourcesmonitoring situations elsewhere. Similarly, a human moral agentaware that maximizing utility is the object of moral activitywill be well able to spot the advantage of monitoring the welfareof sentients affected by its actions, whereas an otherwisesimilar agent unaware of the importance of maximizing utilitywould be significantly more likely to spend its resourcesmonitoring other events, e.g., soap operas or sports programs.Another advantage for humans of the epistemic criterion of131utility maximization used as a top rule is its help ininterpreting other epistemic criteria even in the absence ofconflict. As Diggs says, the follower of instrumental rulesis often expected to know the goal to whichhis rule—directed action supposedlycontributes——to know “what he is doing” inthis sense. Not always, to be sure, butoften he could not make a sound judgment ofwhen and how to apply the rule without thisknowledge. (1978, p. 214)Though unrecognized by Mill (1861, p. 33), this advantage issignificant; imagine trying to teach someone a canoepaddle—stroke without telling them that the point of theoperation was to push water rearwards. Brandt agrees that histop rule should perform this function (1963, p. 133), but hissuggested rule seems inadequate: by analogy it would advise thestudent of paddle—strokes to:‘move the paddle thus if f a person who knewthe relevant facts and had them vividly inmind, had been carefully taught the otherstrokes of this paddle, and was uninfluencedby interests beyond those arising fromlearning the strokes, would feel inclined tomove the paddle thus’A canoe—paddler taught in this way may well get across the lake,but is unlikely to set a speed or safety record.132Thus while Brandt does implicitly accept ‘maximize utility’as specifying utilitarianism’s ontic criterion of right acts, hedoes not think it obvious that this ontic criterion is also themost efficient top rule epistemic criterion for humans. CU holdsthat this efficiency is safely inferrable from currentlyavailable data (Hare agrees — 1981, p. 50). Thus the differencebetween CU and Brandtian RU is minor.CU does not differ greatly from (J.J.C.) Smartian AU either:it differs chiefly in holding that some rules should be followedalmost exceptionlessly rather than merely used as guidelines,even when there’s time to think carefully about the consequencesof one’s actions. Rules such as ‘keep your promises’, ‘providefor your children’, and ‘don’t punish the innocent’ should befollowed more regularly than mere rules of thumb or roughguidelines.One game—theoretic justification for following such rulesalmost exceptionlessly even when there’s time to think carefullyis that some such rules are like ‘drive on the right’ in that oneof their chief benefits is their tendency to coordinatecooperative behavior. The exact nature of the rule is lessimportant than that everybody follow the rule. Some similarrival version of the rule might have worked just as well, mutatismutandis, e.g., ‘drive on the left’ rather than ‘drive on theright’; likewise, rules such as ‘provide for your left—handneighbour’s children’ are just as intrinsically morallyacceptable as ‘provide for your children’; as long as all133children are properly cared for by some adult, and the work isshared around, everybody’s happy and morality is satisfied. Forvarious contingent reasons of psychology, the latter rule is morelikely to endure in a society than the former, hence isextrinsically morally better. The past enduring presence in asociety of one rule from a set of such intrinsically equallyefficient rivals is evidence that the rule will continue toendure. The rational moralist will therefore endorse and followwhichever rule of that set of rivals is predominant in localsociety, not worrying too much about which rule happens topredominate. Since the rule works best when followedexceptionlessly, and since there is no need to worry aboutalternative rules, the utilitarian follows the rule every time,even when there is plenty of time to think about alternativeaction, save in rare emergencies.Another game—theoretic justification for following suchrules almost exceptionlessly even when there’s time to thinkcarefully is that some such rules, e.g., ‘keep your promises’,are like ‘use tit—for—tat strategy in iterated prisoners’dilemmas’ in that they work best when followed exceptionlessly,and can be proven to be more efficient mechanisms of desiresatisfaction than their rivals, so that again there is no pointin thinking the matter over afresh each time one encounters therelevant situation. The institution of promise—keeping amongnormal humans always increases happiness over the lack of such aninstitution. The institution works best when promises are always134kept save in rare emergencies, mainly because while some lapseswould be tolerable, we have little idea where lies the thresholdbetween tolerable and intolerable lapses, and little idea how ourlapses would affect the tendency of others to lapse (see Lyons1967, p. 74). Once one gets in the habit of promise—breaking,one (and one’s partners in promising) regularly lose benefitsregularly secured by other sets’ of promise—keeping partners. Therational moralist will therefore know that there is nothing togain by rethinking the matter every time a promise comes due.Thought is only necessary when deciding whether to make a newpromise to someone with a spotty record of promise—keeping toyou. Once a promise is made, it is always right to keep it savein rare emergencies. Such solemn promises must be distinguishedfrom joking promises, only lightly—binding promise—likeassurances, and so forth; this is most efficiently done by meansof ritual oaths which by social convention bind when uttered,e.g. ‘I swear on my honor to do x’.The option of breaking a lower—level epistemic moral rule(such as ‘keep your promises’) in rare emergencies must beincluded partly to handle the catastrophic—alternative objectionto demanding that any such rule be followed exceptionlessly, andpartly to allow agents to secure the occasional large windfallbenefits that help make possible the pleasant life that sentientsshould have, but which sometimes can be had, due to windfalls’arising unpredictably, only at the cost of breaking a lower—levelrule.135Regarding the first of these two considerations, thecatastrophic—alternative objection, it must be conceded thatsometimes lower—level rules must be broken. Any plausibleutilitarianism must make this concession to the complexity of theworld, for history shows that sometimes catastrophe the onlyalternative to breaking the lower—level rules of commonsensemorality. For instance,.it is part of common—sense morality, apart that presumably would find rational expression as alower—level utilitarian rule, that cannibalism is wrong, thathumans should not use human bodies as food. Yet survivors of aircrashes in remote areas have sometimes found it necessary to eathuman bodies in order to avoid starvation. Community moralintuitions tend to endorse such action. When rescued, suchsurvivors generally receive sympathy rather than censure. AsDiggs says of instrumental rules, “it not only makes sense tospeak of its being proper to violate a rule, “successfulviolations” tend to be commended” (1978, p. 215). Suchviolations make sense as applications of the top rule, ‘maximizeutility’, to adjudicate on difficult cases falling underlower—level rules.Thus those who hold that “fairness demands that there be noexceptions” (Lyons 1967, p. 163) to moral rules are simplywoolly—minded or fanatical, for if one takes their wordsliterally, then one must infer that they would march innocentsoff cliffs for the sake of exceptionlessness. If fairness doesrequire such fanaticism, then it cannot be the only or highest136value. A utilitarian version of fairness, interpreted as anepistemic criterion of right action that is widely useful inpractical life, is more plausible, for it admits of exceptions.However, to prevent abuse of this rare—emergency loophole,human moral agents need training and guidance in the use of theircode’s ontic criterion as a top rule:Although it is difficult to specifyconditions in which the violation of aninstrumental rule is proper, surely the barefact, “that by doing so one can betterpromote the goal,” is not sufficient. Therule follower is not the sole or finalauthority on the propriety of breaking arule, even when it is for the benefit of theother party. (Diggs 1978, p. 214)Many possible sources of human abuse of ‘maximize utility’ as theutilitarian top rule could be acceptably minimized by thedevelopment of social norms for such use. Emergencies may berare, but their general natures are forseeable, hence may bediscussed and planned for. To extend Mill’s nautical analogy(1861, p. 31), sailors encounter big storms only rarely, and haveat such times to act quickly in ways that greatly affect whethertheir fellow mariners live or die, sometimes having to choosebetween acting in accord with a high precept of ordinary goodsailing, and acting to save many other people. Even so, sailorsas a group can build up a tradition of knowledge to deal with137such crises, and as individuals become experienced at using thisknowledge during crises. There is no reason to think that thegeneral method sailors use to do this could not be equally wellapplied to dealing with rare moral crises (see Hare 1981, p. 52).Regarding the second consideration, that agents must be freeto secure occasional large windfalls at the cost of breaking alower—level rule, again the utilitarian calculus and commonsensemorality unite in endorsing such action. The calculus makes itplain that a sufficiently large windfall will greatly outweighthe modest harm a single violation of a lower—level rule tends tocause. Potential windfalls are not circumstances in whichcatastrophe looms, in the sense that a large amount of pain couldbe avoided but will occur as things stand; rather, they arecircumstances in which a large amount of pleasure could berealized, but will not as things stand. In acatastrophe—avoidance emergency, prompt action is needed to avertpain; in a windfall—acquisition emergency, prompt action isneeded to gain pleasure. A utilitarian moral code should try tohandle cases of catastrophe—avoidance and windfall—acquisitionroughly similarly, for according to the utilitarian calculuspleasure and pain are equally components of utility. The equalimportance of pleasure and pain is reflected in utilitariansupererogation (discussed in detail in ch. 5), in which agentsare praiseworthy for doing good deeds beyond what is morallyrequired; the good deeds may be either the creation of pleasureor the prevention of pain.138Windfall—acquisition emergencies are likecatastrophe—avoidance emergencies in that a little rule—breakinggains a lot of utility, so since it is, as Diggs suggests,justified to break a moral rule in a catastrophe—avoidanceemergency (e.g., to break the moral rule of obeying the law ofthe land, by safely breaking the speed limit to drive a strokevictim to hospital), it would also seem justified to break amoral rule to secure a windfall. It might be objected that: onthe one hand, we condone rule—breaking in catastrophe—avoidanceemergencies because while rule—breaking tends to cause a littlelong—term pain, a lot of pain will occur anyway if prompt actionis not taken; thus pain is unavoidable though minimizable. Onthe other hand, the rule—breaking to secure a costly windfallwill cause pain that was avoidable; if the rule were followed,the windfall of pleasure would be lost, but no pain would occur.Since, for contingent reasons of human psychology, it isextrinsically more important not to cause pain than to createpleasure (Sikora, unpublished), there is a utilitarian reason toforego costly windfalls that does not apply to emergencies.Similarly, supererogation does not involve rule—breaking, so doesnot involve the pain that rule—breaking tends to cause, so doesnot run afoul of the utilitarian reason to forego costlywindfalls.Nonetheless, inspection of everyday human attitudes andbehavior suggests that if the windfall is big enough and therule—breaking cost small enough, we endorse the rule—breaking139behavior, and may even rebuke an agent who refrains from breakingthe rule. For instance, if an unemployed person with children tosupport passes up an unexpected offer of a great job, in order tokeep a promise to meet a friend for coffee, then even the friendis likely to chide the promise—keeper. Thus the utilitarianreason mentioned, that counts against costly windfalls but notagainst catastrophe—avoiding emergency rule—breaking orsupererogation, seems to have the moral effect not of making thepursuit of costly windfalls irrational given utilitarianpremises, but merely of setting the amount of utility gained (inthe form of windfall pleasure), necessary to justify the costlywindfall rule—breaking, as larger than the corresponding amountof utility gained (in the form of pain avoided), forcatastrophe—avoidance emergency rule—breaking. This differencebalances the subtle long—term harm that acquiring costlywindfalls causes by creating avoidable pain.These two considerations in defense of emergencyrule—breaking represent the two extremes on the continuum of allpossible utility packages to be avoided or acquired inemergencies. Catastrophe avoidances are avoidances of much pain,and windfall acquisitions are acquisitions of much pleasure. Inpractice, the utility packages to be acquired or avoided inemergencies are seldom composed of either pure pleasure or purepain. Usually they are composed of a mixture of pleasure andpain, though so preponderantly of one or the other as toprecipitate an emergency. In the case of a mixed catastrophe140avoidance, the utilitarian calculus unproblematically deems itworthwhile to avoid a little pleasure in order to avoid a lot ofpain. In the cased of a mixed windfall acquisition, theutilitarian calculus adds the small quantity of pain in theutility package gained to the small quantity of pain caused bythe rule—breaking, and again deems the sacrifice worthwhile ifthe ratio of pleasure gained to pain endured is sufficientlygreat.It might seem that there is a further non—emergency class ofcases in which the breaking of moral rules is permitted, namelythose cases in which only small amounts of utility stand to begained, but the agent is justifiably sure that this utility wouldoutweigh the utility lost by the rule-breaking. However, thereare two.objections to this view. First, as discussed in moredetail elsewhere in this chapter, the small amounts of utilitythus gained would be outweighed by the utility lost by other,less able agents being unjustifiably sure of gaining utility insimilar situations in which utility would in fact be lost.Second, there is no need for such a practice, for there alreadyexists a type of weak, flexible moral rule that covers suchsituations adequately.This type of rule is the moral equivalent of the legal ruleagainst jaywalking. Such rules are available to governsituations that are usually unimportant but sometimes veryimportant, even potentially lethal. The important subsets ofthese situations are impossible to pick out accurately by141expressions concise enough for ready human use, hence must becovered by simple but greatly overreaching expressions. It isfor contingent psychological reasons easier for humans toassimilate the more detailed criteria of the rule’s use not aslinguistic expressions but as habits of application. Thus weremember the linguistic rule ‘don’t jaywalk’ but break this rulevery often, e.g., when traffic is light and no crosswalk ishandy.Jaywalking—type rules function similarly to proverbs such as‘many hands make light work’. Proverbs are easy to remember andapply, but if taken literally and applied exceptionlessly wouldbe hopelessly awkward and disruptive, as evidenced by thecontrary proverb ‘too many cooks spoil the broth’. These twosample proverbs coexist despite formal inconsistency preciselybecause each can be safely ignored in most of the situations towhich it literally applies. The detailed behavioral criteria ofuse for jaywalking—type rules can be verbalized if necessary,e.g., when one is explaining one’s actions to a traffic cop, butnormally are not verbalized. A moral rule of this type is ‘don’tspeak ill of the dead’. Unlike the stronger moral rulesdiscussed earlier, jaywalking—type moral rules should not befollowed with almost no hesitation — rather, a good deal ofhesitation is desirable.Given that there are various kinds of rule that autilitarian ought to follow with almost no hesitation, and thathumans are weak—willed and suggestible at times, it follows that142utilitarians should encourage humans in their society to regardsuch rules with respect, even to venerate them somewhat, andritualize their performance. Such cultural practices areeffective in encouraging willing human compliance with suchrules. Venerating such rules makes breaking them in rareemergencies stressful, but this disutility is overall slightprecisely because emergencies are rare; since veneration makes iteasy and natural to follow these rules on other occasions, on thevast majority of which utility is gained, veneration short offanaticism is helpful. Rules treated in this respectful way aretherefore being used as more than mere rules of thumb or roughguidelines. Some rules therefore play stronger roles withinutilitarianism than Smartian AU suggests is rational: However,since Smartian AU is perfectly consistent with rule—use per Se,and since CU’S improvement in the details of utilitarian rule—usefollows from the application of game—theory to the task ofutility—maximizing in human society, an application J.J.C. Smartendorses (1973, p. 57), the difference between CU and Smartian AUis minor.(More precisely, I term this a minor difference because itis a difference at a low theoretical level, merely on the detailsof application of metaphysically well—understood higherprinciples to which both sides agree; hence, the disagreementwould seem to be easily resolved. Only a little empiricalspade—work is required, much as in a disagreement between twochess players about whether one should castle early in the game143as a matter of course, almost exceptionlessly, or whether oneshould castle early in the game only after thinking how this fitsinto the rest of one’s game plan as constrained by thedevelopment of one’s position thus far. The answer in chessturns out to be similar to the answer to the moral question.Weak players should castle religiously, and strong players shouldplay it by ear. Similarly, agents of merely human intelligenceshould follow the relevant moral rules religiously, and agents ofsufficiently superhuman intelligence should use the relevantrules as rules of thumb).Thus Lyons’ claim that rules endorsed by AU are eithertrivial or not rigorously generated (1967, p. 119—20),interpreted as an objection to CU (as Lyons intends — p. 145),trades upon the fact that the generation of CU rules, which usesontic AU as a premise, also requires as premises some contingentfacts about the psychology of the agents whom the rules will beused by and about the habitat these agents live in. Withoutthese facts only trivial rules follow (e.g., ‘lying is wrongexcept when it gains utility’), but with these facts someimportant high—level rules follow, as shown above. The identityof the top rule is shown to be ‘maximize utility’, and rules suchas ‘keep your promises’ are shown to be worth following withoutexception save in rare emergencies for which identifying criteriacan be taught. Such moral rules are as well—evidenced as chessrules such as ‘trap opponent’s king’, ‘castle early in the game’,‘advance weak pieces before strong pieces’, and ‘control the144center of the board’, hence these moral rules are generated asrigorously as any useful noncontingent rules can be. It might beheld that nonetheless rules of this general type do not deservethe label ‘rigorous’, but since such rules, in chess and inethics, achieve their objectives when followed and have rationalgame—theoretic justifications, this linguistic deprivation lackssting.Lyons also objects that ethics needs non—utilitarianconditions for justifiably breaking rules such as ‘keep yourpromises’, for if only the utilitarian “condition had to besatisfied, we would be justified in breaking rules much moreoften than we normally think we are’ (1967, p. 186). However,Lyons offers no empirical evidence for this contingent claim, andgiven the frequency with which people in fact normally do breakpromises, presumably thinking in most cases that they arejustified in doing so, this accusation is harsh. I can offer nodirect empirical evidence to the contrary, but am strongly of theopinion that indirect evidence, by arguments too lengthy to gointo here, suggests that the reverse is the case, so that if a CUmoral code were generally accepted by humans, fewer promiseswould be broken.Another objection to the justification of epistemic RU byontic AU is that while ontic AU would admittedly entail that mosthumans should use RU for their epistemic moral code, it(allegedly) would also entail that exceptionally intelligenthumans should not be encumbered by epistemic RU, for their145superior abilities at spotting beneficial ways of diverging fromthe conduct required by epistemic RU would thus be wasted. Suchgeniuses would therefore be justified in adopting personal moralcodes, secret as far as possible, diverging from epistemic RU.However, the problem of self—serving rationalization byindividual code—choosers would remain for this view, forintelligence is a characteristic possessed independently of basichuman emotions. Geniuses are as likely to be tempted to actselfishly as most other people. Moreover, some nongeniuses thinkthat they are geniuses, and would choose individual moral codeswithout being able to thus behave significantly closer to perfectutility—maximization than they would have under ordinaryepistemic RU. Meanwhile, the gains to be had from the use ofexceptional moralities by real geniuses would be much smallerthan might first appear. For instance, since these geniuseswould still have to support epistemic RU for general public use,and since public subscription to epistemic RU would be underminedby public observation that geniuses were in practice oftenignoring epistemic RU, the moral code geniuses publiclyadvocated, geniuses would therefore have to publicly obeyepistemic RU.If geniuses were readily distinguishable from other humans,then some aspects of this problem would disappear, for geniuseswould not have to keep their individual moral codes secret. Forinstance, some loss of utility due to unpredictability would beavoided, for no one would expect geniuses to act just like other146people — they could be predicted to behave unpredictably, withoutnongeniuses’ behavior also being unpredictable. However,geniuses are in fact indistinguishable from other people in allsurface characteristics, so this would not work; hence geniuseswould have to keep their individual codes secret.Another reason why the opportunities. for diverging fromepistemic RU would be rare, and the resulting gains small, isthat CU, by its explicit inclusion of ‘maximize utility’ as thetop rule, allows a highly intelligent human to act properly inany emergency that this person’s unusual gifts allow them todetect, without transgressing the moral code. Because all peoplein the society are trained to deal within the moral code withemergencies, rationalization respecting emergencies can be keptto an acceptable minimum.Thus opportunities for geniuses’ esoteric moralities todiverge from epistemic RU would not be during emergencies, whenthe potential gains would be great, and in any case would berare, for the reasons given earlier. Rare, individually smallgains add up to not much - a quantity easily outweighed by thedisadvantages mentioned.Since a form of utilitarianism, CU, the form assumed by OCU,can thus be defended that combines the main philosophical virtuesof AU and RU while retaining the vices of neither, the thirdobjection to OCU is refuted.147Chapter FiveSUPEREROGATION1. Small beneficial sacrificesThe fourth objection to OCU is the objection fromsupererogation.A supererogatory act is, to define it in as much detail asis here needed, an act, beyond the call of duty, that benefitssomeone other than the agent, and is altruistically motivated.Saints and martyrs are extreme examples of the kinds of people towhom supererogatory acts have traditionally been attributed,though at the other end of the spectrum, very small altruisms maybe supererogatory, e.g., taking a stray cat to the SPCA. In thecontext of OCU, a supererogatory act is an altruisticallymotivated act, uncalled for by any rule in the short moral codeof the agent’s society (i.e., beyond the call of duty, hence, ina sense defined below, not obligatory), that both increasesoverall utility and increases utility for other sentients. Theagent typically sacrifices some of its own utility to make theseincreases, but self—sacrifice is not essential to supererogation.The fourth objection to OCU is that it has the unworkablefeature of deeming some acts supererogatory. OCU’s form of148supererogation can be criticised in three ways, some waysapplying to all forms of supererogation, other ways applying onlyto some forms, including OCU’s.The first criticism, which might be termed the objection ofsmall beneficial sacrifices, is that under this definition of‘supererogation’, any personal sacrifice, no matter how small,for the sake of any gain for others, no matter how great, wouldcount as supererogatory as long as it increased utility butwasn’t called for by the short moral code of the agent’s society;by the intuitions of some, that doesn’t seem right (Sikora,personal communication). However, my proposed definition of‘supererogation’ agrees with others in the literature in thisregard, so is intuitively acceptable at least to some. Mydefinition agrees with other definitions in the sense that eachof these other definitions conjoined with OCU entails that anypersonal sacrifice, no matter how small, for the sake of any gainfor others, no matter how great, would count as supererogatory aslong as it increased utility but wasn’t called for by the shortmoral code of the agent’s society. For instance, Montague writesthatx is supererogatory for y (the personperforming x) if and only if x is notrequired and y is praiseworthy for performingx. And given the nature of praiseworthiness,y is praiseworthy for performing x only if xhas moral value and is not prohibited. Some149little benefactions are neither required norprohibited, but are praiseworthy if performedwith appropriate motives and beliefs. Hencesome little benefactions . . . aresupererogatory for their agents. (1989, p.108)Making a very small, uncalled—for--by—the—rules sacrifice for avery great gain for others is analytically not required by therules. Since such sacrificial action gains utility, it has moralvalue, hence, in at least some cases let us suppose, is notprohibited by the rules of the short moral code. Since suchaction has moral value and is not prohibited, it is praiseworthyby Montague’s definition. Since such action is praiseworthy butnot required, it is also by Montague’s definition supererogatory.Thus I use ‘supererogatory’ to pick out a region of acontinuum; anything over the relevant boundary, no matter howslightly over, is strictly speaking supererogatory. Analogously,near—tepid bathwater may still be on the hot side of thezero—point of the intensity continuum exhausted by hotness andcoldness. The boundary OCTJ uses is the conjunction of thezero—point of the quantitative continuum of overall utilitygained, and the zero-point of the quantitative continuum ofutility gained for other sentients. If both boundary conjunctsare crossed, no matter how slightly, so that both quantities arepositive, no matter how slightly, then the altruistic act inquestion is supererogatory. The amount of utility accruing to150the agent is irrelevant. It is typically negative, but may alsobe zero or positive. For instance, the agent may enjoy helpingothers, enough that the utility gained by the agent equals oroutweighs that lost in the doing of the deed, e.g., the mailingof a small check to a charity. (At least, these quantities ofutility are the ones relevant to our initial thinking onsupererogation, when we take a God’s—eye view of events; later,qualifications will be added to take account of the epistemicproblem that acts do not always yield the quantity of utilitythat seemed probable by a given agent’s evidence).Another definition of ‘supererogation’ agreeing with mine inthe relevant respect is Langford’s definition:works of supererogation . . . comprise thekinds of actions that it is generally heldthat we ought to perform for our closefriends, but that it is not generally heldthat we are normally obliged to perform forour neighbour or a stranger, but which aredone, nevertheless, for a neighbour orstranger. (1988, p. 442)It is surely required by OCU’s short moral code that, if theopportunity arises, one make a very small sacrifice to secure avery great benefit for a close friend, yet not required that onedo this for a mere neighbour or stranger. For example, if anelderly close friend is moving, then (if I am healthy) I amremiss if I do not offer to help shift the larger items, though151not if I notice that a similar mere neighbour or stranger ismoving, and make no offer. While it is doubtful that every actdeemed supererogatory by OCU is of a kind that one ought toperform for one’s close friends, it is very plausible that allacts involving very small sacrifices for very large gains areobligatory for close friends (since one has few close friends, sowill not be overburdened), yet are merely permissible forstrangers and suchlike (whose numerousness would overburden theagent often enough to prevent the relevant rule from enduringvery long in the moral code).Similarly, Dancy states that:supererogatory acts are those that, thoughthey have merit or value, still lie beyondthe call of duty. At the limit, these arethe actions of saints and heroes, actionsthat, though of supreme merit, cannot besaid to have been the agent’s duty. Thereare also supererogatory actions of lessexceptional value; quite ordinary actionscan exceed the demands of duty, while stillattracting approval. (1988, p. 173)I shall not argue the point in detail, but this definition alsoseems to agree with OCU’s on the salient point.Thus there is among those who have studied the subjectsignificant intuitive support for the view that any personalsacrifice, no matter how small, for the sake of any gain for152others, no matter how great, would count as supererogatory aslong as it was gainful on balance but wasn’t called for by therelevant moral code. Applied to OCU, this would pick outsuitable acts that weren’t called for by the short moral code forthe agent’s society. It is therefore incumbent on those who findthis view of supererogation unintuitive to develop a moreexplicit theoretical objection to it. Meanwhile OCU assumes theview, and incorporates it into the definition of‘supererogatory’. Thus the objection of small beneficialsacrifices fails.2. Valuable but not obligatorySome hold that the practice of supererogation is unjustifiedbecause the concept of supererogation is paradoxical, in that asupererogatory act would make things better all thingsconsidered, but is not obligatory. This nonobligatoriness seemsinconsistent withthe resilient intuition that if an action isgood, this gives us a reason to do it, and ifit is the best available, we have more reasonto do it than to do anything else and soought to do it. If this intuition is sound,the best available action will always be theright one, the one that one ought to do.153(Dancy 1988, P. 176)This general criticism of any morality’s form of supererogationis starkly applicable to OCU’s, for according to OCU all and onlyutilities intrinsically ground reasons for action. Call thiscriticism of supererogation the ‘justification objection’. Onemight save OCU from this serious metaethical objection byabandoning supererogation, but this seems too high a price.Supererogation is familiar enough as a feature of benigncommonsense moralities around the world that we may speculatethat it is in some way justified. I shall argue that certainevidence renders this speculation plausible. Kagan writes:according to Heyd, the justification ofsupererogation has two “aspects”: thenegative aspect shows the justification forlimiting what is morally required; thepositive aspect shows “the value ofnon—obligatory well—doing as such” (1984, p.241)I shall follow this program of justification.The negative aspect of OCU’s justification of supererogationis the aspect of limiting what is morally required by showingthat the short moral code optimal for human societies makesobligatory only a proper subset of the many valuable acts open toa typical human agent.This objection to OCU’s form of supererogation, that itcannot so limit obligation, might be termed the objection of154pervasively many obligations (adapting Montague’s phrase — 1989,p. 104), and is due to Baier (1958, p. 203—04). It is theobjection that since it is obligatory for the utilitarian agentto act on every occasion on which it has the opportunity toincrease the world’s utility, the utilitarian agent will find itsentire waking life dense with obligatory acts. Duty wouldpervade the morally obedient agent’s life to the exclusion offrivolity and relaxation. In McCarty’s words,the fantastically virtuous person rises dailywith moral determination, eats with moraldiscrimination, organizes all activitiestoward moral destinations, and retires inmoral deliberation. (1989, p. 44)OCU’s response is essentially that rules making so many actsobligatory would seldom be obeyed, hence that such rules wouldtend to come into disrepute, and gradually get dropped from themoral code, both because of fading by neglect, and because ofactive defensive suppression by moralists since flouted rulestend to bring morality in general into disrepute. The details,however, are complicated.For instance, the rules in question are divisible into twoclasses, flouted for different reasons. J.J.C. Smart’s responseto the objection of pervasively many obligations is effective inthe large first class of cases:Baier holds that (act—) utilitarianism mustbe rejected because it entails that we should155never relax, that we should use up everyavailable minute in good works, and we do notordinarily think that this is so. Theutilitarian has two effective replies. Thefirst is that perhaps what we ordinarilythink is false. Perhaps a rationalinvestigation would lead us to the conclusionthat we should relax much less than we do.The second reply is that act—utilitarianismpremisses do not entail that we should neverrelax. Maybe relaxing and doing few goodworks today increases threefold our capacityto do good works tomorrow. So relaxation andplay can be defended even if we ignore, as weshould not, their intrinsic pleasures.(1973, p. 55)J.J.C. Smart’s reply is sufficient to defeat the objection ofpervasively many obligations in all cases in which doing goodworks for others would cost the agent significantly, e.g., intime and effort. While many people have more utilitarianobligations than they fulfill, they do not have so many as tomire them in ceaseless work.In the second class of cases, where little or no sacrificeof utility, e.g., via time and energy, would be needed from theagent for each of the very numerous acts in question, OCTJ holdsthat, nonetheless, agents would not be obliged by the short moral156code to do all these acts. For instance, the billionaire neednot be obliged to sign a check for charity, even though the actwould take only a few seconds and would help sentientsconsiderably. A rich person would no doubt be required by theshort moral code to give some amount to charity, but beyond thatamount, even though the person had a great deal of money left,donation would be supererogatory. It would be supererogatorybecause a rule making such action obligatory would not be obeyedenough to endure, hence would be deleted from the moral code. Wecould logically possibly do that many things, but would notactually do them, for the causal fact is that humans are seldomso generous as to disperse their worldly goods like snowflakes.For instance, as the fall of communism showed, people deprived ofthe fruits of their labor cease to labor, even when the seizedgoods are dispersed to needy people. Therefore, as with thefirst class of cases, the obligations thus ignored would bebrought into disrepute and disappear from the moral code.It might be counterobjected to OCU’s reply to the objcctionof pervasively many obligations that in the long run it would notbring moral rules of obligation into disrepute to make therelevant acts morally compulsory rather than morally optional,for in fact humans, with proper preparation and training, couldand would act very much more diligently than they now do topromote utility. Consider, therefore, some more detailed pointsof psychology, which suggest that a moral code successfullycommitting its followers to relentless pursuit of duty would not157be a moral code conducive to utility maximization.First, relentless hard work is causally incompatible withthe balanced, thoughtful, reflective habit of mind, includinganappreciation of pleasure and emotion, that leads one to be autilitarian in the first place. John Stuart Mill and hiswell—known nervous breakdown form a typical example of a humanmind in which both types of activity were encouraged by misguidededucators (see Mazlish 1975, especially ch. 10).The belief that happiness is the only goal worth promotingfor its own sake is the central belief of hedonisticutilitarianism, so cannot be abandoned without abandoningthemain premise intellectually supporting the view that oneshouldwork ceaselessly for others’ happiness. Yet without properintellectual support, the belief that one should work ceaselesslyfor others’ happiness cannot itself survive — it graduallychanges into a related belief able to derive intellectual supportfrom non—hedonistic beliefs. Anthropology reveals thatculturesin which people work very hard are not hedonistic butdisciplinarian, not individualistic but conformist. Happinessdoes not thrive in such a society.People tend to seek happiness for themselves; just asphotosynthesizing plants are phototropic, humans arehappiness—tropic. To get people by working ceaselessly toactagainst their drive for happiness, a pro—work belief mustbemaintained. This can only be done by threat of some worsealternative, e.g., hunger or torture or death, or by persuasion158that work is better than self—happiness, e.g., by means ofcultish organizations glorifying work. Some examples that cometo mind are puritanical Protestant religious sects, the Japanesesociopolitical unit, and the German traditional nationalistmovement, with its slogan ‘Arbeit macht frei’ (= ‘work makes onefree’). The threat of a worse alternative produces a siegementality; just as many drafted soldiers in continuous combatexperience battle fatigue after a while and cannot function,previously normal people who work ceaselessly tend to crack upafter a while. People can be trained to work very hard, but mustbe raised in that mentality from childhood. Such people must beraised to be psychologically stunted; too much education willcause them to ask questions awkward to the cultish organizationdoing the training, and too wide a variety of emotions will causelongings and dreamings incompatible with ceaseless work.Psychologically stunted people en masse form societies that arerigid and insensitive to the needs of individuals, particularlythe weak or mute; such societies contain much needless sufferingand needlessly forgo much pleasure. If the resources availableto such a society suddenly increase, due to, e.g., trade accessto a hitherto unexploited land, or the acquisition of scientificknowledge permitting better utilization of local materials, thenthe hard work such a society engages in may produce prosperity.However, belief in the value of ceaseless work does not thrive ina habitat of human prosperity; it is not a coincidence that asthe three work—glorifying ideological organizations noted above159began to achieve prosperity, permitting education and spare time,their members began to lose their zeal for work.Thus the objection of pervasively many obligations,respecting cases of supererogation requiring significant agentsacrifice, depends upon a false factual premise about the amountof altruistic good work human beings are on average capable of inthe long run. Utilitarians agree, or ought to, that humansshould do as much altruistic good work as they are capable of inthe long run, but in fact most humans aren’t capable of verymuch. Humans who work relentlessly hard tend not to do work thatcreates happiness, and humans who do work that creates happinesstend not to work relentlessly hard. There are sound scientificreasons for thinking that this situation cannot be changed.Some humans are capable of both relentless hard work andethical sensitivity, but they are very much in the minority.With appropriate psychological and genetic engineering, perhapsall humans could be made this way. However, it seems unlikelythat a society of such humans would be stable. Relentless hardwork may help us get to utopia, but once we are there it isunnecessary. Hi—tech humans, like humans in some hunter—gatherersocieties, need not work more than 2 or 3 hours a day to keepsociety running. The bulk of the work performed by contemporaryhumans either is needless, irrational busywork, or is needed onlyto counteract the blunders of the irrational majority — but inutopia, irrationality would be rare among humans. Theinhabitants of utopia would have no reason to maintain the160artificial controls required to sustain the capacity forrelentless hard work, so being rational, would not maintain it(if they had ever begun it, which is doubtful).Second, similar considerations of psychology apply to casesin which little or no time or energy need be expended toaccomplish the supererogatory act. In a society in which manyare poor, so that there is significant need for such acts, humancompetition for worldly goods will tend to be intense enough tokeep human nature sufficiently selfish on average that moralrules making such generous action obligatory would not be obeyedvery often. The moral rules in question would therefore fallinto disrepute and disappear from the moral code. In a societyin which few are poor, so that there is no significant need ofsuch acts, there will hence be no significant need for moralrules making such generous action obligatory. Ordinarytax—supported programs will be sufficient to ensure that no oneinvoluntarily goes without the basic goods of life.These facts about human nature will be reflected in arational utilitarian moral code by means of rules specifying themodest number of altruistic good deeds the average human shouldfeel obliged to do. Individuals are permitted to do more —supererogation is not banned by utilitarianism — but individualsusing this liberty should be warned of the possible harmfulside—effects of supererogation, and trained in testing themselvesfor possible damage, much as athletes are trained in elementarysports medicine so that they can watch for possible harmful161side—effects of the unusual strains their bodies undergo. Theharm to supererogators might be physical, such as when a chronicvolunteer ruins their bodily health, or mental, as when a chronicdonator of money loses their ability to say ‘no’, and sendschecks to charity so often as to squander their savings for theirold age, as many people have in response to televangelists’ pleason behalf of pseudocharities. No rule requires strong—willedhumans to perform these supererogatory acts, even though they arecapable of them without self—harm or harm to others byside—effect, because it is too difficult, hence costly, toidentify these rare individuals among a population of ordinaryhumans. The extra cost involved would cancel the gain in utilityotherwise resulting.AU has received unwarranted criticism via the objection ofpervasively many obligations, because it has not been widelyappreciated that AU can make such moderating rules for citizens.RU is liable to the same technical objection, but has largelyescaped criticism because it is easier to see that RU rules mustbe designed to endure and be generally obeyed in a society ofordinary, weak—willed humans. As we have seen, AU and RU are,properly construed, the same moral theory, CU, which hence is thebeneficiary of all the intuitive support RU enjoys on this issue,as hence is OCU also.The objection of pervasively many obligations thus dealtwith in the abstract, we may now examine just what sorts of rulewill bring about the appropriate rate of self—sacrificing action162in a utilitarian society. While it is clear that for the averagehuman the sacrifices involved in much supererogation are too highfor the gains made, in that the average human could not keep upthis level of work unrewarded by personal happiness for long, andthe gains made would be insufficient to warrant asking for thesacrifices anyway in order to get even a small measure ofcompliance, it might be asked just how the correct rate ofself—sacrificing action is to be determined. The correct rate ofself—sacrificing action depends upon the correct sacrifice/gainratio, but what ratio is that? The straightforward answer isthat the correct ratio is anything less than 1; that is, actionin general is prima facie demanded by utilitarianism anytime anet gain in utility would result, and this applies no matter whogains and who loses. Thus if I would lose 100 net hedons(measures of happiness) by act A, and others would gain 101 nethedons, then I ought to do act A, in the sense that it would bebetter if I did act A, hence that the short moral code ideallywould find a way of directing me to do act A. However, and theimportance of this qualification cannot be overstressed, it isvery difficult to tell whether this ratio obtains of a given act,and, because of obscure long—term considerations, many sorts ofact of which this ratio appears to obtain do not in fact yieldany net gain when performed by an average human. Thestraightforward answer given above is an ontic criterion, whereasfor practical purposes we need to know what epistemic criteriaidentify the acts thus specified.163Since matters of general human behavior are, according toOCTJ, best governed by rules from a short moral code, we are thusasking what sorts of rule will, if generally followed, result ina sacrifice/gain ratio of less than 1. One obvious rule in thecode will command citizens to do volunteer work for altruisticgood causes at a specific regularity. For instance, humans mightbe commanded to do such work for a shift of 2 or 3 hours once aweek. However, many of the rules involved — for it is the neteffect of the whole code that matters — would be far fromobviously connected to altruism. For instance, the general ruleto obey the law of the land (save for conscientious objection toparticularly bad laws) contributes enormously to altruism. Inparticular, payment of taxes according to tax laws enables thegovernment to carry out all sorts of worthy social programs ofmost benefit to those who pay the least tax money (for theirincome is low); by the law of diminishing returns, net happinessis thereby increased. It takes constant reminding to maintainpublic awareness that despite government inefficiency, tax moneyis on average spent wisely enough to gain utility over leavingthe money in the public’s pockets.Involvement in politics is also a duty that contributesgreatly though indirectly to altruism. To illustrate: after therecent fall of communism in the Soviet bloc, an East Germancitizen visiting West Germany for the first time said somethinglike, ‘I thought that everyone here would be working very hardout of the profit motive — but they don’t work any harder than164us, not hard at all. I don’t understand where all the money iscoming from.’ Educated people know that the money comes from theefficiency of the whole system. The invisible hand of the marketdirects wealth to productive enterprises, the governmental socialsafety net prevents human talent from being wasted or destroyedby bad luck in personal circumstances, and far—sighted lawsprotect valuable objects of national heritage, such as forests,cultural treasures, and clean water and air, from being lost tothe tragedy of the commons (at least, this is the ideal — nocountry has yet achieved perfection in these matters, though somecome tolerably close in most respects). The value to altruisticcauses of such a system is enormous, and the creation andmaintenance of the system depend on the involvement of individualcitizens in democratic government. Therefore altruism is greatlyadvanced by the rule that citizens should get involved inpolitics, and, so that their involvement is productive, the rulethat citizens should educate themselves to the full degree oftheir intellectual talents.The game—theoretic reason underlying these sorts of rule isthat by dividing self—sacrificing altruistic behavior into smallchunks and getting people to do them at regular intervals over alifetime, the moral code can structure the giving and receivingof benefit by and to individuals as an iterated prisoner’sdilemma, making cooperation rational. Individuals can observeeach other’s behavior and take action against those falling toofar behind in their giving. Taken as a whole, an individual’s165giving largely ceases to be self—sacrificial — itis mereenlightened self—interest (though not all citizens areenlightened enough to regard taxpaying and so forth in this way).One gives up a lot in, e.g., some promise—keepings or refrainingsfrom theft, but over a lifetime an average human is more thancompensated for such loss by the accumulated gains, caused byone’s givings—up, of promises kept by others and by enjoyment ofproperty that others do not steal.There is a small class of humans for whom such sacrifice isnot in their long—term self—interest; namely, humans selfishenough to be untroubled by abandoning other sentients to theirown devices, and cunning enough to make their way in the worldwithout obeying moral rules, including those moral rulesunderwriting legal laws. It is insufficient to respond that thesystem would collapse if everyone scoffed at morality and law,for not everyone will scoff at morality and law. OCU nonethelessrequires such selfish yet cunning humans to obey moral rulesanyway, for it is rational to help others whether or not thepotential helper realizes that it is rational (as discussed inch. 3). The blaming of such egregiates is in itself useless, forthey tend not to change, but is an unavoidable part of the systemof blaming that keeps most citizens well—behaved.Thus appropriate, moderate rates of altruisticself—sacrifice, which nonetheless pick out, to the highesthumanly achievable degree of accuracy (i.e., less than fullaccuracy), all and only acts for which the sacrifice/gain ratio166is less than 1, are brought about by utilitarian rules commandinghumans to perform the sorts of action described above. Some suchrules are transparently altruistic, some are not transparentlyaltruistic. These rules ensure that the correct sacrifice/gainratio is achieved because they are designed to maximize presenthappiness without stealing resources from the future (e.g.,without using more than the present population’s fair share ofplanetary resources, such as by exhausting a valuable substancethat cannot be quickly regenerated, such as oil or the ozonelayer, or by extinguishing a major species of organism, such asthe blue whale or the passenger pigeon). Maximizing on asteady—state basis is favored over a pulse pattern of resourceconsumption by the law of diminishing returns. While there isthe trickle—down effect whereby losses to resource—stealing bypast generations can be outweighed by gains fromresource—processing knowledge discovered by those generations,this effect works only in the early stages of a civilization, andis negligible in the succeeding billions of years. Thus rulesfor general use will run society in a steady state, ensuring thatuniversal utility is maximized. As mentioned earlier, this is akind of success that cannot be improved upon.Thus the optimal moral code for a society would be short;the code’s rules would not pick out more than a general frameworkof behavior for a society, for any framework more detailed wouldcausally necessitate a code too complicated for the average humanto absorb. Since it is not, in the relevant sense, obligatory to167do what is not commanded by the rules of the short moral code, itis not obligatory to do the acts OCU deems supererogatory.I have briefly and informally introduced some relevantsenses of ‘obligation’ (and, mutatis mutandis, interdefinableterms such as ‘obligatory’ and ‘permissible’). To complete OCU’sreply to the objection of pervasively many obligations, let menow formally distinguish the sense of ‘obligation’ in terms ofwhich ‘supererogation’ is defined from two other senses useful toOCU and many related moral theories.One’s oritic obligation is to do an act that will maximizeutility. One’s epistemic obligation is to do an act that one’sevidence suggests will maximize utility. One’s rule obligationis to dà an act of a type that is in conformity with the rules ofthe short moral code for one’s society.An ontic obligation is an ontic criterion of right action;it says what a right act is. An ontically supererogatory act, abeneficial altruism that goes beyond one’s ontic obligation, istherefore impossible, as suggested by the intuition Dancydescribes. One’s ontic’ obligation is to maximize utility; no actcould go beyond that amount of utility. The sample condensedderivations of some common moral terms given in §1.2 werederivations of these terms taken in their ontic senses:for ‘ought to do X’ read ‘would do the mostgood by doing X’; for ‘is obligatory underany circumstances’ read ‘would uniquely dothe most good’; for ‘is obligatory all else168equal’ read ‘all else equal will do the mostgood’ . .. ; for ‘has a duty to do X’ read‘all else equal would do the most good bydoing X’.Ontic obligations are indeed pervasive; action is partitionedexhaustively into sin and duty (though including, e.g., duties torelax and enjoy).An epistemic obligation is an epistemic criterion of rightaction; it says how a right act may be known; that is, it iscorrelated with right acts, hence is evidence of their presence,yielding a reason for action. Any act picked out by such acriterion as ontically right is one’s epistemic obligation. Somehumanly knowable epistemic criteria are completely accurate; theacts they pick out are all ontic obligations (note that the agentneed not be certain in order to have knowledge; the completeaccuracy lies in the truth of knowledge, not the evidence of itsjustification). Humanly knowable completely accurate epistemiccriteria are not rules, which if simple enough to be humanlyknowable would encounter exceptions sooner or later in a complexworld such as the actual world; instead, such criteria pick outacts by means of definite descriptions (see Russell 1912, p. 52)of particular acts, e.g., ‘the casting of one’s vote forcandidate X in tomorrow’s election’. Other humanly knowableepistemic criteria are not completely accurate; not all the actsthey pick out are ontic obligations. Such criteria to be usefulin acquiring moral knowledge must pick out a good weight of ontic169obligations, hence to be simple are usually rules, expressablevia descriptions of wide application, e.g., ‘a keeping of apromise one has made’. These criteria pick out enough onticobligations of sufficient weight that the humans using thecriteria do more good, even allowing for the harm their mistakescause, than they would using other epistemic criteria, even morecomplex wholly accurate epistemic criteria, in their place in theshort moral code (a claim closely related to Sikora’s errortheory discussed in §4.1). An epistemically supererogatory act,a beneficial altruism that goes beyond one’s epistemicobligations, is therefore impossible. To be supererogatory, anact must be done intentionally; the agent must be aware that itsact will gain utility and benefit another. But this awareness asknowledge requires evidence; since the agent’s evidence in thepresent case suggests that the act in question is an onticobligation yielding a reason for action, the agent possesses anepistemic criterion of the act’s rightness. The agent may alsopossess other epistemic criteria, e.g., incompletely accuraterules, suggesting that the act is not an ontic obligation, but byhypothesis such criteria are in this case mistaken andepistemically outweighed by others of the agent’s criteria. Thusany putative epistemic supererogatory act reduces to an epistemicobligation, hence does not go beyond one’s epistemic obligations,hence is not really supererogatory.However, rule supererogatory acts are possible. Ruleobligations are all and only those acts commanded by the rules of170the moral code, which rules though based on utility may, as inthe human case, be prevented by the relevant agents’ averagelimitations of evidence or intelligence from commanding all andonly ontic obligations or even epistemic obligations. It is thuspossible for rule permissible acts to fail to maximize utility,and for the agent to possess non—rule epistemic criteria givingthe agent knowledge of this failure by way of knowledge of otherrule permissible acts gaining more utility. Rule supererogationis thus possible for agents in sufficiently subomniscientcultures, including all human cultures. Thus, in my discussionof utilitarian supererogation, by ‘supererogatory’ I mean, unlessotherwise indicated, ‘rule supererogatory’.To my explication of OCU’s version of supererogation itmight be objected that I have merely equivocated on ‘obligation’,so that the objection is merely transferred to OCU’s code ofrules, which seems not to require some obligatory acts to bedone. I reply that this construal of the objection depends onclinging dogmatically to the exact traditional intuitive sense of‘obligation’, whereas the nature of reforming definism is todepart a little from traditional, intuitive usage (and to departin order to more closely approach other intuitions relevant toethics). The reforming clarificatory moves OCU recommendsrespecting the word ‘obligation’ and its attendant practices arepart and parcel of the broad utilitarian program of discoveringand implementing new socioethical practices that approximatetraditional practices, allowing them to harness human emotions as171they are, yet that are pointed in utility—gaining directions(e.g., Hare 1981, ch. 2). Utilitarian interpretations of thesocial institutions, including reactive attitudes, of praise,blame, respect, contempt, guilt, shame, regret, remorse,repentance, punishment, rehabilitation, virtue, rights, justice,fairness, compunction, gratitude, forgiveness, esteem, and soforth, are arrived at in much the same way. One might, ofcourse, merely enlarge the objection to apply to this wholeprogram, but given the practical success of this program (forutilitarians flourish psychologically and are surely no worsethan average behaviorally) combined with its conceptualcoherence, it seems wiser to abandon the exact traditionalintuitions in question. Some of the traditional senses of moralterms turn out to be incoherent when examined closely (e.g.,Kantian ‘freedom’ and ‘desert’ — Sikora, unpublished). Moreover,some other prominent ethics besides OCU require much theoreticalinterpretation of traditional moral terms and practices —contractarian theory, for instance. If these ethics too must berejected, then we are left with little of philosophical merit.If we must choose between blind traditionalism and philosophicalanalysis as metaethical approaches, surely it is more reasonableto choose the latter.As it happens, OCU’s defence of supererogation is closelyrelated to its defence of praise and blame. Rather than treatingpraise and blame as items appropriate to agents for suitableaction on the ground of Kantian desert, OCU follows the general172utilitarian solution of treating praise and blame as usefulpractices. Recall J.J.C. Smart’s admonition:I beg the reader . . . to bear in mindSidgwick’s . . . most important distinctionbetween the utility of an action and theutility of praise or blame of it. Theneglect of this distinction is one of thecommonest causes of fallacious refutations ofact—utilitarianism. (J.J.C. Smart 1973, p.56; see also Lyons 1967, p. 27).Utilitarianism justifies praise by abandoning the traditionalview that there is a deep Kantian metaethical link between agentand praise, and taking up the view that it gains utility to makea formal normative practice of encouraging morally successfulagents on certain kinds of occasion by means of the language ofpraise; complementarily with blame. These are efficient ways ofharnessing human emotions as they are and are likely to remain.Praise functions well as positive reinforcement of choice andbehavior, yet for easy assimilation is couched in the language ofhuman reactive attitudes (see Strawson 1982). There is somequestion how we are to make metaphysical sense of these attitudesgiven that all events, including human moral choices, are eitherdetermined or random. However, that is a problem for any moraltheory, not just utilitarianism, and OCU, by locating allintrinsic value in hedonic items, renders itself independent ofthe existence of a deep Kantian basis for reactive attitudes.173OCU could accommodate such a basis if it existed, but since itseems not to, my discussion of OCU omits description of such anaccommodation.To be morally useful, praise should not necessarily beissued on every occasion on which an agent acts so as to gain ormaximize utility. First, there are very many such occasions, sothe institution of praising would be impractical if it requiredpraise on every such occasion. Second, some but relatively fewsuch occasions are positive (rule—) obligatory acts (as opposedto negative obligatory acts, such as not robbing, which are verynumerous and usually trivial to perform). Since these rulesspecify the supporting framework of a morally optimal society,positive acts commanded by the rules are particularly worthpositively reinforcing. Therefore it is plausible that praise bedemanded or at least encouraged on such occasions, or on somesignificant percentage of such occasions (positive reinforcementseldom achieves maximum return on resources expended when doledout on 100% of successful trials, and even leaving the cost ofreinforcement aside, performance is often improved by lesserpercentages of reinforcement — Timberlake and Lucas 1989, p.269 — by a sort of ‘tease’ effect). Praise is also useful nowand then to reinforce negative action, but far more rarely —mostly on occasions of resistance to unusually strong temptation.On the remaining occasions, on which utility is gained ormaximized but the act is not obligatory, it is plausible thatpraise is permissible but not required, nor encouraged so174strongly as in similar cases where rules are obeyed by positiveaction. Praise is permitted partly to avoid cognitive dissonancein the potential praiser, since the act is known to gain utility,which is posited to be the only intrinsic value, and partly topromote the gaining of a little utility beyond what the rulesspecify, for a hedon more is a hedon more. Praise is notrequired or strongly encouraged, for fear of creating socialexpectations of beneficence beyond duty, which in turn wouldcause pointless guilt in the majority not doing such actions.Such guilt would be pointless because it could not usefullychange their behavior; by hypothesis most humans could not liveup to standards significantly higher than those of the shortmoral code.Utility—promoting acts beyond what is called for by therules are therefore praiseworthy, hence eligible forsupererogatory performance.To this account of supererogation it might be objected thatit is actually harmful to praise acts beyond the agent’s ruleobligations, since the praise will encourage them to try to dowhat in the long run they cannot. However, agents typicallyacting supererogatorily by so acting demonstrate a supernormalquantity of moral perception and strength of will that makes itmore likely than not that they could keep up good deeds at a paceabove that prescribed in the short moral code for their society.As well, it is part of the convention that praise of asupererogatory act is praise of that particular act, not of the175practice comprehending the act. Agents typically decliningsupererogation by so doing demonstrate a suboptimality of moralperception and strength of will that makes it more likely thannot that they could not have kept up good deeds at a pace abovethat prescribed in the short moral code for their society. Theselikelihoods rationally warrant both our not blaming those whodecline supererogation, and our praising those who performsupererogation. Thus the acts picked out are both praiseworthyand not obligatory, in that the acts which are done tend to bepraiseworthy, and the other acts which are not done tend not tobe blameworthy; the potential acts though all falling under thesame act—description, are separated by which are done and whichare not. Therefore the objection fails.It would take detailed empirical studies to prove that thistype of praising function would optimally harness human emotions,but it is at least plausible on present knowledge — plausibleenough to ground OCU’s version of supererogation. Whileempirically unproven, this praising function has not beenempirically refuted, is supported by some empirical evidence, andis metaphysically uncontroversial hence metaethically sound.To Dancy’s concern that any act better supported by reasonsfor action than its alternatives must be obligatory, OCUtherefore replies that while this is true respecting onticobligation and epistemic obligation, it is false respecting ruleobligation. Rule obligations arise not only from reasons butalso from the complex web of motivating institutions described.176A reason for an action, even a conclusive reason, is necessarybut insufficient for the act’s being a rule obligation. A reasonfor action may provide some motivation, but not always enough toget the potential act chosen for actualization over competingacts. The complex utilitarian system of rules, praise, blame,etc., is designed to shunt available motivation around so that inthe long run the acts with the most compelling reasons tend toget chosen and done (thus acts are effectively chosen betweeneven when they are not colocated alternatives, but distant inspace—time). Acts with reasons to be done but little prospect ofobtaining sufficient motivation to get done are not made ruleobligatory, for there would be no advantage to doing so. Theintuition to be opposed to the intuition Dancy mentions is thatought implies can. Without motivation one cannot will the act;hence without motivation it is not the case that one ought towill the act.Thus the first half of the second criticism of OCU’ssupererogation, the justification objection, has been answered.The second aspect of OCU’s program of justification ofsupererogation is the positive aspect of showing the value ofnon—obligatory well—doing. In light of the explanations alreadygiven, this is easily shown: it has been admitted that theshortness of the code of rules prevents it from making all onticor epistemic obligations into rule obligations, and surely someof these left—over obligatory acts are both ontic epistemicobligations, hence utility—gaining and justifiably believed so by177the relevant agents. These agents then have-the option of doingsupererogatorily any of the many such acts that are permitted bythe rules, and that gain more utility than other permissibleacts. Since these acts gain utility, they are valuable.Some acts that are utility—gaining and permissible but notobligatory by ordinary rules are nonetheless reasonable responsesto emergencies, hence are rule—obligatory by the top rule‘maximize utility’. To this view it might be objected that ifthe top rule covers emergencies, then it covers supererogationtoo; just saying that it doesn’t cover supererogation doesn’tmake it not cover supererogation. The covering or lack of itissues from the rule itself. The rule says to maximize utility,and the supererogatory act is known to produce more utility thanits nonsupererogatory alternative. To this objection OCTJ repliesthat the human tendencies to rationalize, etc., discussedearlier, prevent the rule ‘maximize utility’ from justifiablybeing applied directly in cases of supererogation, though therule does apply indirectly, e.g., in the background rationale forrestricting its direct application. The agents’ motivation forfollowing this restriction is guided and amplified by praise andblame. This guidance is done by means of learned criteria andcointexts for rule—following. Humans have the causal capacity tolearn criteria and contexts for rule—following, as shown by thedifficulty humans have using new rules until they have learntcriteria and contexts, and by the unfortunate ease with which theprocess can be subverted by irrational criteria and contexts,178producing an immoral double standard. Human legal systems, withtheir close attention to often—obsolete precedents, offer manyexamples of both phenomena. Thus OCU’s solution to the objectionhas been shown to be rationally justified, agent motivatable, andcausally possible.It. might also be objected that in most cases where an agentby declining to act supererogatorily declines what seems likeachievable utility, the utility was not in fact achievable; theshort—term gain would be accompanied by outweighing long—termloss. The option of such agents doing such acts with no suchassociated downside is a mirage. Thus no supererogatory act waspossible; there was no available act which would have gainedutility, hence nothing valuable beyond the rules.Analogously, for an alcoholic, the act of having one drinkmay seem more beneficial than the act of having no drink, but infact is a mirage, for the act cannot be done independently.Having had one drink, the alcoholic would end up having manydrinks, and the world would suffer a net loss of utility. Theoption of having just one drink was not open.To this objection it can be replied that while in many casesno utility could have been gained, in fewer but still many casessignificant utility could have been gained. Some achievableutility is lost by OCU. It sometimes happens that an able,strong—willed human knows that a utility—gaining action wouldbenefit others, yet not benefit them enough to constitute anemergency. The act is permissible but is commanded by no rule of179the moral code, hence is supererogatory. On such occasions theagent would have sufficient reason for action, and would know it,yet might or might not find enough motivation to act onticallyrightly. Similarly, a champion high—jumper, though well able tojump a certain height, may or may not do so on a given attempt.It is not the fault of the coach when the jumper fails; we maysuppose that no other coaching method would improve the averageperformance of the jumper. Likewise, it is not a deficiency ofOCTJ when achievable utility is lost on the occasions describedabove. It is, from the point of view of the moral code, simplebad luck. No further instruction issued by the moral code wouldimprove the situation on balance. One can only hope that theagent will independently recognize the facts and act onticallyrightly. On some such occasions the agent will have sufficientmotivation; the agent will thus have a reason and motivation, butno obligation. On other occasions the agent will not havesufficient motivation; the utility thus lost is achievable inthat though determined not to actually exist, it would haveresulted from an act that the agent was free to perform.Thus the second half of the second criticism of OCU’ssupererogation, the justification objection, has been answered;since both its halves have been answered, the objection as awhole fails.3. The sacrifice/gain ratio180It might be objected that it is impossible that theresulting set of rules could ensure that the correctsacrifice/gain ratio is on average achieved by everybody, simplybecause there is no such thing as a sacrifice/gain ratio that iscorrect for everybody (Sikora, personal communication). This isthe third criticism of OCU’s form of supererogation; call it thesacrifice/gain ratio objection.Non—cognitivist versions of CU are hard—pressed to answerthis objection, for, as will be discussed in more detail in §6.1,non—cognitivist utilitarianisms cannot consistently claim (andtend not to claim) that people are mistaken in their assessmentsof the comparative value of different experiences containingdiffering durations and intensities of pleaure and pain, becauseaccording to non—cognitivism there are no objective values outthere in the world for people to be mistaken about, and noobjective reasons for preferring one thing to another. If ethicsis thus a branch of aesthetics, then claiming that Jones ismistaken about how much happiness it is rational to personallyforgo to bring X amount of happiness to Smith is like claimingthat Jones is mistaken in preferring to look at black velvetclown paintings rather than Picassos. Mistakes are impossible inthis sort of judgment.One sort of non—cognitivist utilitarianism might claim toavoid this objection. Suppose that one held, first, that to saythat I ought to do a given thing is, as a reforming definism, tosay that it is in accord with the morality that I would choose181for my society on the hypothesis that I were calm andepistemically rational, and, second, that the moral code onewould choose under these hypothetical circumstances is some formof rule utilitarian short moral code. It is stipulated of theepistemically rational person that its choices will not beaffected by logical errors or by failure to make the logicalinferences relevant to (i.e., capable of influencing) them; itwill have all the conceptual apparatus relevant to these choices;and it will have the relevant factual information as well as avivid idea of the relevant consequences of its choice(paraphrasing Sikora 1978, p. 148—50). By this version ofnon—cognitivist utilitarianism, even if pleasure and pain werecompletely objectively measurable, some people would have moregenerous sacrifice/gain ratios than others. Proponents of thistheory might hold that since each person has a subjectivelyrational sacrifice/gain ratio, a ratio they personally want touse to guide their own behavior, the utilitarian set of rulescould ensure that the correct sacrifice/gain ratio is, to anacceptable approximation, achieved by everybody. The set ofrules would do this by specifying some minimum sacrifice/gainratio, low enough that almost anyone will want to achieve it (andwith modest effort will be able to), as obligatory, andpermitting individuals to sacrifice more if they wish bysupererogation. However, this sort of non—cognitivistutilitarianism has the same flaw as so many other contemporarymetaethics, namely that it is unwittingly a form of strong moral182objectivisin, a cognitivism. It assumes that certain facts yieldreasons for action, in that if I want x to be the case, then Ihave a reason for acting to bring about x. Consider thefollowing facts:1. Jones wants x to occur2. I am Jones3. That person over there is JonesSomehow, according to the sort of utilitarianism underdiscussion, facts 1 and 2 jointly make rational my choice ofbringing about x, in that they provide me with a reasonfor action, such as the act of pursuing a certain sacrifice/gainratio. But to call something a reason for action is to suggestthat it provides some consideration beyond the mere causal powerof the want to generate the act, namely some sort of conceptualfittingness between want and act. If one allows conceptualfittingness as a coherent notion, then one must also admit themetaphysical adequacy of altruistic reasons for action, derivedfrom, e.g., facts 1 and 3 above, resulting in Jones’ wants givingme a reason to act to bring about the things Jones wants. Theseobjective, interpersonal values might then turn out to yield, asOCU claims, the same sacrifice/gain ratio for everyone. If theabove sort of ‘non—cognitivist’ utilitarianism wishes to rejectits unwitting cognitivism, then it must stop calling the agent’swants reasons for action, hence must give up the claim that thereare differences in individuals’ sacrifice/gain ratios, for it hasonly shown that there are differences in what individuals want183their sacrifice/gain ratios to be, not that there are differencesin what individuals have reason to make their sacrifice/gainratios.Thus this sort of ‘non—cognitivist’ utilitarianism cangenerate no coherent objection to OCU’s claim that there is asacrifice/gain ratio that is correct for everyone. Moreover, OCUcan offer a coherent explanation why this claim is true. As acognitivist metaethic, OCU can and does claim that when peoplediffer in their opinions about how much happiness one shouldsacrifice to bring about a given quantity of happiness forsomeone else, at least one of the people differing must bemistaken. Humans’ asking themselves what sacrifice/gain ratiowould be epistemically rational for them would be like humans’asking themselves what square root of 25 would be epistemicallyrational for them.Thus I claim that a non—cognitivist theory can allowinterpersonal comparisons of many aspects of pleasure and pain,e.g., their duration, their intensity, and their status asqualia, but cannot allow interpersonal comparisons of theirobjective value (entified or not), for according tonon—cognitivism they have none to compare. Thus undernon—cognitivism, it is not the case that pleasure and paincontain any concrete intrinsic value entities, or are rationallyobjectively preferable to one another. Under non—cognitivismpleasure and pain have subjective value, and these subjectivevalues can be interpersonally compared, but that is a different184matter. Thus on a non—cognitivist theory one can say, ‘I likePicassos twice as much as you like them’, but not ‘Picassos aretwice as good you think they are’ or ‘quantity x of pleasure istwice as good as you think it is’.A moral cognitivist, on the other hand, can say, ‘quantity xof pleasure is twice as good as you think it is’ with the fullweight of objective rationality.I make use of the epistemic/ontic distinction here. I usesome of the same preference tests that a non—cognitivist woulduse, though these tests must be reinterpreted to accommodateobjectivist phenomenology and metaphysics. I claim that thereare concrete, objective value existents (capable of groundingontic obligations), and objective value experiences (capable ofgrounding epistemic obligations); value experiences involveentertainments of the concept of objective value. Objectivevalue experiences, as opposed to aesthetic value experiences, arethe raw material of tests for ontic values; though not alwaysveridical, objective value experiences are evidence for theexistence and nature of objective value entities, as discussed inch. 3.I leave it open that other tests of objective value mightexist over and above those adaptable from non—cognitivists, but Ido not at present know of any. The two sorts of test,cognitivist and non—cognitivist, are closely related in that eachcan be derived from the other by suitable reinterpretation, butneither has any important nonethical priority over the other —185they have both been around for a long time.An intuitive fear about OCU’s sort of solution to theproblem of supererogation is that some unlucky person is going toget stuck in life circumstances that make it clear that byundergoing a vast amount of unhappiness they can create aslightly larger amount of happiness for others. If thesacrifice/gain ratio of anything less than 1 is correct, thensurely OCU must tell the unlucky person that they must make thesacrifice and have a most unhappy life. However, this fearignores the fact that no rule demanding such an enormoussacrifice from a single human would be obeyed very often, becausethe slight net gain in happiness would seldom be able to causallyovercome the vast fear of pain that most humans do and shouldhave as a means of detecting and minimizing bodily injury.Therefore such a rule would not long endure, hence would not beincluded in the set of rules endorsed by OCU for humans (or, forthat matter, any other causally possible species of intelligentsentient). A chief purpose of describing in some detail thekinds of rules that would govern supererogation and altruismunder OCU was to show how an appropriate set could achieve thecorrect ratio for society asa whole yet distribute the burdenfairly evenly. No intrinsic value is attached to distributingthe burden evenly — it is merely a means to getting humans totake up the burden at all.Thus OCU can maintain that since pleasure and only pleasurehas intrinsic goodness, one should seek to bring it about without186intrinsic regard to who gets it. Yet OCU can also maintain thatfor contingent causal reasons one should have some extrinsicregard for who gets it.OCU does not maintain that every human should make exactlythe same sorts and quantities of sacrifices. The average humanshould follow the basic rules as set out in the short moral code,but able, strong—willed persons are permitted as mentioned tosacrifice more, supererogatorily, and those disabled in mind orbody are permitted to sacrifice less, since they often have lesshappiness to make sacrifices from. This does not involve privatemoralities; it involves a single public morality that recognizesdifferent social roles based upon the different kinds of mindsand bodies that humans have. Other relevant social roles arethose of male, female, doctor, lawyer, teacher, soldier, adult,child, and so on. A given human may occupy many of these roleseither serially or, in some cases, simultaneously. Minordifferences in the lower—level rights and duties of occupiers ofthese roles have rational bases, yield similar happiness levels,and do not supersede the equal dignity and equal higher—levelrights and duties of every human. It would be a lamentably rigidand impractical moral code that did not have some such flexiblefeatures.An essential property of such a code’s recognizing differentsocial roles is that all roles are in principle open to anyone.We are all children early in life, and merely by living longenough we can later enjoy the rights of adulthood. We may not be187born male or female, but can in principle have sex—changetreatments giving us access to a new sexuality. We can all go toschool and learn to become doctors or lawyers, etc. Anyone canapply to join the military. Admittedly, the facts of one’s mindor body may prevent one from achieving a given social role; adunce will fail the law exams, and a paraplegic will fail themilitary physical exam. However, analogously to thecompatibilism by which one is free to do x, though determined todo otherwise, I propose a compatibilism by which one is free topursue any of the social roles of epistemic RU, though onoccasion determined by contingent personal characteristics not toachieve a given social role.Able, strong—willed people are permitted to declinesupererogatory acts partly because it is so hard for a human,even a strong—willed one, to be sure that they are in a situationin which an uncalled—for, genuinely gainful, hencesupererogatory, sacrificial act is possible. Even when ablepeople are intelligent and well—educated enough to be rightlysure that large gains of utility would result, they may not bestrong—willed enough to overcome their strong, generally usefulfear of pain.Thus the third criticism of OCU’s form of supererogation,the sacrifice/gain ratio objection, fails.Thus all three criticisms of OCU’s form of supererogationhave been answered, hence the fourth objection to OCU is refuted.188Chapter SixTWO METAPHYSICAL BACKGROUND ISSUES1. Goodness and pluralismThe fifth objection to OCU is that moral pluralism might betrue, but that OCU is a form of moral monism, hence isimplausible to the degree that moral pluralism is a live option.OCU takes no position on the issue whether a plurality ofkinds of thing are good. While there is no metaphysicalnecessity to there being only one sort of intrinsic good, no morethan that there should be only one sort of color, in the cases ofputative goods other than pleasures, parsimoniousentity—reduction seems in order. Few reliable intuitions wouldbe disturbed by regarding such objects, e.g., knowledge, love,freedom, and honor, as having only extrinsic value, so long ascertain relevantly related other entities, pleasures, areregarded as having intrinsic value. OCU is a form of hedonism(taken as the moral doctrine that only pleasure is objectivelyintrinsically good, not as one of some other doctrines oftenconfused with moral hedonism — e.g., the psychological doctrinethat all desire is for pleasure, which Nowell—Smith, 1954, p.135, calls ‘hedonism’). However, it is not obvious that one189cannot be both a hedonist and a pluralist, and it would be unwiseto unnecessarily commit OCU to a premise that is not obvious,particularly when many philosophers have endorsed that premise’scontrary (see Goldstein 1985, p. 49). I shall therefore show inthis subsection how one might be a pluralistic hedonist, thusshowing that the objection that moral pluralism might be true,and is inconsistent with OCU, is unfounded.The question whether a hedonist is a monist or a pluralistis the question whether all pleasures are pleasant equally and infundamentally the same way, so that pleasure is a singledeterminable property in its own right (though also, according toOCU, at a different level in the relational hierarchy the uniquedeterminate of the determinable goodness); or whether theentities called ‘pleasures’ in English are not all pleasant inthe same way, but fall under two or more sante—level pleasuredeterminables, each also a determinate of the determinablegoodness. For example, Ryle claims:that ‘pleasure’ can be used to signify atleast two quite different things.(1) There is the sense in which it iscommonly replaced by the verbs ‘enjoy’ and‘like’. To say that a person has beenenjoying digging is not to say that he hasbeen both digging and doing or experiencingsomething else as a concomitant or effect ofthe digging; it is to say that he dug with190his whole heart in his task, i.e., that hedug, wanting to dig and not wanting to doanything else (or nothing) instead.His digging was his pleasure.(2) There is the sense of ‘pleasure’ inwhich it is commonly replaced by such wordsas ‘delight’, ‘transport’, ‘rapture’,‘exultation’, and ‘joy’. These are names ofmoods. . . . Connected with such moods,there exist certain feelings which arecommonly described as ‘thrills of pleasure’,‘glows of pleasure’ and so forth. (1949, p.108—09)These two types of pleasure might be tagged, respectively,‘desire—satisfaction’ and ‘joy’. Many ethicists are persuadedthat certain kinds of experience such as these, all loosely knownas ‘pleasures’, are intrinsically good, and are unlikely to bedissuaded by learning that no one pleasure property underlies allthese experiences; for example, Baylis 1967, p. 444. Occam’sRazor favors the collapsing of two entities into one whereverpossible in a theory, and many hedonists find that theirintuitions are not seriously strained by the collapsing of, e.g.,desire—satisfaction and joy, into one entity, pleasure. Forinstance, one might adopt Nowell—Smith’s claim that “to desiresomething is to expect it to be pleasant” (1954, p. 137), and addthat the satisfaction of a desire is the obtaining of the191expected pleasure when the desired object is achieved. However,some hedonists may not find this or any comparable moveplausible, and yet may still view both hedonic entities,desire—satisfaction and joy, as objectively intrinsically good.On behalf of pluralistic hedonism, it might be objected tomonistic hedonism that our pleasures are far too disparate tofall under the same pleasure determinable, that we take pleasurein far too many different things, that we like the feel of fartoo many different experiences. This objection stems from amisapprehension of the plausible premise that pleasures are, orare relevantly analogous to, objects capable of intentionality.A paradigmatic object capable of intentionality is a sentence,which can be about some external existent (‘The cat is on themat’), or about nothing of this nature (‘Wow!’); likewise’apleasure can be about some external existent (‘I find gardeningpleasant’), or about nothing of this nature (‘I’m in a pleasantmood today’). The objects of pleasure may indeed be radicallydifferent. However, this does not entail that the pleasuresthemselves are radically different. Just as the nature ofsentencehood is unified and independent of the nature of theentities, if any, referred to in the sentence, the feel ofpleasure is the same no matter what, if anything, occasions thepleasure (that is, the feel of pleasure is the same takingpleasure in abstraction from its intensities). On this accountof pleasure, it is not the case that an experience is pleasantiff you like the way it feels (which would multiply pleasures192indefinitely— digging, etc., would all be distinct pleasures);rather, pleasure is a liked feeling that is sometimes occasionedby an experience (the feeling liked being the same for anyexperience occasioning it). Or alternatively, pleasure is likingthe way you feel, whether that feeling includes or is occasionedby an experience or not (the liking is the unchanging pleasure).Goldstein 1985, p. 52, claims that various kinds of pleasuremight be radically different but nonetheless encompassed by thesame defining account. I agree, but this claim does not entailthat pleasures are similar enough to fall under the same pleasuredeterminable. I am inclined to think that they are similarenough, that the pleasures have something profoundly in common,hence that monistic hedonism is more plausible than pluralistichedonism. VIf a unified account of pleasure cannot be made to work,then, if one wishes to remain a hedonist, one must be apluralistic hedonist. How might hedonistic pluralism be objectedto?It might be objected that if two or more independent hedonicentities are each objectively intrinsically good, then it willsometimes be impossible to rationally calculate what to do, sincesome of the several hedonic goods will be incommensurate. Thisimpossibility would be inconsistent with OCTJ and many other formsof utilitarianism, according to which it is always possible inprinciple to rationally calculate what to do. This objectiontakes several variations depending how incommensurability is193defined.One definition of incommensurability is given by Raz: “A andB are incommensurate if it is neither true that one is betterthan the other nor that they are of equal value” (1986, P. 117).Such ‘ordinal incommensurability’ (adapting Hall’s term — 1967,p. 38) does permit rational action of a weak sort: “rationalaction is action for (what the agent takes to be) an undefeatedreason. It is not necessarily action for a reason which defeatsall others” (Raz 1986, p. 132). However, ordinalincommensurability does preclude rational choice of the strong,positive sort OCU claims to provide. OCU, like other strongmoral objectivisms, claims not merely that the actions itprescribes can’t be proven to be foolish choices, it claims thatthese actions can be proven to be wise choices, that theseactions can be given reasons which defeat all others. Eventhe arbitrary tie—breaker choice between options of equal highestvalue by adherents of OCU (or of any utilitarian version ofentifying moral objectivism) is strongly, positively rational inthat the agent is assured that all options other than theequivaluable set yield less value, and that within theequivaluable set the choice is merely between different tokens ofthe same type (goodness embedded in a pleasure); no wild card isin play, so one is rationally guaranteed that no importantdifference lurks between the options. Ordinally incommensurateoptions are unsettling precisely because we know that importantdifferences may lurk between them, yet have no rational way of194assessing these differences.Ordinal incommensurability arises frequently in epistemiccontexts. A human, even a convinced utilitarian, may be forcedto choose between a friendship and a better—paying job, yet dueto limited evidence have no way of rationally preferring oneoption to the other. This situation passes the transitivity testof incommensurability (Raz 1986, p. 120—21); although reason isindifferent between the two options, it does not rank themequals. Increasing the job’s salary would make the job betterthan it had been, but would not by transitivity thereby make thenew option superior to its predecessor’s rival in the eyes of thehuman agent. The agent choosing simply has inadequate evidenceto yield a rational preference between friendship and money.However, epistemic ordinal incommensurability does notentail ontic ordinal incommensurability. The unavailability ofevidence permitting, e.g., friendship and money to be humanlycompared may be merely contingent. OCU holds that all goods arein principle ordinally commensurate, hence that an omniscientagent could choose rationally from any set of options. This lineassumes that people may be mistaken in their assessments of avalue; to this assumption Raz objects that,the very assumption of a possible gapbetween people’s considered judgments of thecomparative value of options and their realvalues presupposes that values have areality which is independent of people’s195perceptions of their lives and the valueof options which are in principle open tothem. This presupposition, the Platonicassumption we can call it, is to berejected. (1986, p. 132)However, as shown earlier in this thesis, there is no compellingreason to reject real moral values, and much to be metaethicallygained from accepting them. OCU is therefore unharmed by thisobjection derived from the first definition ofincommensurability.A second type of incommensurability is ‘linearincommensurability’ (again adapting from Hall 1967): ‘A and B areincommensurate if they lack an appropriate linear dimensionpermitting numeric quantification of value’. This definitiongrounds several objections to OCU, including an objection toOCU’s version of hedonic pluralism. First, it might be objectedto OCU (and many other utilitarianisms) that they speak freely ofnumeric quantities of pleasure, e.g., in locutions like ‘supposethat Jones obtains 100 hedons of pleasure from an hour’s walk inthe woods’, implying that pleasures are commensurate in thissecond sense. Yet, goes the objection, pleasures lack anappropriate linear dimension by which their value might benumerically quantified, henceare incommensurate in this secondsense; therefore OCU must be false. This objection arises fromthe fact that though we may casually say things like ‘it feelstwice as hot today as it did yesterday’, such talk is literal196nonsense since heat sensations and their ilk have no intrinsicquantitative values. The correspondence of heat sensations tonumeric temperatures such as those of the centigrade orfahrenheit scales is merely conventional; unlike thermometers,heat sensations seem to have no appropriate intrinsic lineardimension. Locutions such as- ‘it’s twice as pleasant to watchplayoff hockey as it is to watch regular—season hockey’ have asimilar form to such nonsense locutions, which suggests that asimilar confusion may underlie such pleasure locutions, via thelack of an appropriate intrinsic linear dimension to pleasure.For purposes of argument, let us grant that the only lineardimension of pleasure available even as a candidate forappropriateness is the temporal dimension; then we shall seewhether the objection can nonetheless be answered. Under theassumption granted, every pleasure has this temporal dimension,for every pleasant event takes place in time, hence is temporallyextended.Yet hedons (= utils, etc.), interpreted as measurements ofpleasure, measure more than the mere temporal length of apleasure; we want to be able to say that Jones obtained 100hedons from an hour’s walk in the woods with Smith, but thatSmith obtained 200 from the same walk, so that Smith’s mentalstates were on average twice as pleasant as Jones’s during eachtime—slice of the walk. The extra thing being measured is theintensity of the pleasure; twice as intense is twice as good.Hedons are numerically quantitative units, so presuppose a linear197dimension (a number line) in whatever they measure, yet byhypothesis pleasure has no potentially appropriate intrinsiclinear dimension beyond time. Since the extra thing beingmeasured by hedons therefore (goes the objection) both is and isnot an intrinsic linear dimension, the concept of a hedoncollapses in self—contradiction. And without hedons, OCU lacks away of comparing the quantities of different pleasures.To help answer this objection from linearincommensurability, let us examine in its light a popularutilitarian method of accomplishing the numeric quantification ofpleasure (e.g., Hall 1967, p. 44—45). According to this method,one pleasant mental state is correctly measured as containing,e.g., twice the amount of pleasure per time—unit as anotherpleasant mental state if, all else equal, experienced agents areindifferent between obtaining one time—unit of the formerpleasant mental state and two time—units of the latter. Thecombination of this way of quantifying the intensity of pleasureand the normal quantification of time—segments allows allpleasures to be quantitatively compared; thus hedonisticutilitarians can make rational choices between options.OCU’s version of this sample method of comparison might becriticized on the ground that it is liable to the error of beinga form of ideal observer theory, in that it tries to substitute,for an objective intrinsic consideration, the subjectiveresponse, to the relevant entities, of some type of observer.OCU responds that while this comparison method must, to be198acceptable, show that the measures produced are correct becausethey correspond to some unstated underlying objectiveconsideration between the relevant entities, this correspondencearguably obtains. However, positing such an underlying onticfeature of pleasure exposes OCU’s version of this comparisonmethod to the objection just described that pleasures arelinearly incommensurate.To escape this objection, OCU must hold either that (1)hedons despite their superficial numeric character do not measurenumeric quantities, or (2) that numeric quantities may measuresomething other than a linear dimension. To see which premiseOCU should embrace, let’s examine what OCU requires from hedonsin pursuit of rational choice.What OCU requires is a rational value ordering of themembers of the power set of all possible tokens of simple hedonicitems (hence including repetitions of some such items), so thatin any possible ethical choice between two sets of consequences,one set will be objectively more valuable than the other, or thetwo will be equally valuable. Since the set of all possibletokens of simple hedonic items may be infinitely large, thisexpression of OCU’s requirements is logically problematic, forstandard logic applies only to finite systems. But let’s assumethat this problem can be overcome; it is common to allconsequentialisms, hence can be overcome if, as they do,consequences make a practical moral difference. Moreover, logicapplies to the universe of physical objects, which may well be199infinitely large; if we do discover it to be infinitely large,then physical science will liable to the same objection, andsurely we are not prepared to give up physical science justbecause of the discovery of a lot more stuff pretty much the sameas the stuff physical science already explains so well.Given such a value ordering, moral agents may rationallychoose a course of action in any possible moral dilemma. Hedonsfunction to express such a value ordering in such a way thatadding and subtracting numbers of hedons, in correspondence withthe addition or subtraction from an option of the hedonic eventsassigned these hedonic numbers, preserves the accuracy of thevalue ordering. For example, assume the following options andvalue orderings (event ‘2X’ is 2 tokens of event X; event ‘0’ isa mental event that is neither pleasant nor painful):option 1 = events A, B, and Coption 2 = events A, B and Doption 3 = events B and Doption 4 = events A, B, and 2C0 < A < B = 2A < C = 3B< D = 2CAssume that each of these events is one time—unit’s worth of therelevant event type. Assume also that these events do not changevalues when in combination with other events. This assumptionstill allows the possibility of valuable organic wholes; however,200any such organic wholes would be listed separately. Thus, e.g.,the value of option 3 is captured entirely in the values ofevents B and D, so that the ‘event’ B&D has no value (if it did,option 3 would be described as events B, D, and B&D). Option 2is thus of greater value than options 1 or 3, and of the samevalue as option 4. What OCtJ requires is an assignment of hedonicvalue numbers to events A—D such that for any two options w andz, the sum of the hedonic value numbers of w’s events is lessthan the sum of the hedonic value numbers of z’s events if f w z’. But giventhe assumptions made about value orders logically ‘behavingthemselves’ via association, transitivity and the like, it iseasy to construct such an assignment. Simply assign some hedonicevent X of 1 (arbitrary) time—unit’s duration the value of 1hedon, and assign hedonic value numbers to all other hedonicevent types in inverse proportion to the length of thetime—segment they need to equal X in value. This assignmentyields the comparative worth of a standard temporal length of anyhedonic event type, in such a way as to preserve the valueorderings of all possible options through all possible additionsor subtractions of events and their associated hedonic numbers.And as explained in reply to the objection from ordinalincommensurability, the mild logical ill behavior of humanlyperceived values is, in the context of hedonism, an epistemicphenomenon, consistent with complete logical well—behavedness inthe underlying ontic phenomena of objective values; moreover,201this assumption of complete logical well—behavedness is rational(see Hall 1967, p. 45—51). OCU thus gets what it needs fromhedonic numbers without assuming that pleasures have a secondlinear dimension of the sort objected to.In its escape from the objection from linearincommensurability, OCU thus holds that regarding the comparativeworths of hedonic events of the same temporal length (in effect,the comparative worths of pleasure intensities), hedons despitetheir superficial numeric character do not measure numericquantities. Instead, they use numbers to express orderings.These order expressions are true because they correspond to realmoral value orderings in pleasures. The full numeric potentialof hedons is used only regarding the temporal dimension. This isnot to say that, regarding intensity, hedons do not measurequantities, merely orders; as C. Stuinpf says, “increase anddiminution are names for quantitative changes” (see Husserl 1970,p. 440). Since intensities may increase and diminish,intensities are quantitative entities. It seems natural toexpress this as intensities having non—numeric quantities, andline segments having numeric quantities. However, one mightalternatively express the distinction as intensities’ quantitiesbeing merely ordinally numeric, and line segments’ quantitiesbeing linearly numeric; either way, there is a clear boundarywhich makes no trouble for hedons. Note that if the boundary didmake trouble for hedons, it would also make trouble for somestandard theoretical entities from physical science, such as202measures of heat—energy (Hall 1967, P. 40).Thus so long as quantitative attributions of value tohedonic consequences are interpreted merely as expressions ofvalue equi—ordering between certain temporal quantities ofvarious hedonic items, these quantitative attributions will notentail that the valuable hedonic items have, in addition to anintrinsic value ordering, an intrinsic linear dimension presenteven in a time—slice of the hedonic item, as is the case in atime—slice of a yardstick. OCU (and many other utilitarianisms)may therefore avoid the objection that hedonic items have nointrinsic linear dimension. Confirming evidence comes from thefact that in practice we have no difficulty making at least somesorts of quantitative comparisons between hedonic consequences;e.g., torturing a person to death is at least twice as bad asstepping on their toe. Exact quantitative comparisons are oftendifficult or impossible in practice, but this problem is commonto all the quantitative sciences.A final objection to this method of quantitatively comparingpleasures is that if, as deemed possible earlier, pleasures areencompassed not by a single property of pleasure but by aplurality of properties such. as desire—satisfaction and joy,hedons could not possibly in any way measure amounts orintensities of pleasure, because there is no unified pleasureentity to be measured. OCU replies that what hedons rank arenot, strictly speaking, pleasures, but their goodnesses.Goodness is instantiated in time, so may be temporally203quantitatively measured, and is instantiated in variousintensities, so may be ordinally ranked as explained above. Andgoodness is part of each hedonic entity whether or not thesepleasures fall under a single pleasure property. For each kindof pleasure, one or many, a certain intensity of pleasure isnecessarily accompanied by a certain intensity of goodness; asthe intensity of the pleasure increases, so proportionately doesthe intensity of the goodness. As explained in ch. 2, OCU leavesit an open question whether the goodness of a pleasure is foundas a part within a whole, or merely by entailment supervening onthe pleasure; either way, any indexical location in whichpleasure is instantiated necessarily also instantiates goodness.And these goods are what hedons measure; it is always rational tochoose the most goodness available — the most temporal length andthe most intensity. This system of measurement would work evenif other mental items were good besides hedonic items, though Icannot imagine what such things might be (in agreement with Blake1967, p. 437). Since, as Sidgwick notes, “it is reasonable toprefer pleasures in proportion to their intensity” (1874, p.127), hedonic judgments can be made by agents cognizant of theirawarenesses of pleasure but not of goodness, as is sometimes thecase. The form of hedonic pluralism thus produced would bebenign because it retains the traditional strength of hedonicmonism: “one of the main attractions of traditional hedonism wasits attempt to reduce all value to a single source” (Goldstein1985, p. 50), while allowing, as some find phenomenologically204necessary, a plurality of pleasure properties.A third definition of incommensurability is used inmathematics: ‘A and B are incommensurate quantities if theirratio cannot be expressed as an ordinary fraction’. Any rationalnumber is incommensurate in this third sense with any irrationalnumber (call this ‘mathematic incommensurability’); for instance,the numbers 1 and root 2 are mathematically incommensurate.However, mathematic incommensurability poses no significantproblem for any type of utilitarianism, for it is not a claimthat the relevant quantities do not exist, merely the claim thatthey cannot be expressed in a certain way. The numbers 1 androot 2 pick out certain points in a linear dimension thatundeniably exist, hence measure line segment lengths (Neserve etal 1989, p. 114). These line segments exist and form certainproportions of each other whether or not these proportions areexpressable as ordinary fractions. We can easily use othersymbols to name such proportions (Meserve et al 1989, p. 125).Likewise, claims utilitarianism, certain temporal segments ofintrinsic goods exist and form certain proportions of each otherwhether or not these proportions are expressible as ordinaryfractions. There is even less difficulty for hedonic quantitiesexpressing ranked intensities, for no underlying numericquantities are posited.A fourth definition of incommensurability is given byKenshur:to prove incommensurability, one has,205ultimately, to prove the nonexistence of anystandpoint outside individual theories orconceptual systems from which they can bemeasured against one another. (1984, p. 376)Such incommensurability (call it ‘theoretic incommensurability’)could be asserted of the moral theories that desire—satisfactionis good and that joy is good (and so forth for whatever goods thepluralist posits). However, it need not be asserted; thepluralist might be a rigorous rational objectivist about basicepistemology and metaphysics, and claim only to havedistinguished between two (or more) sorts of mental state easilycomprehended within one overall world—view. If, on the otherhand, the pluralist opts for theoretic incommensurabilityregarding moral goods, they are committed within that sphere to adeep Feyerabendian relativism that appears to contradict itself(Kenshur 1984, p. 378), being yet another incarnation of naiveCartesian skepticism, the view that anything that can be doubtedshould be doubted. One might attempt an intermediate positionaccording to which an overall world—view is fine, but not anoverall moral view, so that every moral agent’s assignment ofvalue is equally well justified in the weak, negative sense thatno justification is possible. But such an intermediate viewamounts to the reduction of ethics to a branch of aesthetics,which view was rejected in §5.3 as inconsistent with moralcognitivism, and is false if even the weakest form of strongmoral objectivism, e.g., Mackiean non—entifying moral206objectivism, rational prudence, is true.Another form of incommensurability is embodied in the viewthat goodness is not the only fundamental moral value. Forinstance, one might hold that goodness and rightness are bothfundamental moral values, that goodness is a consequentialistconsideration, that rightness is a deontic consideration, andthat there is no rational way of settling the conflicts betweenthe two that arise in normal human life. Perhaps this form ofincommensurability is assimilable to one of the previous fourtypes, perhaps not; in either case, no such problem arises forOCU or any similar utilitarian morality, for in such moralitiesrightness is always derived from goodness, as explained for OCTJin §1.2 (Moore 1903, p. 25, 147, agrees).A general form of incommensurability objection has been madeto utilitarianisms both hedonic and non-hedonic, e.g., by Taylor(1989, p. 230), who claims that things such as integrity,charity, and liberation are worthy in ways some of which areincommensurate, hence jointly uncapturable in the utilitariancalculus. Again, perhaps this form of incommensurability isassimilable to one of the four types discussed, perhaps not; ineither case, no such problem, arises for OCU or any other hedonicutilitarianism. Hedonic utilitarianism answers this objection byholding, first, that such things as integrity, charity, andliberation have no intrinsic value beyond their hedonic content,if any, and that most or all of their value is extrinsic,realized in their hedonic consequences, and second, that no207hedonic values are incommensurate. Also, as noted by Blake(1967, p. 434—35), the general objection applies to any form ofconsequentialism, and it is surely false that consequences areirrelevant to practical moral thinking, as they would be ifincommensurability prevented us from telling better from worseacts by their consequences.Thus the pluralistic hedonist who subscribes to OCU hasnothing to fear from the various objections to valuecommensurability. As a result, OCU has nothing to fear frompluralistic hedonism. Since OCU is consistent with (at least oneform of) moral pluralism, the fifth objection is refuted.2. The Synthetic A PrioriThe sixth objection to OCU is. that OCU assumes the existenceof some synthetic a priori propositions, but that no suchpropositions exist. OCU does make this assumption, so mustdefend the synthetic a priori. OCU makes this assumption byassuming that determinable properties exist; it will be seen thatdeterminable properties, whether construed as supervening ontheir determinates or as existing within them as parts, stand inmetaphysically necessary relations with their determinates, orwith those parts of their determinates that are ontologicallyindependent, and that the propositions describing these states ofaffairs are synthetic a priori. OCU also assumes the existence208of analytic a priori propositions, but these are much lesscontroversial entities, so will not be defended in this thesis.The question whether any propositions are synthetic a priorihas been answered positively and plausibly by Sikora (1981), butlet me provide a brief argument for this positive answer,incorporating, but not restricted to, some of his points. Aproposition was characterized in §2.1 as analytic if its logicalform, as revealed by the meanings or meaningful uses of itsterms, reveals its truth—value, and synthetic otherwise. I takea priori knowledge of a proposition’s truth—value to restessentially on a correct. understanding of the meanings ormeaningful uses of the proposition’s terms. Any propositionwhose truth—value is knowable in this way, i.e., by reasonunenriched by experience beyond the bare understanding of theproposition, is a priori; all other propositions are empirical.The concepts of analyticity, syntheticity, and a prioricity havebeen defined in many ways (Pap 1958, ch. 5). The reasons why thedefinitions given above are right for present purposes arecomplex, but let me say at least a few words on the subject.First, I take a proposition to be analytic if its logicalform, as revealed by the meanings or meaningful uses of itsterms, reveals its truth—value, and synthetic otherwise; mydefinitional reference to logical form might be objected to onthe ground that:on the basis of such a definition, to saythat logical principles are analytic would be209no more informative than to say that logicalprinciples are logical principles, and (thevarious) philosophers who maintained thatlogical truths are analytic surely intendedto make a significant statement, a statementclarifying the nature of logical truth. (Pap1958, p. 106)However, two considerations justify the definitional reference tological form.First, one can maintain that it is significant that logicaltruths and falsehoods are analytic, and hold that thesignificance of the remark is that it expresses the interestingtruth that two metaphysical terms from different contexts arebeing used to point at the same fundamental state of affairs.Likewise, it is interesting to learn that ‘x2 + y2 = 1’ and‘circle’ point at the same fundamental state of affairs. Thissort of discovery may allow proofs to link up hitherto separatedomains of inquiry. As Sikora points out in this context (1981,p. 447), the members of some perfectly legitimate sets of termscannot be defined via terms from outside their set. There arevicious circles and benign circles; one form of benigndefinitional circle is a dictionary.Second, we may further illuminate the terms ‘analytic’ and‘logical’ by linking them with the qualon/indexon distinctionintroduced in §1.3. Recall that a qualon is by definition anobject composed wholly of qualitative properties, and that an210indexon is by definition an object composed wholly of indexicalproperties. Qualitative properties ontically express what athing is, what kind of concrete stuff it is made of; indexicalproperties ontically express where, in a field of existence,possible or actual, a thing is. The logical form in virtue ofwhich the truth—value of an analytic proposition is revealed is aform manifested as an indexon. All states of affairs manifestingindexons, and nothing but indexons, thereby manifest logicalforms. Sikora makes the same point in different language when hesays thatin the case of analytic truths, unlike sap[synthetic a priori) truths, you don’t needto know what particular properties areconnoted by the various terms so long as youknow which of the various propertyrequirements are identical and which are not.(1981, p. 451)Husserl, also, links his version of the qualon/indexondistinction (see §1.3) to logic and the analytic/syntheticdistinction (1970, p. 455—58). Such links may be discerned lessexplicitly in the writings of many philosophers; e.g., Hare: ‘Noamount of logic will show us the difference between blue and red’(1981, p. 3).My second comment on why I use the definitions given ofanalyticity, syntheticity, and a prioricity is that it is betterto make these definitions accommodate both the truth—values,211truth and falsity, rather than just one. I take a proposition tobe analytic if its truth—value is suitably revealed to theinquiring mind, whether the truth—value revealed is truth orfalsity, whereas some define analyticity in such a way that onlytrue propositions are eligible (e.g., Bennett 1959, p. 172; Lewis1946, p. 35). My feeling here is that while one may say someinteresting things on the subject of analyticity whilerestricting the scope of one’s remarks to truths, there is noobvious reason to so restrict one’s remarks, for the points ofinterest are not about truth itself, but about some ways in whichtruth is manifested and revealed. These ways apply equally,mutatis mutandis, to truth and falsity, hence are better statedin the more general form, so as to better focus our attention onthe relevant phenomenon. Also, it would be impossible to define,e.g., ‘a priori’ in a way that usefully distinguished betweentruths and falsehoods according to how we learn them. If, forexample, we learn that it is true that ‘-i(p.-ip)’ by reason alone,then surely we may also learn that it is false that ‘p.—’p’ by.reason alone. Since each such truth learned generates afalsehood learned, and vice versa, most of our comments on the apriori may for brevity be phrased in terms of only onetruth—value. However, such comments strictly speaking apply toboth truth—values; likewise, mutatis mutandis, for the terms‘empirical’, ‘analytic’, and ‘synthetic’.My third comment on why I use the definitions given ofanalyticity, syntheticity, and a prioricity concerns the third of212these items. I take a priori knowledge of a proposition’struth—value to rest essentially on a correct understanding of themeanings or meaningful uses of the proposition’s terms. Anyproposition whose truth—value is knowable in this way, i.e., byreason unenriched by experience beyond the bare understanding ofthe proposition, is a priori; all other propositions areempirical. One of my reasons for using this definition is that,as Sikora says (1981, p. 444), the notion of a prioricity is anepistemological notion about how we know certain things. Mydefinition gives an account of the process by which we come toknow the relevant things. Another way of putting it would be todefine, as Sikora does (1981, p. 445), a priori knowledge asknowledge which lacks or does not require any inductive support.While it is true that a priori knowledge lacks or does notrequire any inductive support, it, as knowledge, does requiresome kind of support. A definition in the present context mayhelpfully give a positive account of the kind of support a prioriknowledge rests on; my definition tries to give such an account.Sikora’s definition accurately locates the boundary between apriori and empirical knowledge, and gives a positive account ofwhat is on the empirical side of the boundary, but says littleabout what is on the a priori side. For some purposes this isfine — it may be sufficient to locate the boundary accurately,which Sikora’s definition does concisely, and have a positiveaccount of what is on the empirical side of the boundary. Butfor other purposes, such as those of my argument of §2.3213concerning the third account of resemblance, it helps to alsogive a positive account of the kind of support a priori knowledgerests on. Part of the purpose of this section is to see thateach side of the a priori/empirical boundary is given a positiveaccount. Similarly, my definition of ‘synthetic’ was negative; apositive account will be given later in this chapter.My fourth comment on why I use the definitions given ofanalyticity, syntheticity, and a prioricity also concerns thethird of these items. The a priori/empirical distinction issometimes restricted in scope to propositions whose truth—valuesare humanly knowable, so that propositions whose truth—values arenot humanly knowable are neither a priori nor empirical (e.g.,Bradley and Swartz 1979, p. 150). My definitions avoid thisrestriction because the nonexhaustiveness of the distinction soconstrued is unnecessary and moves one’s attention away from theepistemic properties of the proposition, where attention shouldbe, and toward the epistemic properties of the potential knower,where it should not be. Attention should be on the epistemicproperties of the proposition in the sense that in order toappreciate the full epistemic potential of propositions, we mustconsider their interactions with the widest possible selection oftypes of agents seeking knowledge. This precludes ourconsidering only the human type. As philosophers we should seekto appreciate the full epistemic potential of propositions, andthe terms ‘a priori’ and ‘empirical’, if intended to be be aboutpropositions’ amenability to rational inquiry, should thus214consider propositions’ interactions with all types of rationalagent, including the most fully, hence superhumanly, rational.take ‘a priori’ and ‘empirical’ tobe terms of metaepistemology,whereas questions about the limits of human rationality areaddressed in normative epistemology.There is no metaepistemologically useful place to draw theline between what is humanly knowable and what is not. If whatis humanly knowable is limited to what humans know, then theknowable/known distinction collapses and ‘humanly knowable’ losesits utility. If what is humanly knowable includes propositionsunknown to humans, then it must, to be metaepistemologicallygeneral, include any proposition it is possible for a human toknow. Since it is logically and metaphysically possible for ahuman to have any rational power, this construal collapses‘humanly knowable’ and ‘rationally knowable’, again removing ourreason for coining the former term. It is possible for a humanto have any rational power in the sense that whether one isconsidering logical possibility, causal possibility, or whateverscope of possibility one pleases, any rational power within thatscope can be combined with human possession of that power.Humans can change; we can, if we wish, add devices for physicalabilities we lack, for senses we lack, or for rational powers welack. There is no obvious inconsistency between the concepts ofan agent’s being human and an agent’s possessing a given rationalpower. We already count, as humanly knowable, propositionsrequiring the epistemic cooperation of devices such as215binoculars, electron microscopes, and computers (e.g., Bradleyand Swartz 1979, P. 149); how could we consistently admit thesedevices but not a device yielding some other epistemically usefulpower, such as a given rational power?Assuming these accounts of the analytic/synthetic and apriori/empirical distinctions, what examples might be given ofsynthetic a priori propositions? They would be propositionswhose truth—values are knowable by reason unenriched byexperience beyond the bare understanding of the proposition, butare not revealed by its logical form. A candidate synthetic apriori proposition is:(1) Every hue has an intensity.One may know what a color hue is and what an intensity is, andknow that the logical form of (1) does not reveal itstruth—value, yet also see that anything with a hue must also havecopresent, intermingled, and, in Husserl’s term,‘interpenetrating’ (1970, p. 437), an intensity, hence see that(1) is true.It might be objected to the claim that such propositions as•(1) are synthetic that they are expressions from a naturallanguage whose words have meanings and meaningful uses that arein their details poorly understood, evolved for social ratherthan purely philosophic functions, and perhaps have or lackanalytic entailments in such fashion as to render my claims aboutsuch propositions false, even though such entailments or theirlacks are distracting linguistic baggages which might be216discarded to the gain of conceptual precision — that is, to thegain of clarity and conciseness, and without begging the questionin favor of the synthetic a priori. I agree that this ispossible, hence stipulate, in agreement with Sikora, that thepropositions in question are to be regarded as expressed in anartificial language purged of such distracting analyticentailments or their lacks. For instance, in the case ofproposition (1), saying in ordinary English that something has agiven ‘hue’ may well logically imply that it has some intensity;this requirement is explicitly rejected in the artificiallanguage proposed. The linguistic rule of usage discarded, wecan then examine the property of hue — the core propertyostensibly required for the satisfactory application of ‘hue’ —and see whether it contains as a part the property of intensity,and if not, whether it must nonetheless by metaphysical necessitybe accompanied by the property of intensity. Sikora uses hue andintensity to explain how core properties may be identified anddistracting linguistic rules of usage discarded:You are shown two batches of color swatches.All the swatches in the first group have thesame determinate hue but there is no otherdeterminate quality common to all of them.In the second batch they all have the samedeterminate intensity but again this is theonly determinate quality common to all ofthem. You are told that ‘L—red--156’ is to217stand for the common determinate quality ofthe first batch and that ‘L—intensity’ is tostand for the determinate quality of whichthe determinate quality common to all membersof the second batch is a determinate version.Further, to avoid ambiguity, you are toldthat although all the instantiations ofL—red—156 that you have been shown have someL—intensity, saying that something isL—red---156 is not to be taken as logicallyimplying that it has some L-intensity.Saying in ordinary English that something hasa given hue may well logically imply that ithas some intensity, but this requirement isexplicitly rejected for the artificialterminology in question. Given thesestipulations it would not be contradictory tosay that something could be L—red—156 yetlacked an L—intensity. None of the rulesgoverning the use of the two key terms wouldpreclude such an object: you didn’t stipulatethat for something to be L—red—156 it musthave an L—intensity, only that it must havethe determinate quality common to the firstbatch of swatches. . . . Thus it is asynthetic truth that anything that is218L—red—156 must have an L—intensity, and sinceit is not an empirical truth it is asynthetic a priori truth. (1981, p. 450—51)Thus sloppy color wérds from ordinary English, such as ‘hue’ and‘intensity’, may be used to help define theoretical terms preciseenough to permit inferences of philosophic rigor, much as wasclaimed in §1.2 of sloppy moral words from ordinary English.It might be objected to synthetic a priori claims, e.g., ofthe form of (1), asserting that one thing must always beaccompanied by another thing, that the necessity is an artifactof consciousness, that the mere fact that we cannot, e.g.,imagine the one thing without the other, demonstrates a limit ofour imagination, not of external reality. But as Husserl says,differences such as this, that one objectcan be ‘in and for itself’, whileanother can only have being in, or attachedto some other object — are no merecontingencies of our subjective thinking.They are real differences, grounded in thepure essences of things, which, since theyobtain, and since we know of them, prompt usto say that a thought which oversteps them isimpossible . . . (1970, p. 445)As explained in §3.1, Occam’s Razor presumes against theexistence of all intuitions, but once granted their existencepresumes in favor of their veridicality. Thus if we have the219intuition that a thing is impossible, then we should try toinclude in our world—view the hypothesis that the thing isimpossible, not the more complicated hypothesis that the thing ispossible but that to have an intuition that the thing is possibleis impossible or at least counterfactual. Unless independentevidence indicates otherwise, we should accept our intuitions(assuming we have them) that some propositions, e.g., of the formof (1), are synthetic a priori. It makes no more sense to doubtthese intuitions than to doubt our intuitions that analytically,necessarily 2 + 2 = 4. My arguments regarding, e.g., (1), aredesigned to arouse the former intuitions in rational persons, andeliminate contrary intuitions.It should also be noted that the necessity attributed topropositions known a priori is not epistemic, but ontic, alogical or metaphysical state of affairs an a priori propositionexpresses. One can no more be certain of a priori knowledge thanof empirical knowledge — a mad scientist or Cartesian demon mightbe tricking us in either case by supplying misleading evidence.A priori knowledge is necessary only in that the proposition inquestion is claimed to have the same truth—value in all possibleworlds. It is irrelevant to point out that foolish people oftenclaim to be absolutely certain of various propositions; theircertainty is unwarranted by the evidence available to them. Noamount of evidence warrants absolute certainty about anything,and in any case, certainty is a psychological state of abeliever, not a logical or metaphysical state of affairs a220believed proposition expresses (see Pap 1958, p. 125—27).Note that one can consistently hold that while neither apriori nor empirical knowledge can yield justified absolutecertainty, some a priori knowledge can justifiably yield apsychological state nearer to absolute certainty than can anyempirical knowledge. Broad, for instance, seems to hold thisview; he asserts, of a way of acting, that “we know as surely aswe can know anything that is not a priori, that by no meanseverybody will act in this way” (1916, P. 377—8). Thus one canin principle know two propositions beyond reasonable doubt yethave less doubt about one than the other. OCU. allows that apriori and empirical knowledge yield different maximum degrees ofhuman confidence, but does not insist upon this view.The reason OCU assumes the existence of some synthetic apriori propositions is that whether a determinable accompaniesits determinates as a part or as a supervening ontologicallyindependent property, the situation contains some metaphysicalnecessity, expressable in a synthetic a priori proposition. Itwas explained in ch. 2 how the latter state of affairs, with itssupervening ontologically independent determinable, is describedby a synthetic a priori proposition; now consider how the formerstate of affairs, in which a determinate is logically necessarilyaccompanied by its determinable as a part, has an aspectdescribed by a synthetic a priori proposition. Recall that onany such account, determinate—determinable propositions areanalytic; e.g., the proposition ‘All red things are colored’ is221of the form x((Fx.Cx)-Cx) where ‘C’ attributes coloredness and‘F’ attributes some property F which combined with colorednessmakes redness. C and F are thus ontologically independentproperties, and yet it is inconceivable that anything F shouldnot also be C. Therefore x(Fx-Cx) is a synthetic a priori truth.The only way to deny the existence of such synthetic apriori propositions is to deny either that determinate propertiesresemble each other in the way in question, or to deny that thedeterminable—property theory is the correct account of suchresemblances.The first disjunct is false. Determinate properties doresemble each other in various ways. In particular, every colorproperty resembles every other color property in being a color.As mentioned in §2.3, Armstrong is right that “these recalcitrantresemblances seem objective phenomena, demanding an ontologicalanalysis” (1978, vol. 2, p. 116).The second disjunct, that these resemblances exist but havecorrect accounts other than variants of the determinable—propertytheory, is dubious because both the chief rival accounts aredubious. To show this I reviewed in §2.3 the five chiefontological accounts of resemblance. Three turned out to positdeterminables, and the other two turned out to be flawed.Thus whether a determinable accompanies its determinates asa part or as a supervening ontologically independent property,the situation contains some metaphysical necessity, expressiblein a synthetic a priori proposition. While Moore did not make222use of determinable—determinate property theory in his account ofgoodness and goods, he did grasp that the relations betweengoodness and goods always contain some metaphysical necessity:“propositions about the good are all of them synthetic and neveranalytic” (1903, p. 7). While this claim is false because toogeneral (consider the proposition ‘all good things are good’), weare now able to reconstruct the true claim Moore was trying tomake.Finally, let us examine the alternative to the existence ofsynthetic a priori propositions. Since the analytic/syntheticdistinction was taken to be exhaustive of propositions, thefollowing proposition would be true:(2) All a priori propositions are analytic.Moreover, (2) would be knowable a priori, according to at leastsome skeptics of synthetic a priori propositions, e.g., M.Schlick (see Pap 1958, p. 94—95). Therefore (2) itself would beanalytic (leaving aside the problem of self—referentialpropositions). Since analytic propositions are those whosetruth—values are revealed by their logical form, and byhypothesis (2) is analytic, analyticity would therefore have tobe logically contained in a prioricity. A prioricity obtains ofany proposition whose truth—value is revealed by theproposition’s terms’ meanings or meaningful uses; analyticityobtains of any proposition whose truth—value is revealed by theproposition’s logical form. A proposition is an attribution ofproperties to individuals. The logical form of a proposition223does not depend on the nature of the properties attributed. Thenature of a property is expressed in the meanings or meaningfuluses of terms. Thus the logical form of a proposition does notdepend on what is expressed in some meanings or meaningful usesof terms of propositions. Hence such meanings or meaningful usescould be changed without changing the proposition’s logical form.Therefore logical form is not logically contained in somemeanings or meaningful uses of terms. Thus analyticity is notlogically contained in a prioricity, hence (2) is not, as washypothesized, analytic.To put the point slightly differently, a prioricity obtainsof any proposition whose truth—value is revealed by theproposition’s asserted tautological indexons, contradictoryindexons, metaphysically necessary qualons, or metaphysicallyimpossible qualons; analyticity obtains of any proposition whosetruth—value is revealed by the proposition’s assertedtautological indexons or contradictory indexons. Since adisjunct is not logically contained in a disjunction in which itappears (that is, p or q does not logically imply q), aprioricity does not logically contain analyticity, hence (2) isnot, as was hypothesized, analytic.One might try to fall back on the position that (2) is truebut empirical rather than a priori, but this position isdifficult to state coherently.Empirical propositions are usually thought of as being truein some possible worlds and false in some possible worlds,224experiental evidence being needed to determine which of these twosets of worlds the actual world belongs to. But if (2) is falsein some possible world, then in some world some a prioriproposition is not analytic, i.e., is synthetic. But if a givenproposition is synthetic a priori in some possible worlds, surelyit is synthetic a priori in all possible worlds in which itexists, for synthetic a prioricity is a way in which aproposition is via its intrinsic nature knowable. One might holdthat the relevant proposition just happens not to exist in theactual world, but then one must reject the widely—acceptedpremise that any proposition that exists in one possible worldexists in all possible worlds (e.g., Bradley and Swartz 1979, p.13—24).Alternatively, one might characterize empirical propositionsas including propositions which are true in all possible worlds,but nonetheless in need of experiental evidence for assessment oftheir truth—values. But if it is impossible for suchpropositions to be false, why would this impossibility not berevealed in their terms’ meanings or meaningful uses, henceknowable a priori, hence not in need of experiental evidence forassessment of their truth—values? Analytic impossibility wouldbe so revealed, as would synthetic impossibility; is there someother relevant kind of impossibility? A coherent account of suchimpossibility would be needed before this position could beaccepted.Thus the sixth objection to OCU is refuted.2253. ConclusionTo conclude this thesis, OCU is a viable comprehensive moraltheory, able to answer the six objections currently mostinfluential against it among metaethicists and some others. Itwas also revealed in the course of this investigation that manyalleged competitors to OCU for the status of a plausiblecomprehensive moral theory are not in fact comprehensive, butduck or beg central metaethical questions. The remainingcompetitors to OCU suffer important unanswered objections;therefore OCU can justifiably claim to be the most plausiblecomprehensive moral theory.226BIBLIOGRAPHYAngel, Leonard (1989), How to Build a Conscious Machine.Boulder: Westview.Aristotle (1985), Nicomachean Ethics. Trans. Terence Irwin.Indianapolis: Hackett.Armstrong, David N. (1978), Universals and Scientific Realism.Vol. 2. Cambridge: Cambridge UP. 2 vols.(1989), Universals: An Opinionated Introduction.Boulder: Westview.Baier, Kurt E.M. (1958), The Moral Point of View. Ithaca, NewYork: Cornell UP.Baylis, Charles A. (1967), ‘Grading, Values, and Choice’,Problems of Moral Philosophy. Ed. Paul W. Taylor.Belmont, California: Dickenson. 439—51.Bennett, Jonathan (1959), ‘Analytic—Synthetic’, AristotelianSociety Proceedings 59: 163—88.(1984), A Study of Spinoza’s Ethics. Indianapolis:Hackett.Blake, Ralph Mason (1967), ‘Why Not Hedonism?’, Problems of MoralPhilosophy. Ed. Paul W. Taylor. Belmont, California:Dickenson. 431—39.Boyd, Neil (1991), High Society. Toronto: Seal, 1993.Bradley, Raymond, and Norman Swartz (1979), Possible Worlds.Indianapolis: Hackett.Brandt, Richard B. (1959), Ethical Theory. Englewood Cliffs,N.J.: Prentice.(1963), ‘Toward A Credible Form Of Utilitarianism’,Morality and the Language of Conduct. Eds. H.N. Castanedaand G. Nakhnikian. Detroit: Wayne State UP. 107—43.(1979), A Theory of the Good and the Right. Oxford:Clarendon Press.Brentano, Franz (1889), The Origin of our Knowledge of Right andWrong. 3rd ed. Eds. Oskar Kraus and Roderick M. Chisholm,Trans. Roderick M. Chishoim and Elizabeth H. Schneewind.London: Routledge and Kegan Paul, 1969.Broad, C.D. (1916), ‘On the Function of False Hypotheses inEthics’, International Journal of Ethics 26: 377—97.227Chishoim, Roderick M. (1969), ‘The Loose and Popular and theStrict and Philosophical Senses of Identity’, Perception andPersonal Identity: Proceedings of the 1967 OberlinColloquium in Philosophy. Eds. Norman S. Care and Robert H.Grimm. Cleveland: Case Western Reserve UP. 82—106.Dancy, Jonathan (1988), ‘Supererogation and Moral Realism’, HumanAgency. Eds. Jonathan Dancy, J.M.E. Moravcsik, and C.C.W.Taylor. Stanford, California: Stanford UP.170—88.(1991), ‘Intuitionism’, A Companion to Ethics. Ed. PeterSinger. Oxford: Basil Blackwell. 411—19.Danto, Arthur (1963), ‘What We Can Do’, The Journal ofPhilosophy 60: 435—45.Dennett, Daniel C. (1982), ‘Mechanism and Responsibility’, FreeWill. Ed. Gary Watson. Oxford: Oxford UP. 150—73.Diggs, B.J. (1978), ‘Rules and Utilitarianism’, ContemporaryUtilitarianism. Ed. Michael D. Bayles. Gloucester, Mass.:Peter Smith. 203—38.Dolby, R.G.A. (1973), ‘Philosophy and the Incompatibility ofColours’, Analysis 34: 8—16.Foley, Richard (1978), ‘Prudence and the Desire Theory ofReasons’, Journal of Value Inquiry 12: 68—73.Gewirth, Alan (1978), Reason and Morality. Chicago: U ofChicago P.Goldman, Alvin I. (1970), A Theory of Human Action. EnglewoodCliffs, New Jersey: Prentice.Goldstein, Irwin (1985), ‘Hedonic Pluralism’, PhilosphicalStudies 48: 49—55.(1989), ‘Pleasure and Pain: Unconditional, IntrinsicValues’, Philosophy and Phenomenoloqical Research 50:255—76.Gould, S.J. (1993), ‘The Sexual Politics of Classification’,Natural History 102,11: 20—29.Hall, John C. (1967), ‘Quantity of Pleasure’, AristotelianSociety Proceedings 67: 35—52.Hardin, Garrett (1985), Filters Against Folly. London: Penguin,1987.Hare, R.M. (1981), Moral Thinking. Oxford: Clarendon Press.228Harrod, R.F. (1936), ‘Utilitarianism Revised’, Mind 45: 137—56.Heyd, David (1982), Supererogation: Its Status in Ethical Theory.Cambridge: Cambridge UP.Hobart, R.E. (1934), ‘Free will as involving determination andinconceivable without it.’ Mind 43: 1—27.Huine, David (1888), A Treatise of Human Nature. Ed. L.A.Selby—Bigge. 2nd ed. Oxford: Clarendon Press, 1987.Husserl, Edmund (1970), Logical Investigations. Vol. 2. Trans.J.N. Findlay. London: Routledge & Kegan Paul. 2 vols.Johnson, W.E. (1921), Logic, Part I. Cambridge, U.K.: CambridgeUP.Kagan, Shelly (1984), ‘Does Consequentialism Demand Too Much?’,Philosophy and Public Affairs 13: 239—54.Kant, Immanuel (1948), Groundwork of the Netaphysic of Morals.Trans. H.J. Paton. New York: Harper, 1964.Kenshur, Oscar (1984), ‘The Rhetoric of Incommensurability’,Journal of Aesthetics and Art Criticism 42: 375—82.Kim, Jaegwon (1978), ‘Nomological Incommensurables’, .PmericanPhilosophical Quarterly 15.2: 149—156.Langford, Michael J. (1988), ‘Supererogation and Friendship’,Philosophy and Culture 3: Proceedings of the 17th WorldCongress of Philosophy. Ed. V. Cauchy. 441—45.Lewis, Clarence Irving (1946), An Analysis of Knowledge andValuation. La Salle, Ill.: Open Court.Locke, John (1894), An Essay Concerning Human Understanding.Vol. 1. Oxford: Clarendon Press. 2 vols.Lyons, David (1967), Forms and Limits of Utilitarianism. Oxford:Clarendon Press.Mackie, J.L. (1977), Ethics: Inventing Right and Wrong.Harmondsworth, Middlesex, England: Penguin.Mazlish, Bruce (1975), James and John Stuart Mill. New York:Basic.McCarty, Richard (1989), ‘The Limits of Kantian Duty, andBeyond’, American Philosophical Quarterly 26: 43—52.McDowell, John (1985), ‘Values and Secondary Qualities’, Morality229and 0bjectivity. Ed. Ted Honderich. London: Routledge &Kegan Paul. 110—29.Meserve, Bruce E., Max A. Sobel, and John A. Dossey (1989),Introduction to Mathematics. 6th ed. Englewood Cliffs, NewJersey: Prentice.Mill, John Stuart (1861), Utilitarianism. Ed. Oskar Piest. NewYork: Macmillan, 1957.Montague, Phillip (1989), ‘Acts, Agents, and Supererogation’,American Philosophical Quarterly 26: 101—11.Moore, George Edward (1903), Principia Ethica. Cambridge:Cambridge UP, 1959.• (1912), Ethics. London: Oxford UP, 1963.(1922), Philosophical Studies. London: Routledge & KeganPaul, 1965.• (1959), ‘Is Goodness a Quality?’, Philosophical Papers.New York: Collier, 1966. 89—100.Nagel, Thomas (1970), The Possibility of Altruism. Oxford:Clarendon Press.• (1974), ‘What Is It Like To Be a Bat?’, PhilosophicalReview 83: 435—50.Neurath, Otto (1937), ‘Unified Science and its Encyclopedia’,Philosophy of Science 4: 265—77.Nowell—Smith, P.H. (1954), Ethics. Harmondsworth, Middlesex,England: Penguin, 1965.Oddie, Graham (1991), ‘Supervenience, Goodness, and Higher—OrderUniversals’, Australasian Journal of Philosophy 69: 20—47.Pap, Arthur (1958), Semantics and Necessary Truth. New Haven andLondon: Yale UP, 1966.Perry, John (1979), ‘The Problem of the Essential Indexical’,Nous 13: 3—21.Rawls, John (1971), A Theory of Justice. Cambridge,Massachusetts: Belknap—Harvard UP.• (1978), ‘The Basic Structure as Subject’, Values andMorals. Eds. A.I. Goldman, and J. Kim. Dordrecht,Holland: D. Reidel. 47—71.Raz, Joseph (1986), ‘Value Incommensurability: SomePreliminaries’, Aristotelian Society Proceedings 86: 117—34.Rescher, Nicholas (1973), The Coherence Theory of Truth. Oxford:Clarendon Press.230Rorty, Richard (1989), ‘Solidarity or Objectivity?’, Anti—Theoryin Ethics and Moral Conservatism. Eds. Stanley G. Clarkeand Evan Simpson. Albany, New York: State U of New York P.167—83.Ross, William David (1967), ‘Prima Facie Duties’, Problems ofMoral Philosophy. Ed. Paul W. Taylor. Belmont, California:Dickenson. 266—78.Russell, Bertrand (1912), The Problems of Philosophy. New York:Oxford UP, 1966.(1959), My Philosophical Development. New York: Allen.(1986), ‘The Philosophy of Logical .Atomism’, ThePhilosophy of Logical Atomism and Other Essays: 1914—19.Ed. John G. Slater. London: Allen. 160—244.Ryle, Gilbert (1949), The Concept of Mind. London: Hutchinson,1966.Shoemaker, Sydney (1990), ‘Qualities and Qualia: What’s in theMind?’, Philosophy and Phenomenological Research 50,Supplement, Fall: 109—31.Sidgwick, Henry (1874), The Methods of Ethics. 5th ed. London:MacMillan, 1893.Sikora, Richard I. (1962), ‘Reason, Language, and Choice: acritical analysis of contemporary development in ethicaltheory’, Diss. Berkeley.(1978), ‘Is It Wrong to Prevent the Existence of FutureGenerations?’, Obligations to Future Generations. Eds.R.I. Sikora and Brian Barry, Philadelphia: Temple UP.112—66.(1981), ‘Synthetic A Priori Truths in an ArtificialLanguage’ Philosophy Research Archives 7.1449: 443—60.Simpson, George Gaylord (1969), Bioloqv and Man. New York:Harcourt.Sleigh, Nicholas (1984), ‘Brandt and Daniels on Credence in MoralBelief’, Diss. Simon Fraser University, Essay I.(1992), ‘Objective Goodness and Aristotle’s Dilemma’,Journal of Value Inquiry 26: 341—51.(1993), The Bion: the fundamental unit of evolution bynatural selection. Unpublished ms., draft of May 1.Smart, J.J.C. (1956), ‘Extreme and Restricted Utilitarianism’,Philosophical Quarterly 6: 344—54.• (1973), ‘An outline of a system of utilitarian ethics’,Utilitarianism: for and against. By J.J.C. Smart and B.Williams. Cambridge: Cambridge UP. 3—74.231Smart, R.N. (1964), Philosophers and Religious Truth. London:S.C.M. Press.Spinoza, Benedictus de (1967), Ethics. Trans. Andrew Boyle.Revised G.H.R. Parkinson. London: Dent.Strawson, Peter F. (1949), ‘Ethical Intuitionism’, Philosophy 24:23—33.(1982), ‘Freedom and Resentment’, Free Will. Ed. GaryWatson. New York: Oxford UP. 57—80.Szalay, Frederick S., and Eric Delson (1979), EvolutionaryHistory of the Primates. New York: Academic.Taylor, Charles (1989), ‘The Diversity of Goods’, Anti—Theory inEthics and Moral Conservatism. Eds. Stanley C. Clarke andEvan Simpson. Albany, New York: State U of New York P.223—40.Timberlake, William, and Lucas, Gary A. (1989), ‘Behavior Systemsand Learning: From Misbehavior to General Principles’,Contemporary Learning Theories: Instrumental ConditioningTheory and the Impact of Biological Constraints on Learning.Eds. Stephen B. Klein and Robert R. Mowrer. Hilisdale, NewJersey: Lawrence Erlbaum. 237—75.Tocqueville, Alexis de (1840), Democracy in America. Vol. 2.New York: Random House, 1990. 2 vols.Trivers, Robert (1985), Social Evolution. Menlo Park,California: Benj amin/Cummings.Williams, C.J.F. (1969), ‘Are Primary Qualities Qualities?’,Philosophical Quarterly 19: 310—23.232