Metaethics Quentin Smith

From pp. 160-179 of Ethical and Religious Thought in Analytic Philosophy of Language, by Quentin Smith, Yale University Press, 1997.

As noted above, Adams inaugurated essentialist ethics in it with his article on divine command ethics, which uses the essentialist method to argue for a certain type of moral realism. He applies the essentialist linguistic method to ethical terms, specifically, to the word “wrongness,” to arrive at a divine command ethics. He wants to argue that it is an a posteriori and metaphysically necessary truth that wrongness is contrariety to the commands of a loving God. His conclusion is that it is necessary that the property of wrongness be identical with the property of contrariety to the commands of a loving God, but that this identity is known only a posteriori. Although I shall criticize Adams’s conclusion, I shall endorse his method of ethical knowledge and shall later use it myself because I believe it is superior to the method of reflective equilibrium and other methods of ethical knowledge that have been practiced in the 197os, 198os, and 199os.

Adams plausibly argues that every competent user of “wrong” need not know the nature of wrongness; indeed, people have different views about the nature of wrongness, and some have no views at all. This fact must be distinguished from views about what actions possess the property of wrongness. You and I could agree that stealing has the property of wrongness but disagree about the nature of the property designated by “wrongness” or even be ignorant of this nature. Every competent language-user, however, does know that wrongness is a property of actions and that people are generally opposed to actions they regard as wrong and generally count wrongness as a reason for opposing an action.

These commonly understood aspects of wrongness belong to what Adams calls the concept or understanding with which the word “wrong” is used. Using the distinctions previously made in my account of the essentialist philosophy of language, I may say that what Adams calls the concept of wrongness is both (a) the cognitive significance of “wrong” and (b) the reference-fixing description that governs the use of “wrong.” The concept of wrongness is both (a) what speakers have in mind when they say something is wrong and (b) determines which property is the direct referent of the word “wrong.” The semantic content or direct referent of the predicate “wrong” is the property contrariety to the commands of a loving God.

As I suggested in chapter 4, the cognitive significance of a locution may diverge from its reference-fixing description. For example, some items may be referents of proper names and the names may not have their reference fixed by any description; the reference of “John Doe” may be fixed by ostension (that baby), not a definite description, and yet the (or a) cognitive significance of “John Doe” may be the definite description the president of University U in 1996. What speakers “have in mind” when they think of John Doe is not that baby but the definite description the president of University U n 1996 (or the husband of Jane, and so forth). Although Adams does not make this distinction, it is both plausible and consistent with what he does say, namely, that “wrong” has a reference-fixing description and that this is also its cognitive significance.

The property designated by “wrong” is the property that best fills the role assigned to wrongness by the concept (set of descriptive conditions). Adams lists several descriptive conditions for this role, some of which I have already briefly mentioned:

i. Wrongness is a property that actions possess independently of whether we think they do. (This is a moral realist criterion and conforms to the arguments I presented in chapter i, 2, and 3 that our first—level ordinary moral beliefs are realistic. I argued that this pertains to both our particular attitudes to particular actions and to our general attitudes to basic types of actions.)

ii. Wrongness is exemplified by an important central group of those actions that are generally thought to be wrong, for example, stealing, torturing children for fun, murder.

iii. The property of wrongness plays a causal role in the relevant actions coming to be regarded as wrong.

iv. Understanding the nature of wrongness (that is, the referent of “wrongness”) should give us one more, rather than one less, reason to oppose wrong actions.

v. Adams ’s fifth condition is a general condition that the theory about wrongness satisfy other intuitions about wrongness as far as possible.

Adams allows that many different theories about the nature of wrongness meet his five criteria but notes that “given typical Christian beliefs about God, it seems to me most plausible to identify wrongness with the property of being contrary to the commands of a loving God.” Adams continues, “Ethical wrongness is (i.e. is identical with) the property of being contrary to the commands of a loving God. I regard this as a metaphysically necessary, but not an analytic or a priori truth. .. . It purports to be the correct theory of the nature of ethical wrongness that everybody (or almost everybody) is talking about” (115).

It is clear from this passage that Adams is relying on linguistic essentialism for his ethical conclusions. Although Adams does not go into detail, one may construct the following argument about wrongness. I shall assume as a suppressed premise, as Adams means to assume, that typical Christian beliefs are true. The argument goes

1. The reference-fixing description of “wrong” is that this word directly and rigidly designates whatever property (a) is exemplified by some actions, (b) gives us a reason to oppose those actions, (c) is an objective (moral realist) property, (d) is exemplified by certain paradigmatic actions, such as stealing, murder, rape, torturing children for fun, and so on, and (e) plays a causal role in our coming to regard the relevant action, and so on as wrong.

2. The property that best fulfills the reference—fixing rule of use of “wrong” is contrariety to the commands of a loving God.

3. “Wrongness” rigidly designates contrariety to the commands of a loving God.

4. What is said by the statement “Wrongness is identical with contrariety to the commands of a loving God” is the same thing (proposition) that is said by the statement “Contrariety to the commands of a loving God is identical with contrariety to the commands of a loving God.”

Although Adams does not make these distinctions, we may say that this is a case of a posteriori necessary identifications, such as “Hesperus is Phospherus” and “Water is H₂O The object—level fact is an identity proposition and thus a priori. The metalevel fact, namely, that the statement “Wrongness is identical with contrariety to the commands of a loving God” states this identity proposition, rather than some other proposition, is a posteriori and contingent.

In his article, Adams makes the comment that it is a metaphysically necessary a posteriori truth that wrongness is contrariety to the commands of a loving God. This does not seem to be correct, given the argument implicit in the article. Correctly formulated, one should say that what is said by the statement “Wrong ness is contrariety to the commands of a loving God” is not an a posteriori truth but an identity proposition that is known a priori.

Second, this is a logical necessity in the sense of first-order predicate calculus with identity and does not involve the modality of metaphysical necessity any more than any other logical tautology.

Third, the a posteriori feature belongs to the metalevel proposition that the sentence states this proposition rather than some other proposition. The reference—fixing rule of use of “wrong” allows that this word refers to different properties in different possible worlds. Adams says, “There may be other possible worlds in which other properties best fill the role by which contrariety to a loving God’s commands is linked in the actual world to our concept of wrongness” (a i8). Adams gives the example that in some possible worlds, “wrong” refers to the property of failing to maximize human happiness. In these worlds, the sentence “Wrongness is contrariety to the commands of a loving God” expresses the logically false proposition, Failing to maximize human happiness is (identically one and the same property as) contrariety to the commands of a loving God. Thus, the sentence “Wrongness is contrariety to the commands of a loving God” is a contingent a posteriori truth, even though it states a logically necessary proposition.

Apart from this restructuring of Adams ’s argument, I must reject at least one of the elements that Adams believes to belong to the concept of wrong, namely, that the property of wrongness plays a causal role in our coming to regard certain actions as wrong. This seems to be a doubtful condition, even though endorsed by Putnam (“I would apply a generally causal account of reference also to moral terms”).[1] Adams offers no argument for this condition or even an explanation of what it could mean. Suppose I observe a woman beating her child, and I come to believe that her action is wrong. Light rays reflect from the woman and child, hit my retina, institute a brain process, and this results in my visual perception of her beating the child, If we suppose this action also possesses the property of being contrary to a divine commandment, how is this moral property supposed to play a causal role in my coming to believe that the beating is wrong? There is no physical causal process that extends from the beating’s being contrary to a divine commandment, no physical information that is carried by light particles, air waves, or some other physical means and reaches my sense organs. For the wrongness of the beating to be causally effective, it must be identical with a certain sort of physical property that can causally interact with my sense organs. But Adams ’s divine command property cannot causally affect me in this way. (And God cannot directly intervene and cause my moral belief because that is inconsistent with the libertarian free will that Adams ascribes to people.)

Moreover, it seems that this causal condition is not a part of every competent language-user’s concept of wrongness. For example, I believe right now that some wrong actions are being performed in China , and yet I also believe, quite correctly, that there is no causal process emanating from China (faster than the speed of light) to my current belief state.

As I have mentioned, Adams does not argue that his divine command ethics is true, but merely that a certain conditional statement is true, namely, “Given typical Christian beliefs about God, it seems to me most plausible to identify wrongness with the property of being contrary to the commands of a loving God” (i is). I argued in my chapter on essentialist philosophy of religion that God does not exist. Thus, even if Adams ’s conditional is true, the falsity of the antecedent leaves us with no reason to assume that wrongness is identical with the divine property Adams mentions. If one wants to establish moral realism, one cannot do so in a supernaturalistic context. Nonetheless, the linguistic essentialist method that Adams uses, properly reconstructed, can be abstracted from a supernaturalist context and used to establish a different ethical theory, as I shall argue later in this chapter. But first in the order of things is to examine Brink’s interesting attempt to establish a naturalistic moral realism.

In chapters i, 2, and 3, I argued that ordinary language and our first—level moral beliefs imply moral realism. Since the decline of ordinary language analysis, with its antirealist account of ordinary ethical discourse, several philosophers have realized that ordinary moral talk is in fact realist. Even some moral antirealists, such as Mackie, concede this point.[2] John Post, a moral realist, plausibly notes that “the presumption should be that there are objective values. For this is what ordinary usage of moral terms overwhelmingly presupposes, as does most of our actual moral reasoning.” Post observes that “Mackie is far more candid than most subjectivists in acknowledging that the burden of proof is on those who advance any such thesis [that is, moral antirealism].”[3]

Brink likewise adopts this stance: “We begin as (tacit) cognitivists and realists about ethics. . . . We are led to some form of antirealism (if we are) only because we come to regard the moral realist’s commitments as untenable, say because of the apparent occult nature of moral facts.”[4] The defense of moral realism, accordingly, amounts to responding to arguments against moral realism brought by antirealists. This is a project Brink undertakes in his book Moral Realism and the Foundation of Ethics, using in part the methodology of linguistic essentialism. There are several other recent defenses of moral realism, most importantly Butchvarov’s Skepticism in Ethics and some articles by Geoffrey Sayre-McCord, Nicholas Sturgeon, Richard Boyd, and others.[5] Only Brink’s defense, however, explicitly uses the method of linguistic essentialism, and for this reason I concentrate on his theory.

The two most familiar contemporary arguments against moral realism are the argument from queerness and the argument from moral disagreement. In the following sections, I shall critically discuss Brink’s response to these arguments and his naturalist version of moral realism, and I shall formulate what I believe to be plausible responses to these arguments.

The argument from queerness, developed by Mackie, was primarily directed against nonnaturalist versions of moral realism, such as were espoused by Moore, Prichard , and other logical realists. Mackie’s argument, in part, is that moral properties are so unusual that there is no good reason to postulate them. They are very unlike natural properties. A nonnatural property is queer. Further, the relation between moral properties and nonnatural properties is queer. Nonnatural proper ties do not entail moral properties or have any other sort of familiar relation to moral properties.

Brink’s theory meets these two charges of queerness by implying that moral properties are natural properties and are related to other natural properties by being constituted or composed by them. Since “being natural” and “being com posed of natural properties” are familiar or nonqueer predicates, moral realism is not metaphysically queer.

Brink argues for the naturalistic thesis that moral properties are constituted by natural properties, where the “is” in “The moral property F is the natural property G” is the “is” of constitution. Brink writes,

A table is constituted by, but not identical with, a particular arrangement of microphysical particles, since the table could survive certain changes in its particles or their arrangement. Similarly, moral properties are constituted by, but not identical with, natural properties if, though actually constituted or realized by natural properties, moral properties can be or could have been realized by properties not studied by the natural or social sciences. [Moral properties might have been constituted by supernatural properties, for example.] Moral properties may well be constituted by natural properties; they may be nothing over and above organized combinations of natural and social scientific properties. (158)

Although Brink’s language in this passage is tentative (“Moral properties may well be constituted by natural properties”), this is the position with which he most sympathizes and in effect endorses. Brink adds to this account of compositional naturalism that one and the same moral property may be constituted by different natural properties; for example, “The property of injustice . . . could have been realized by a variety of somewhat different configurations of social and economic properties” (158). This gives a certain sense to the claim that moral properties “supervene” on natural properties. Brink continues, “Moral properties supervene on natural properties because moral properties are constituted by natural proper ties” (158). The supervenience relation is a metaphysically necessary relation. If the natural properties G and H constitute the moral property F (and in this sense are the base properties upon which F supervenes), then in every metaphysically possible world in which x is G and H, x is F.

Brink’s argument for his “compositional naturalism” is based on the essentialists’ philosophy of language. Brink uses this philosophy of language to undermine the thesis of Moore and other logical realists that moral realism requires a nonnaturalism in ethics.

In chapter 1, I pointed out that Moore ’s argument that “moral realism implies nonnaturalism” is based on the “open question” argument and the charge of a naturalistic fallacy. Moore accepts a certain semantic thesis of the theory of refer ence that dominated analytic philosophy prior to the work of Marcus and others. The semantic thesis implies there are no relations of synonymity or meaning implication between moral and nonmoral terms. As Moore indicates, the “open question” argument shows that “pleasure” is not synonymous with “good” or any other natural predicate. An open question is one about which it is possible for a competent speaker to doubt an affirmative answer. “Is pleasure pleasant?” is a closed question. Moore argues that for any natural predicate N and any moral predicate G, it is an open question as to whether G things are G just insofar as they are N. It follows from this semantic argument that “good” is not synonymous with any natural predicate, and therefore that goodness is not identical with any natural property. Because “good” has a sense and the sense of a term is its referent, “good” has a nonnatural property for its referent, the property ought—to--exist.

Brink undertakes to refute this argument, using the distinction between semantic content and cognitive significance made by the linguistic essentialists. Brink points out that Moore assumes that if two predicates are synonymous, then all competent speakers associate the same properties with both terms. Moore assumes that properties which are the referents (semantic contents) of the predicates are also the cognitive significance of the predicates. Thus, if “pleasant” and “good” refer to the same property, they have the same cognitive significance for me—which is just the property they designate. Brink notes that this “appears to be a consequence of the traditional theory of meaning according to which the meaning of a term is the set of properties that any speaker competent with the term associates with it [the term]” (157).

Given the distinction between cognitive significance and semantic content, however, the open question argument does not show nonnaturalism is true. “Good” and some natural predicate may refer to the same property but have a different cognitive significance. There may be no a priori connection between the cognitive significance of “good” and the cognitive significance of some natural predicate, for example, “pleasure,” and it may require a posteriori knowledge to learn that they are coreferring or that the semantic content of the natural predicate constitutes the semantic content of the moral predicate.

A further feature of Moore ’s argument for nonnaturalism that is inconsistent with the essentialasts’ philosophy of language is his “semantic test of properties” (162). The semantic test implies that “meaning implication is a test of constitution or necessitation” (162). Thus a moral property F is constituted by the natural properties G and H only if the moral property F is part of the meaning (semantic content) of the conjunction of the natural predicates “G” and “H.” This is based on the idea that all necessary truths are analytic and thus that if the base property B necessitates the moral property M, then M is part of the meaning of “B.” Given this semantic test, M is constituted by and supervenes on B only if “x is B” analytically entails “x is M.”

Brink reconstructs Moore ’s argument as the claim that no natural sentence of the form “x is B” analytically entails a moral sentence of the form “x is M.” In other words, a natural base property B necessitates and constitutes a supervening moral property M just in case it is part of the meaning of the base predicate “B” that B things are M. Because there are no such analytic entailments, however, it follows (given Moore ’s philosophy of language) that no natural properties constitute moral properties and that constitutional ethical naturalism is false.

Brink responds that Kripke and Putnam have shown that analyticity and necessity are distinct notions; analyticity is a semantic notion and necessity is a meta physical (modal) notion. Given this, one cannot maintain that meaning implication is a necessary condition of a metaphysically necessary relation of constitution. The necessary connection can be synthetic (i66) and a posteriori (175).

Such reasoning suggests that moral truths are synthetic, a posteriori, metaphysically necessary, and involve the “is” of constitution. “Institutionalized slavery is unjust” may be analyzed as saying that the property of institutional slavery con stitutes the property of injustice, but injustice is not part of the semantic content of the predicate “institutionalized slavery.” Correlatively, an a priori analysis of the semantic content of “institutionalized slavery” will not enable us to know the proposition expressed by “Institutionalized slavery is unjust.” This proposition is know a posteriori, by examining the institutional structure and determining that it constitutes injustice.

Thus, Brink seems to arrive at an ethical naturalism, a compositional naturalism, and thereby rebuts Mackie’s argument for queerness. Natural facts and the relation of constitution are not queer, and this challenge to moral realism seems to be met.

Brink misunderstands Moore on one point; Moore held that some necessities are synthetic and a priori. But we are examining Brink’s essentialist ethics, and we may assume Brink’s understanding of Moore . But even assuming this, does Brink’s argument succeed?

Note that Brink does not hold that injustice can be constituted only by institutional slavery; many different sets of natural properties can constitute injustice. For each such set 5, if the properties in S are exemplified by x, then it is metaphysically necessary that x exemplifies injustice, such that injustice is composed of the prop erties in S. For example, the properties of institutional anti—Semitism or taxation without representation may also constitute injustice. Just as a table is constituted by a particular arrangement of microphysical particles but can survive changes in its particles or its arrangements, so injustice can be composed of different natural properties (157-58).

This theory seems problematic. If injustice is, as Brink says, “nothing over and above organized combinations of natural and social scientific properties” (158) such as the organized combination instantiated in the southern American states in the first half of the nineteenth century, then it seems to be identical with the organized combination of properties. This seems to be a general principle: If F is nothing over and above the organized combination of G and H, then F is identical with the organized combination of G and H. For F is “nothing over and above ( )“ means “F is nothing but ( ),“ which means “F is identical with ( ).“If F is not identical with the organized combination of the natural properties G and H, then F is something over and above this organized combination, and the part of F that is over and above the organized combination of the natural parts G and H will be (by definition) something that is not natural. (If it is natural, it will be part of the organized combination of natural properties that constitutes F.) On the other hand, if injustice is identical with this combination, it is necessarily identical with this combination, as Marcus’s theory of the necessity of identity implies, and consequently injustice cannot be identical with a different combination of properties, say the combination instantiated in Nazi Germany.

These considerations about moral properties being “nothing over and above” organized combinations of natural properties suggest that Brink’s naturalism is really an identity version of ethical naturalism, rather than a constitutional natural ism. But then it is difficult to see how Brink’s theory could be plausible, for with which organized combination shall we identify injustice? Is injustice identical with the set of properties instantiated in all and only institutions of slavery? Or is injustice identical with the set of properties instantiated in all and only anti-Semitic institutions? Given that both institutions are unjust and are not identical with each other, it follows that injustice is not identical with either set of properties.

Perhaps we could retain Brink’s constitutional naturalism if we said that the property of injustice is the disjunctive property, being an institution of slavery or being a sexist institution or being an anti—Semitic institution and so forth. Even if there are such things as disjunctive properties, which is doubtful, it is highly doubtful that moral properties are disjunctive properties. It seems quite clear that when we say that the Confederate States of America in the first half of the nineteenth century were unjust, we are not saying that they are either sexist or anti-Semitic or slave-owning or homophobic or taxed without representation, and so on.

Brink cannot be defended by invoking the idea that “a whole F is greater than the sum of its parts G and H,” for Brink himself has already invoked this idea; he says the whole F is not just the parts G and H but is the organized combination of G and H. My criticism is of his particular version of the “whole is greater than the sum of its parts” theory. Brink is silent about the different metaphysical theories of the nature of properties (Is F a universal, and the organized combination of G and H a combination of particular properties, tropes? Or are all properties universals? Or are all properties particulars?—these questions are not addressed). Thus, it is not apparent which, if any, metaphysical theory of properties could avoid the problems about F, G, and H that I have mentioned.

I think Brink’s theory is best revised to become a nonnaturalist constitutional ethics. Suppose we reject Brink’s statement that injustice is “nothing over and above” the organized combination of the natural properties G and H. (I use “natural” as an abbreviation for what Brink calls “natural or social or economic.”) Suppose we say that the whole property injustice contains an extra aspect that is “over and above” the organized combination of natural properties that constitute it. Although Brink does not say this, it seems to be a plausible way of viewing constitution theories.

Consider a table that is constituted by an organized combination of microscopic elements. Contrary to Brink’s theory of constitution, the table is not “nothing over and above” this organized combination of atoms because it in addition has the nontrivially essential property of being designed as an artifact by the person x. This nontrivial essence is an “individuating essence” in Marcus’s sense: it is partly definitive of the special character of the table and distinguishes the table from some other things of the same kind (the kind TABLE). The individuating essence of the table enables the table to retain its self-identity through changes and rearrangements of its parts.

Consider the property of institutional injustice (as distinct from unjust acts, unjust people, and so on). We may say that in some cases it is composed of the set of properties involved in institutional slavery, in other cases of the set of properties exemplified in anti-Semitic societies, and so forth. But injustice is not identical with either of these sets of properties. Injustice is composed (in each case) of some organized combination of natural properties and an extra nonnatural property, specifically, a nonnatural property of the various natural properties that partially constitute the different kinds of injustice. This property will be the nonnatural property ought-not-to-be-instantiated. This is a second—order property, a property of certain natural properties. There are different kinds of injustice, and the kinds are distinguished from each other by the different kinds of natural properties that exemplify ought-not-to-be-instantiated. Institutionalized slavery is one kind of injustice; this kind is the relevant set S₁ of properties that exemplify ought-not—to— be-instantiated. Institutionalized anti—Semitism is another kind of injustice: it is a different set S₂ of properties that ought-not--to-be-instantiated.

The genus institutional injustice, accordingly, is the complex property: being a kind of institution that ought not to be instantiated. The species of the genus are determined by the different kinds of institutions that ought not to be instantiated, racist, anti-Semitic, and so on.

No such genus / species distinction is possible on Brink’s constitutional natural ism, if only for the reason that there is nothing to play the role of the genus. Because we cannot include a nonnatural property (ought not to be instantiated), we are left with the natural property, being a kind of institution. Because this would also be the genus of just institutions, we would have a contradiction; just and unjust institutions would be indistinguishable. Perhaps some other sort of constitutional natural ism would be a workable ethical theory, but it is difficult to see what it might be.

Given the problems with Brink’s constitutional naturalism, it seems that his theory has not shown us how moral realism and naturalism can both be true. Thus, this particular response to the argument from queerness does not succeed, raising the question if a nonnaturalist ethics can rebut the argument from queerness. In this section, I shall show how constitutional nonnaturalism can be defended against the argument from queerness. Later in the chapter I shall show how a plausible identity version of ethical naturalism can be developed by a use of the essentialist method of linguistic analysis.

Before demonstrating how the nonnaturalist can refute the argument from queerness, I would first like to place the refutation of the argument from queerness in a larger context. A refutation of the argument from queerness can count as a sufficient reason for believing moral realism only if one first has a sound argument for (a) the thesis that commonsense moral beliefs imply moral realism and (b) an epistemic thesis that may be called the principle of veridical seeming (to be explained below).

(a) Regarding commonsense moral beliefs, Brink, Post, and other moral realists do not devote much of their philosophizing to arguing that common sense is committed to moral realism, but clearly such an argument is necessary if their position is to be tenable. Emotivists, prescriptivists, and constructivists such as Rawls and others have all argued that common sense is not committed to moral realism, and these arguments need to be addressed. I presented several arguments in part I for the thesis that first-order commonsense moral beliefs imply moral realism, and for this reason the thesis may be justifiably assumed in what follows. But this assumption is only one crucial assumption needed in the argument for moral realism.

(b) A second crucial assumption in the argument for moral realism is the conditional: if common sense is committed to moral realism, then it is rational to believe moral realism unless there is a compelling argument for moral antirealism. Mackie, Post, Brink, and others assume this conditional is true. But why should we believe it to be true? Why should what seems to be the case to us in our commonsense attitudes have any evidential force at all? If there is no compelling argument for moral antirealism and no compelling argument for moral realism, should we not remain agnostic on this issue, rather than abide by “what seems to be the case” to common sense? In the following I shall present an argument for this conditional.

I shall call the general principle of which this conditional is an instance the principle of veridical seeming, and the epistemic argument I shall present the argument from veridical seeming. Intuitively, the principle of veridical seeming is that we have reason to believe what seems to us to be true, unless there is good reason for us to distrust what seems to us to be the case.[6] The following argument can be shown to be sound:

1. Ordinary ethical sentences and commonsense first-level moral beliefs imply moral realism (or “Moral realism tacitly seems to be true in ordinary commonsense moral attitudes”).

2. There are no empirical or a priori reasons to believe that first-level moral beliefs are all false.

First, I must show that this argument is valid. A skeptical claim that (1) and (2) do not entail ( would be justified by the principle,

S. If p seems to be true, then, even in the absence of empirical and a priori reasons to not believe p, p’s seeming to be true does not justify a belief in p.

Now what could justify a belief in the skeptical principle (5)? If we grant there are no a priori or empirical reasons to not believe (S) (other than the one I am about to prove), then the skeptic cannot appeal to the fact that (S) appears true to her to justify her belief in (S). For (S) rules out the epistemic legitimacy of this very appeal.

The skeptic might then introduce some reason R that is a reason to believe (S). If so, then it appears to the skeptic that R is true and that R justifies (S). But by (S), what appears to the skeptic to be the case gives the skeptic no reason to believe that it is the case. The skeptic (to be consistent with [S]) cannot trust her own beliefs and thus cannot trust any of her beliefs about the justification of(S).

This shows that the argument (1)-(3) cannot be refuted. But can I also show that we are justified in believing the argument-form instantiated by (1)-(3) is valid? Yes, for this argument-form is self-justifying. The argument seems to be valid, there are no a priori or empirical reasons to believe it is invalid, and therefore it is justified to believe it is valid.

Given this circumstance, there is no reason to believe that (1) and (2) are defective justifications for the thesis (s). There is no undermining defeater for belief in (s). So we have this argument:

1. Ordinary ethical sentences and commonsense first-level moral beliefs imply moral realism (or “Moral realism tacitly seems to be true in ordinary commonsense moral attitudes”).

2. There are no empirical or a priori reasons to believe that first-level moral beliefs are all false.

4. There is no reason to believe that the conjunction of (1) and (z) is a defective reason to believe moral realism.

This argument is valid, but is it sound? The two crucial premises are (i) and (2). I defended (1) in part I of this book. Thus, it is incumbent upon me to say something in behalf of premise (2). If (2) is rendered plausible, then moral realism is rendered plausible.

First, there are no a priori analytic reasons to believe that moral realism is false, for “There are objective moral facts” and “It would be the case that animals ought not to be tortured even if no humans believed this” are not implicitly logically self- contradictory; it cannot be shown, by substitutions of synonyms for synonyms, that the relevant negations of these sentences are substitution instances of first- order predicate logic with identity.

Second, there are no synthetic a priori principles that justify belief in the falsity of moral realism (no such principle has ever been offered, at least that I am aware of, and I cannot think of any such principle).

The arguments usually offered are empirical arguments. The main one is the argument from queerness developed by Mackie and endorsed by others. (I shall discuss below the other main empirical argument, the argument from moral disagreement.) In the following I shall show that a defender of a nonnaturalist constitutional ethics can refute Mackie’s argument from queerness. (The third main empirical argument for moral antirealism, Harman’s argument from “the explanatory impotence of moral facts,” has been given plausible responses by Sturgeon, Sayre-McCord, and Brink, and I would refer the reader to their responses [see note 7 above]).

Mackie grants premise (i) of my argument for moral realism. He writes, “Belief in objective values is built into ordinary moral thought and language. . . . As such, [moral antirealism] needs arguments to support it against common sense.”[7] Mackie’s argument from queerness is in effect an attack on premise (2) of my argument; he wants to show there are empirical reasons to reject the moral realist thesis that is tacit in commonsense moral beliefs.

Mackie’s argument from queerness is specifically designed to counter nonnaturalist theories of moral realism. (This is why the easiest response to his argument is to argue for a naturalistic moral realism, which I shall do in later sections. But here I shall defend moral realism in general by showing that

Mackie’s argument is ineffective even against a nonnaturalist moral realism.) One prong of his argument concerns the connection between natural properties and nonnatural properties. He writes, “A way of bringing out this queerness is to ask, about anything that is supposed to have some objective moral quality, how this is linked with its natural features” (41). Mackie notes that the connection cannot be one of entailment, logical necessity, or semantical necessity. But the constitutional nonnaturalist has a ready response to this claim: the connection is synthetic, metaphysically necessary, and a priori. The constitutional nonnaturalist will say we may plausibly consider the relation between moral and natural properties as a relation between properties and properties of those properties; that is, moral properties are second-order properties, properties of natural properties. For example, goodness is a property of the property being scientifically knowledgeable. The nonnaturalist may analyze “good ness” as meaning “ought to be exemplified,” so that “scientific knowledge is good” means “scientific knowledge ought to be exemplified.” It is not an analytic but a synthetic truth that scientific knowledge is good. And this truth is necessary and a priori. This synthetically necessary a priori relation is not one that Mackie even considered. Why not?

It seems that what lies behind Mackie’s assumptions about queerness is a commitment to an extensional philosophy of language and a physicalist ontology. Linguistic essentialism, however, contains a battery of arguments for an intensional philosophy of language and for such abstract entities as possible worlds—I reviewed some of these in chapter 4. If our ontology is committed to such abstract entities as the possibility of there being Fs, then nonnatural moral facts such as there ought to be Fs are not queer. Indeed, if we reject Mackie’s presupposition of physicalism and extensionalism, then the argument from queerness is a nonstarter.

Given the enormous influence of Mackie’s argument, it is worth showing in detail that it is a blatant petitio principii.

Why does Mackie not consider the possibility that the relation between moral facts and natural facts is a relation of metaphysical necessity? Because he rules out such relations in advance. He writes, “If some supposed metaphysical necessities or essences resist such treatment [ is, reductive analysis in empiricist terms], then they too should be included, along with objective values, among the targets of the argument for queerness” (39). Now, as we have seen, Mackie can conclude that moral values are queer only by assuming there are no metaphysically necessary connections. In the passage just quoted, however, he asserts that the argument from queerness is also supposed to lead to the conclusion that there are no metaphysical necessities. The thesis that there are no metaphysical necessities is both an assumption and a conclusion of his argument from queerness. This vicious circularity vitiates any force his argument might have against moral realism.

Further, I should note that a premise of any argument for the queerness of metaphysical necessities would be that all familiar necessities are logical or semantic (analytic), and thus that metaphysically necessary connections are queer. If this particular argument is not to be a blatant petitio principii, however, Mackie would have to take on the entire tradition of linguistic essentialism from Marcus onward, which has provided numerous arguments that metaphysical necessities are quite familiar, both to common sense and to science. But this is not a task Mackie attempts to undertake.

Another part of Mackie’s argument from queerness is his claim that “a faculty of intuition” is required if we are to apprehend objective values. The thesis that there is a “faculty of moral intuition” is one of the most famous strawmen in twentieth- century philosophy. Even the so-called intuitionists— Moore , Prichard , Russell, Ross, Ewing , and (perhaps) Butchvarov—did not postulate a faculty of intuition. Only Richard Price did in the eighteenth century. Mackie tries to make a great mystery out of how nonnatural moral values could be apprehended, but I suggest the mystery disappears once we eliminate the “perception” model of apprehending moral values. If moral values are nonnatural, they are apprehended analogously to how we come to have beliefs about the motives of people’s actions. We can perceive people’s behavior but not their motives. And yet we are constantly forming beliefs about their motives. We do this by tacitly inducing their motives from their perceived behavior. Likewise, when we come to believe that an action is wrong, we do not perceive its wrongness but come to believe it is wrong by inferring from the behavior we do perceive. (The inference relation is here a synthetically necessary relation.) In some cases, we ascribe wrongness to an action only if we also make a tacit inference about the motives of the act; in these cases, the acquisition of the belief about the moral value is relevantly similar to the acquisition of the belief about the motive. However, whereas we induce (in the sense of a tacit inference to the best explanation) from the behavior to the motive, we infer (grasping a synthetically necessary connection) the moral value from our inductively acquired beliefs about the motives, the action, and the action’s consequences.

Furthermore, the apprehension of nonnatural moral values would be a special case of an apprehension of abstract objects, such as sets, numbers, universals, propositions, possibilities, and so forth. The apprehension of moral values can be seen to be queer only if we beg the question by assuming physicalism and nominal ism—for example, that all we apprehend we apprehend by means of being causally affected by a physical object. The apprehension of moral values is queer only if reductive physicalism and nominalism are true. But reductive physicalism and nominalism have not been demonstrated to be true, and certainly not by Mackie.

Further, nonreductive physicalism can allow for nonnatural moral values, as Post has plausibly argued, and most physicalists today accept a nonreductive version.[8] Mackie’s additional argument is that objective moral values would be intrinsically motivating and action guiding, which is queer. “Something’s being good both tells the person who knows this to pursue it and makes him pursue it” (40). Mackie’s argument here runs together a host of distinctions, as Brink has shown in chapter 3 of his book Moral Realism and the Foundation of Ethics. Brink plausibly argues that there is an external connection between moral values and motives for acting. It does not analytically belong to the concept of a moral value that an apprehended moral value is motivating; more exactly, the concept C₁ expressed by “moral value” does not include the concept C₂ expressed by “The belief that something possesses a moral value is a sufficient motive for acting.” If the moral value is motivating, that depends on such external considerations as the nature of the agent, the content of the moral theory, and so on. Further, the existence or possibility of existence of sociopaths shows it is possible to understand moral values intellectually and yet not find them motives for acting (this has been plausibly argued by Pritchard, On Becoming Responsible).[9] Even if, as I suggested in chapter 3, some moral statements do entail imperatives, it does not follow that understanding and believing these moral statements will motivate one to act on the imperative (I can believe “One ought to go to work” and understand that it implies “Go to work!” and yet feel unmotivated to act in accordance with the imperative.) Moreover, many moral statements do not imply imperatives: for example, “It is morally permissible to become a chemist” does not entail any imperative.

In my argument for moral realism, I noted that we are justified in believing moral realism unless there are a priori or empirical reasons to think otherwise. I have argued that there are no a priori reasons to think moral realism is false and that the main empirical argument, the argument from queerness, is a petitio principii and in any case is directed against only one version of moral realism, the nonnaturalist version. Another main empirical argument against moral realism, one that I discussed in part in chapter a in regard to Moore ’s ethics, is the argument from moral disagreement. The conclusion I drew in chapter 1 is that Moore ’s principles entailed that human life is meaningful but absurd; if there are objective moral facts, there are no possible means of knowing them. The argument from moral disagreement as it is normally understood, however, is an argument that human life is objectively meaningless, that is, that there are no objective moral facts. This argument, of course, is not that there is moral disagreement, but that there seems to be no possibility in principle of resolving moral disagreement. The “argument from moral disagreement” is that the best explanation of this fact is that there no objective moral values and therefore no fact of the matter as to which mo beliefs are true.

In response to this argument, I want first to make some con about extant moral disagreements. First, there is not widespread moral disagreement, contrary to what philosophers and nonphilosophers alike often maintain. Virtually all people in virtually all cultures agree that murder, rape, stealing, lying, torturing children for fun are wrong. Virtually all humans agree about basic moral principles, such as that it is wrong to cause suffering needlessly, that love is morally better than hate, that wisdom is better than ignorance, and that it is good to heal the sick or injured, and so forth. Cases of moral disagreement are in fact isolated, even though they receive much attention (for example, the currently discussed issues pertain to abortion, death penalty, affirmative action, and so on). Indeed, there is more agreement in morals than in many other recognizably fact-stating disciplines. Philosophy of religion is fact stating; it is a fact that there is a god or it is a fact that there is no god. But fundamental disagreements about religious issues are more persuasive than those about morals. If you randomly take one thousand people on the earth you will probably find they all disagree about fundamental religious issues but all agree that it is wrong to cause suffering needlessly, wrong to steal, and so on. The same holds for psychology, economics, and other areas; there is more agreement about moral principles than about the psychological structure of the human personality or about how the economy works.

Another reason philosophers tend to overestimate the extent of moral disagreement is that they do not take into account the hundreds or thousands of “trivial and tacit” moral judgments we make every day, for example, that it is wrong to punch each person who walks by you in the nose, that it is right to behave in an amiable way to guests to whom one is introduced at a party, that it is wrong to drive one’s car over one’s neighbors’ flower gardens, and so forth. If we add up these thousands of cases of tacit moral agreements that occur every day, they will outweigh extant moral disagreements by a very large ratio.

Second, virtually all instances of moral disagreement are about the relevant nonnormative facts, the relevant religious, natural, or social scientific facts. For example, the moral debate about abortion is primarily about the nonnormative issue of whether a fetus has a divinely implanted soul, which is a debate about factual religious claims. Disagreements we have with the Nazis largely concern the empirically false beliefs they held about Jews and Aryans; our disagreements with the nineteenth—century American slave owners primarily concern their empirically false beliefs about African Americans, and so on.

Where, then, do we find the flaunted “moral disagreements” supposed to constitute such a problem for moral realism? Butchvarov has made a good case in Skepticism in Ethics that it is intrinsically plausible that friendship, knowledge, pleasure, health, and existence are good and that enmity, ignorance, pain, sickness, and death are bad. There is virtually universal agreement about such noncompara tive moral judgments, that is, judgments about what general kinds of things are good and what bad. But, as I suggested in chapter 1, the main cases of disagreement involve comparative judgments. Is friendship more valuable than knowledge, as Butchvarov maintains, or are Plato, Aristotle, Spinoza, and Hegel correct that knowledge is more valuable than friendship? Is pleasure more valuable than health or vice versa? There does seem to be widespread disagreement about the hierarchy of values. The argument from moral disagreement is best stated as an argument about widespread and seemingly intractable disagreements about comparative moral judgments.

In chapter 1 I suggested that the argument from moral disagreement would show (in the context of Moore ’s logical realist ethics) that moral realism is true (that human life has an ethical meaning) but that we cannot know which comparative moral judgments are true (and thus that human life is both meaningful and absurd). But outside the confines of Moore ’s logical realism, there is an answer to this argument that is consistent with human life being meaningful and not absurd. The seemingly intractable disagreements about comparative moral judgments are (in principle) a correctable epistemic feature of human moral life. The disagreements are due to a failure to realize that the relevant comparisons are made at too general a level for the comparison to have a truth-value. “Knowledge is more valuable than friendship” does not have a truth-value, but “Knowing the local phone directory by heart is morally better than having any friends at all” has the truth value of false, and “Developing the special and general theories of relativity is more valuable than becoming casual friends for one day with one’s neighbor’s visiting cousin” has the truth-value of true. The same holds for “Pleasure is more valuable than health” and “Freedom is more valuable than love” and other such generalizations. There is no reason to think that just because we can construct ethical sentences of any degree of generality that they therefore all have truth-values.

To be convincing, the limited argument from disagreement should be an inductive argument that there is no agreed-upon method for resolving moral disagreements and therefore that it is likely there is no knowable method for resolving these disagreements. The further conclusion will then be that the best explanation of there being no knowable method for resolving disagreements is that there is nothing to be known, no objective comparative moral facts.

But this argument also fails because there is no agreed-upon method for resolving disputes in metaphysics, epistemology, religion, psychology, and so on, and this does not imply that it is likely that there is no knowable method for resolving disputes in these areas. The fact that there is no agreed—upon method for resolving disputes in these and other areas is best explained by the hypothesis that (i) there is a knowable method for resolving disagreements, but (2) the complexity and difficulty of the subject matter, combined with the limited epistemic abilities of the human species, have so far prevented humans from both learning and reaching agreement about this method.

In addition to these responses to the argument from moral disagreement, we may endorse some of the ideas that Brink presents. Brink addresses the argument from moral disagreement by noting that realism about a discipline does not require that all actual cognizers eventually reach agreement; rather, it is reasonable to expect agreement on all facts only if all cognizers (a) are fully informed, (b) are fully rational, and (c) have sufficient time for deliberation. In cases of moral disputes, very few people (if any) meet these three conditions. For example, although most people hold moral views about domestic and foreign policy political issues, hardly anybody is fully informed about the issues or is fully rational in his thinking about the issues or has sufficient time to deliberate about them. The explanation of moral disagreement by postulating that there are no moral facts would work only if there is pervasive disagreement even after the cognizers met these three conditions; but because these conditions are rarely, if ever, met, there is no reason to think that there is pervasive disagreement of this sort.

For these various reasons, the argument from moral disagreement fails to count as an empirical reason to defeat the justification for moral realism that is provided by our first-level commitment to moral realism. Because the other main antirealist argument, that from queerness, also fails, we have reasonably strong grounds for believing that human life has an objective ethical meaning.

But the grounds for this belief could be made stronger if we could show that ethical naturalism is true. In my response to the argument from queerness, I indicated how a constitutional nonnaturalism could be defended against it. But this does not imply that the preferred metaethical theory for the moral realist is non- naturalism. Moral realism has stronger epistemic weight if it is combined with ethical naturalism. Some good informal and intuitive reasons for believing this are as follows.

Compare the conjunction moral realism and nonnaturalism with the conjunction moral realism and naturalism. Virtually all philosophers hold that some natural properties can be exemplified by some natural items independently of whether humans or other finite minds believe these items are exemplified. Call this position “minimal naturalism.” Minimal naturalism is common to virtually every meta physical theory (be it Platonic realism, Aristotelian realism, nominalism, scientific realism, physicalism, theism, phenomenalism, idealism). For example, even phenomenalists or idealists would grant that the natural property, being a sense-datum and not a mind-independent physical thing, is exemplified b some sense-datum even if the person experiencing it does not happen to hold the belief that this property is exemplified. If moral realism’s metaphysical commitmer are merely to this mini mal realism, then the relevant metaphysical aspect of moral realism will be uncontentious. But if moral realism’s metaphysical commitments are also to the thesis that there are objective nonnatural facts, then the metaphysical aspect of moral realism will be highly contentious and will have considerably less prima facie plausibility than the weaker metaphysical theory of minimal naturalism. This suggests that, all else being equal, the conjunction of ethical naturalism and moral realism has higher epistemic status than the conjunction of ethical nonnaturalism and moral realism. Of course, the nonnaturalist will deny that all else is equal because she will argue that ethical naturalism is false, perhaps even logically in coherent. Nonetheless, it is true that f a logically coherent and plausible ethical naturalism can be developed, then this form of moral realism, by virtue of having a metaphysical aspect that is uncontentious, is epistemically preferable.

This point may also be put by saying that the naturalist moral realist has one less position for which she needs to argue: The moral realist who is a nonnaturalist will not only have to argue, within the field of ethics, for her ethical theory, but will also have to take on the metaphysical task of showing that there are objective nonnatural facts in addition to natural facts. But because the ethical naturalist can take for granted that there are objective natural facts, she does not need to argue for her metaphysical theory and can confine herself to arguing for her ethical theory.

It seems to me that there is a plausible version of ethical naturalism, an identity naturalism rather than a constitutional naturalism, and thus that moral realism should be combined with identity naturalism rather than with the constitutional nonnaturalism I articulated in sections 23.2 and 23.3. This identity naturalism is what I shall call global, naturalist perfectionism. The remainder of this chapter is devoted to developing this perfectionist theory. The theory is partly based on Hurka’s perfectionist ethics and will emerge as a criticism of the nonnaturalistic formulation he gives to perfectionism.

[1] Hilary Putnam, Philosophical Papers, Volume Two ( Cambridge : Cambridge University Press,

0975), p. 290.

[2] John Mackie, Ethics ( London : Penguin Books, 0977).

[3] John Post, The Faces of Existence. An Essay in Nonreductive Metaphysics ( Ithaca : Cornell Uni

versity Press, 1987), 255.

[4] Brink, Moral Realism, 23.

[5] Panayot Butchvaro Skepticism in Ethics ( Bloomington : Indiana University Press, 0989); Nicholas Sturgeon, “Harman on Moral Explanation of Natural Facts,” in Moral Realism, ed. Gillespie, 1986 (The Southern Journal of Philosophy, Supplement 24); Richard Boyd, “How to Be a Moral Realist,” in Essays on Moral Realism, ed. G. Sayre-McCord ( Ithaca : Cornell University Press, 0988); Sayre-McCord, ibid.

[6] G. E. Moore, Principia Ethica, 16-17.

[7] I discuss this principle in The Felt Meanings of the World: A Metaphysics of Feeling ( West Lafayette : Purdue University Press, 0986), 131-134, and in Language and Time (New York: Oxford University Press, 1993), 14-18.

[8] Mackie, Ethics, 48-49.

[9] Post, The Faces of Existence, see, e.g., 281.