CHAPTER 1
CONTEMPORARY MORAL REALISM:
PROBLEMS WITH A COMMON ASSUMPTION
Contemporary versions of moral realism assume that the term âgoodâ refers to a property that falls within the scope of Kripkean and Putnamian developments in philosophy of language and metaphysics (hereafter I will use the less cumbersome KP developments). Put briefly, KP developments refer to a cluster of theses on the anti-descriptivist side of the meaning and referring divide. These theses include the claim that it is possible for certain termsâmost notably proper namesâto refer without descriptive content. Thus, the notion of a rigid designatorâa term that refers to the same thing in every circumstanceâis introduced, and the term âgoodâ is viewed by many as being a plausible candidate for being a rigid designator. Various metaethical theories within the moral realism camp were then developed that took advantage of these insights. But these versions of moral realism face challenges strong enough to warrant an investigation into an alternative account of moral realism.
In order to defend these claims, I first explain the account of moral realism common to contemporary moral realists. Contemporary accounts of moral realism have developed in large part as a response to G. E. Mooreâs Open Question Argument (OQA). The OQA assumes that goodness is a property. Thus, moral realists who share this assumption must respond to the OQA by denying one of its premises. Two representative versions of moral realism, which both assume that goodness is a property that falls within the scope of KP developments and attempt to deny one of the OQAâs other premises, are Richard Boydâs moral naturalism and Robert Adamsâs moral supernaturalism.1 In the second section of this chapter I explain the version of naturalistic moral realism represented by Boyd. I argue that Boydâs moral realism is not plausible because of its reliance on KP styles of analysis. In the third and final section, I explain the version of supernaturalistic moral realism presented by Adams. I argue that Adamsâs version likely fails for reasons that he himself worries about. Thus, contemporary versions of moral realism are either in need of serious repair or should be abandoned. I will leave the repairing to their advocates. In subsequent chapters, I lay the foundations for a new version of moral realism that avoids the problems presented here.
1 Moral Realism: The Contemporary Account
In this section I argue that standard formulations of moral realism assume that âgoodâ refers to a property susceptible to KP developments. More generally, standard forms of moral realism imply that if statements of the form âx is goodâ are true, goodness must be a property, that is, at least capable of figuring into synthetic a posteriori identity statements. With this purpose in mind, I limit my discussion of moral realism to two recent accounts.
In his introduction to the Oxford Handbook of Ethical Theory, David Copp argues that moral realism consists of the following five doctrines: â(1) There are moral properties (and relations). . . . (2) Some moral properties are instantiated. . . . (3) Moral predicates are used to ascribe moral properties. . . . (4) [M]oral assertions express moral beliefs. . . . (5) The moral properties, in that they are properties, have the metaphysical status that any other property has, whatever that status is.â2
Coppâs account of moral realism includes metaphysical and semantic claims. The metaphysical part of the account states that moral properties are on a par with other non-moral properties and that some things have moral properties. In a footnote Copp writes, â. . . [T]he first realist doctrine is to be interpreted such that the term âproperty,â as it occurs there, ascribes the same metaphysical status to moral properties, such as wrongness, as it ascribes to a non-moral property such as redness when it is predicated of such a property.â3 Note that Coppâs account is perfectly compatible with versions of moral realism that take the property goodness to be identical to some other property, where the identity is either analytic or synthetic.4
The semantic component of Coppâs account of moral realism is comprised by points (3) and (4). With the assumption that moral beliefs are truth-apt, point (4) commits the Coppian moral realist to cognitivism. Cognitivism does not entail moral realismâin Coppâs senseâbut moral realismâagain in Coppâs senseâdoes entail cognitivism.
Coppâs account of moral realism implies that moral realists must countenance moral properties and that moral properties are no different, in terms of their metaphysical status, than non-moral properties. Since, there are non-moral properties that are susceptible to KP developments, Copp leaves it open that moral properties may be susceptible to such developments as well. Copp aims to provide a general account of the commitments of moral realism, and thus leaves room for moral realists to take advantage of the semantic and metaphysical developments introduced by Kripke and Putnam. Whatever its merits or demerits, Coppâs account is evidence in favor of my claim that contemporary versions of moral realism build in a commitment to moral properties that are capable of receiving a KP analysis.
Geoffrey Sayre-McCord gives a different definition of moral realism:
Sayre-McCordâs characterization differs from Coppâs in some crucial respects. For our purposes, the most important difference is that Sayre-McCordâs account does not mention moral properties. The omission of moral properties from his account suggests that one could be a moral realist and deny the existence of moral properties (but not moral facts). Nevertheless, appearances are misleading.
Sayre-McCord explains moral facts in terms of moral properties. He provides two accounts of moral facts. According to the first, â. . . moral thought and talk [is] committed to properties, and facts, and truths, that could just as well be expressed in nonmoral termsâ.6 The first account goes further and claims that moral properties are identical with either natural or supernatural properties depending on who is articulating the view. The identity between moral properties and non-moral ones is thought to be knowable through analysis of moral concepts. Mooreâs OQA challenged the idea that an analysis of moral concepts reveals an identity between moral properties and non-moral ones. Thus, assuming Moore is right, if moral properties are identical to non-moral ones, mere analysis of moral concepts is not sufficient to reveal the identity.
Sayre-McCordâs second account of moral realism is an attempt to bypass the OQA by showing that moral properties can be identical with non-moral ones even though the identity cannot be known via conceptual analysis. This account of moral realism relies on Saul Kripkeâs pioneering work in 1972.7 Kripke introduced a new way of discovering property identities. Most assumed that the only way to discover property identities was via conceptual analysis. Thus, a claim that properties F and G are identical was, if true, thought to be both necessary and analytic. Kripke argued that many property identities are not knowable a priori and hence not knowable via conceptual analysis. This new wave in philosophy of language and metaphysics opened the door for a new response to Mooreâs OQA. Moral properties could be identical with non-moral properties without there having to be synonymy between the moral predicate and the non-moral predicate that refer to these properties. In Sayre-McCordâs view:
Despite the fact that Sayre-McCordâs characterization of moral realism does not mention moral properties, it is clear from his exposition that moral facts include moral properties. Indeed, as is clear from the above quotation, Sayre-McCordâs account of moral realism explicitly includes reference to moral properties that are capable of KP analysis.
Standard accounts of moral realism include, either explicitly or implicitly, reference to moral properties that are capable of figuring in analytic or synthetic property identities. In particular, most moral realists are committed to the claim that goodness is a property that is either analytically or synthetically identical to some other property. Consequently, the moral anti-realist need only show that each project fails in order to establish her/his position.9 If the anti-realist can show that moral properties are neither analytically equivalent to non-moral properties, nor synthetically equivalent to non-moral properties, then moral realism is false, given the assumption that these are the only two options.
The chief aim of this book is to argue that the standard account of moral realism implicitly assumes a false dichotomy (or trichotomy if we insert moral non-naturalism back into the list of options for moral realists). There is another option for moral realists besides the analytic and synthetic ones currently offered. Before providing the details of this third option, however, it is important to see some of the shortcomings of the analytic and synthetic accounts.
2 From Mooreâs OQA to A Posteriori Moral Naturalism
In order to be a moral realist one must be a naturalist, non-naturalist, or a supernaturalist. If G. E. Moore is correct, moral naturalism is a non-starter. Mooreâs OQA aimed to show that no naturalistic definition of a moral term is possible. Here is a reconstruction of Mooreâs main argument: Either goodness is simple or goodness is complex.10 If goodness is complex, then analysis should reveal its constituents. But analysis does not reveal the constituents of goodness. Hence, goodness is not complex. The OQA is Mooreâs defense of the minor premise. Any identification of the property goodness with some other property (i.e. goodness = F) will always leave it an open question whether the other property is indeed good. But since property identifications are a matter of analysis, and if goodness is in fact identical with F, the question âis F good?â should never be open. Since it is open, there is no property that goodness is identical with, and hence goodness is not complex. Since analytic moral naturalism and analytic moral supernaturalism, which claim that the definition of âgoodâ is both analytic and purely descriptive, appeared to be the only options for the moral naturalist and supernaturalist, these versions of moral realism looked doomed. Hence, if one opts for moral realism, and the OQA is sound, then one must be a non-naturalist. But non-naturalism falls prey to a host of objections that have convinced most that non-naturalistic moral realism is plainly false.11 Hence, given the soundness of the OQA, moral realism is false.
Through the work of Saul Kripke and Hilary Putnam, a plausible semantics for synthetic property identities was developed.12 Moral naturalists and supernaturalists attempted to appropriate this work in order to show that the nature of goodness is entirely natural or supernatural but cannot be known a priori. Hence, it is possible, according to their theory, to provide a definition of âgoodâ that is purely descriptive but not analytic. If this appropriation is successful, a version of moral naturalism emerges that avoids the OQA. To date, Richard Boydâs âHow to be a Moral Realistâ is the most discussed account of a semantics for âgoodâ that attempts to show that like âtiger,â âwater,â and other natural kind terms, âgoodâ may refer to a property or kind that can be synthetically identified with a natural property.13 In the next section I present a novel argument that Boydâs account fails.14
2.1 Mooreâs OQA, Boydâs Response, and Trouble
The motivation behind moral realistsâ turn toward the work of Kripke and Putnam comes primarily from G. E. Mooreâs OQA. By allegedly showing that neither moral naturalism nor moral supernaturalism is correct, Mooreâs commitment to moral realism forced him to embrace a version of moral non-naturalism. But Mooreâs conclusion that goodness is simple, unanalyzable, and knowable only through intuition has left most moral realists unsatisfied.15
In the early 1970âs Kripke and Putnam put forth a new semantics that explained how synthetic property identities are possible. Kripke and Putnam supported their semantics by arguing against the then prevalent descriptivism. One of the assumptions of descriptivism is that necessary truths and a priori truths are co-extensive. The identity of water and H2O appears to show that it is possible for at least some statementsâthat is, water is H2Oâto be necessarily true but not knowable a priori. This advance opened the way for those moral realists not willing to embrace non-naturalism to respond to the OQA.
If synthetic property identities are possible with respect to natural kinds and if âgoodâ is a natural kind term (or something like one), it may be that the proposition expressed by âgood = Fâ is necessary but not knowable a priori. Thus, the OQA is unsound, since one of its assumptionsâthat property identifications are a matter of analysisâis false. Richard Boydâs semantics for âgoodâ argues just that. According to Boyd: