CHAPTER ONE
Naturalism and analytic moral philosophy from Moore to Hare
âLooking back . . . one may be surprised and a little sad, that this particular conflict, about âfact and valueâ, has occupied so much of our time. We seem to have rushed on to the field without waiting to map the territory supposedly in dispute, ready to die for some thesis about commendation or approval, about pro-attitudes or evaluation before anyone had done much detailed work on the specific, and very different, concepts involved.â (TE 12)
In this chapter, I will discuss the philosophical context of Philippa Footâs early writings. In her early work, she argues that moral judgments characterize persons or actions as good, bad, right, or wrong in view of the impact of the personâs dispositions or actions on human well-being. In taking this approach, Foot advocates a form of naturalism, which is the view that moral judgments describe facts about the world that we observe through our senses. In Footâs view, moral judgments are a distinctive class of judgments because of their subject matter, and that subject matter is the world described by the natural sciences. Footâs naturalism conflicts with the view that was dominant at the timeânoncogntivism. For noncognitivists, moral judgments are distinguished from other judgments, not by their subject matter, but because they stand in an essential connection to attitudes of approval or disapproval and in that they are used in distinctive ways, for example, to bring about similar attitudes in others. These views, for which I will use the standard name ânoncognitivism,â in turn emerged from a response to the moral philosophy of G. E. Moore. Moore argued stridently against all forms of naturalism. Yet, like Foot, he thought that moral judgments are about a distinct subject matter. For Moore, moral judgments are about what will bring about the most good, using the word âgoodâ in a distinctive sense. As Moore is often read, âgoodâ in this special sense designates a nonnatural property belonging to a Platonic realm that, though, separate from the natural world, still somehow pertains to whatever we call âgood.â As we will see, although Mooreâs argument against naturalism was widely accepted by philosophers in the early to mid-twentieth century, his non-naturalism was seen as abhorrent because it invoked a mysterious non-natural property standing in an equally mysterious relation to the natural world.1 Noncognitivism is an attempt to avoid the alleged problems of naturalism without the supposed mysteries of Mooreâs non-naturalism.
Moore: Analysis against naturalism
Foot claims that G. E. Mooreâs moral philosophy set the stage for her early work (TE 1). Indeed, Mooreâs 1903 Principia Ethica is widely viewed as the starting point for all analytic moral philosophy.2 Mooreâs Principia is consequential for all subsequent analytic moral philosophy in two ways. First, methodologically, Moore establishes the possibility of treating questions of metaethics independently of normative issues.3 Much moral philosophy through the 50 years after Principia appeared focuses on metaethicsâspecifically on the question of the relation of facts and value. This agenda is not only the result of Mooreâs methodological innovation of separating metaethical issues from normative ethics, but also reflects the influence of his arguments within metaethics. Second, Mooreâs moral philosophy is consequential in that he made an argument against naturalism in ethics that many moral philosophers regard as conclusive. As we will see in what follows, Footâs moral philosophy is influenced by both of these aspects of Mooreâs views.
Moral philosophy before Moore had certainly offered views on metaethical issues; yet, Mooreâs approach to moral philosophy is pioneering in that it treats questions concerning the meaning of âgoodâ separately from questions about which things are good. By separating the questions âWhat does âgoodâ mean?â and âWhich things are good?â Moore launched analytic moral philosophy and, for better or worse, split the enterprise of moral philosophy into two sub-disciplines. As we will see, the separation of metaethics and normative ethics leads some to believe that metaethics could be pursued with neutrality toward the various values one might hold, and because of this apparent neutrality, some philosophers take metaethics to be a more scientific enterprise than normative ethics. For Moore, unlike those influenced by him, the task of figuring out the meaning of âgoodâ was âof the last [i.e., highest] importanceâ because failure to sort out such matters had been the âcause of the acceptance of false principles . . .â4 For him, then, metaethics is no neutral theoretical science, but is itself an endeavor with practical consequences.
In Mooreâs view, moral philosophers before him had overlooked a crucial question. While treating the question âWhich types of conduct are good?â they had overlooked the question of the meaning of the word âgoodâ in general. Hence, they tended to generalize, from the meaning of the word in the context of âgood conductâ to all contexts, with disastrous results. Without asking what âgoodâ means generally, they identified âgoodâ with one of the salient properties of good conduct, for example, producing pleasure. Moore thinks that this is analogous to the mistake one makes when identifying the meaning of âyellowâ with reflecting light waves of a certain wavelength.5 For though it may well be that all yellow things reflect light of certain wavelengths, that is not what âyellowâ means. Likewise, though all good actions may produce pleasure, âproducing pleasureâ is not the meaning of âgood.â Moore therefore proposes to askâwhat is the meaning of the word âgoodâ itself?
His answer to this question isââgoodâ cannot be defined. By this, Moore means to say that that âgoodâ cannot be shown to be synonymous with concepts that identify other properties, either natural or supernatural. âGood,â like âyellow,â is a simple notion.6 This claim certainly suggests that Moore identifies âgoodâ as a distinctive non-natural property, but one must determine first how one establishes the sweeping negative conclusion that no definition of âgoodâ is adequate. To establish this conclusion, he uses what is now called the âopen question argument.â This argument points out the difference between asking âThat may be pleasant, but is it pleasant?â which is nonsense, and âThat may be pleasant, but it is good?â which is clearly meaningful. One who defines âgoodâ as âpleasantâ commits himself to viewing these concepts as substitutable. Yet, the open question argument shows that they are not substitutable. The two questions just quoted differ in meaning. With the open question argument, then, Moore believes he has found a way to reject any proposed definition of the good, whether it invokes natural or supernatural properties. For one who identifies goodness with preserving the survival of the species, Moore can point out that it makes sense to ask of any species-preserving action, âThat may preserve the species, but is it good?â Likewise, for one who identifies goodness with conforming to the will of God, Moore says that it makes sense to ask of any case, âThat may be an act in conformity with the will of God, but is it good?â The answer to one of these questions may always be âyes, it is good,â so that being species-preserving or conforming to Godâs will is a reliable method of tracking what is good, but Mooreâs point stands, he thinks, by the mere fact that these questions are meaningful. They show that âgoodâ means something different from the names of these qualities.
Anyone who defines âgoodâ commits what Moore misleadingly calls the ânaturalistic fallacy.â This name is misleading in two waysâfirst, even those who define âgoodâ in supernatural terms commit the naturalistic fallacy, so it is not truly a naturalistic fallacy. Second, it is not, strictly speaking, a fallacy, as William Frankena showed in a justly famous paper. The perception of the openness of a question on any given run at the open question argument depends on a prior conviction that there are two distinct properties rather than one. In other words, anyone who is fully convinced that âgoodâ means âproducing pleasureâ will find the question âThat may be pleasant, but is it good?â to be nonsensical. Hence, the argument begs the question. There is only an errorâa fallacy if you likeâif two distinct properties are conflated, but that is precisely what is at issue. Yet, many readers of Moore share his conviction that the questions are indeed open. Further, as Stephen Darwall has pointed out, the apparent openness of the question may stem not so much from the necessary falsehood of any attempt at defining the good, but from the fact that goodness has a connection with action, which the proposed definitions do not have. Because we are self-reflective agents, Darwall argues, we can stand back and examine any proposed value; we can ask, ought we to value that?7 And this ability gives the open question argument its force. Although Frankena is correct that the naturalistic fallacy is not a fallacy, this is compatible with Darwallâs point that the open question argument has some force against naturalism. Still, it should be noted that in the 36 years between the publication of Principia Ethica and Frankenaâs article, Mooreâs argument against naturalism was widely perceived as knock-down.
As mentioned above, Moore appears to commit himself to the view that âgood,â in the sense that is at issue in moral judgments, refers to a non-natural property. This property is not perceivable by the senses or by its causal effects on the natural world; instead, it would be detected by intellectual intuition. Indeed, this is the interpretation of Moore that Foot adopts (TE 2). To many philosophers following Moore, this view involves two objectionable claims. First, the idea that there are some properties that are not part of the natural world is at odds with the prevailing empiricist epistemology. After all, to say that there is a property apart from the natural world is to say that there is something that does not have any causal interactions with the natural world, and so could not be understood in terms of empirically funded scientific laws. Second, the appeal to intuition seems to ignore the fact that people have different intuitions about what is good. Admitting those disagreements, it is not clear that there could be a method for resolving moral disputes rationally. Parties to a disagreement would seem to be left with thumping their chests and appealing to their separate intuitions.
Although this damning assessment of Moore is widely shared, it is not clear that it is based on a charitable interpretation of Mooreâs claims. As Robert Shaver argues, it is possible to read Moore as denying that goodness is a natural concept, but not denying that it is a natural property.8 That is, the concept of goodness describes a feature of the natural world that is not accessible to the natural sciences. Goodness cannot be analyzed in terms of any one natural property, according to Moore, but that does not mean that goodness is not natural. This interpretation of Mooreâs view turns on distinguishing an analytic connection between goodness and a natural property, on the one hand, and a constitutive connection between goodness and natural properties, on the other hand. In this view, any natural property constitutive of goodness in a given case may not always be sufficient to make something good, and such a property may not be the only property that makes something good. Hence, in Shaverâs interpretation, non-naturalism rejects the idea that moral matters can be settled scientifically, but does not posit a property or class of properties that is metaphysically irreconcilable with a commitment to scientific naturalism.
As for the epistemological problems with Mooreâs view, Shaver argues that although Moore leaves himself open to these charges, these problems are not inherent to Mooreâs position. The non-naturalist can appeal to precisely the same justifications that naturalists appeal toâmost importantly, systematizing common sense moral judgments.9 That is, the non-naturalistâs appeals to intuition are not necessarily infallible, and for any intuitively plausible instance of something that is deemed good, we can ask whether that claim can be squared with other intuitions about what is good. Overall, then, Mooreâs non-naturalism is probably more defensible than has been generally acknowledged by those who rejected it, including Foot.
Ayer and Stevenson: Emotivism
In the widely received version of Mooreâs views, goodness is a property of things, albeit a property of a special kind. It is like âyellowâ in being simple and indefinable. Yet, unlike yellow, it is not accessible through our senses. For those who find Mooreâs open question argument compelling, but who cannot accept the implications of his non-naturalism, the only solution appears to be this alternative responseâto deny that goodness names any sort of property at all. They propose that we should seek to explain moral judgments by identifying a characteristic function they perform, rather than a specific class of property that they designate. Much moral philosophy understands moral judgments to have a descriptive function. In such a view, some moral judgments describe a thing as featuring the property of goodness. Yet, clearly not all meaningful language describes the world; for example, we use language to express feelings (e.g., âHow wonderful!â) and to command (e.g., âOpen the door!â). These other functions of language do not ascribe properties. In the former case, the words express a positive feeling or attitude toward something. Similarly, the latter does not ascribe a property, but rather, commands someone to do something. By exploiting these disparate functions of language, some moral philosophers hope to avoid problems with descriptive theories of moral judgment, such as the difficulties with Mooreâs view under its standard interpretation.
For these moral philosophers, working around 30 years after Mooreâs Principia, moral language serves a purpose that combines an expressive and a commanding function. A. J. Ayerâs 1936 Language, Truth and Logic contained an early, though brief exposition of this view of moral judgments. He writes:
. . . if I say to someone, âYou acted wrongly in stealing that money,â I am not stating anything more than if I had simply said, âYou stole that money.â In adding that this action is wrong I am not making any further statement about it. I am simply evincing my moral disapproval of it.10
We see in this passage that there is, in Ayerâs view, a descriptive component to a moral judgment (in this caseââyou stole the moneyâ). However, such a descriptive statement is clearly not a moral judgment. What distinguishes a moral judgment from an ordinary descriptive statement is the expression of an attitude toward the descriptive content. Further, at least some moral judgments convey commands. According to Ayer, the sentence, âIt is your duty to tell the truthâ is equivalent to âTell the truth!â11 The view expounded in Ayerâs treatise came to be known as emotivism because it defined moral judgments in terms of a function of expressing feelings or attitudes. Note that the moral language does not describe feelings that the speaker has; it expresses them. ContrastââI approve of xâ with âhooray for x!â The former describes my attitude; the latter expresses it.
C. L. Stevenson, in his 1944 Ethics and Language and a number of papers that appeared between 1937 and 1962, developed the idea in Ayerâs brief discussion of emotivism into a complete metaethical theory. For Stevenson, moral terms are used in two central ways. In the first use, terms like âgoodâ are used to say âI approve of x. Do so as well.â Stevenson is careful to point out that this translation of âgoodâ is at best crude, and for reasons to be discussed, he came to think of all such translations as seriously misleading.12 Nevertheless, the translation captures, according to Stevenson, something of what he calls the emotive meaning of moral terms, which he defines as, âthe power that the word acquires, on account of its history in emotional situations, to evoke or directly express attitudes, as distinct from describing or designating them.â13 So, in Stevensonâs view, âgoodâ is, in part, supposed to express my positive attitude toward something. Unfortunately, âI approve of xâ does not quite capture that. Instead, it expresses my belief that I have a positive attitude toward something, and thus the two claims must have quite a different meaning, as Stevenson later came to realize. After all, âI approve of xâ is a descriptive claim about my psychology, and there is a specific type of evidence that would support or undermine that claim. This is quite unlike an utterance that actually expresses my approval of x, such as âHooray for x!â Since that exclamation is purely expressive, we cannot gather evidence for or against it. For utterances that express attitudes, the whole idea of evidence seems out of place. Likewise, the second part of the translation, âdo so as well,â suggests that saying âx is goodâ has an imperatival function, but Stevenson was wary of this claim even as he stated it. After all, an actual command to adopt an attitude is likely to backfire. Something subtler is going on with âgood,â as Stevenson readily admitted. To say that something is good, then, suggests to another that she adopt the speakerâs positive attitude as well, and it is to exert a causal, rather than rational effect on its hearer. In exuding my posi...