It is my pleasure to introduce our next Featured Philosopher, John Deigh. John is a Professor of Philosophy and Law at the University of Texas, Austin, and he is widely known for his insightful work in moral psychology, the history of philosophy, and for his valuable work as the editor of Ethics from 1997-2008. Please feel free to share your comments or questions!
I am grateful for the opportunity to share with the PEA Soup community some ideas about the history of meta-ethics in the twentieth century that I’ve been working out recently. These ideas are part of a larger project that began with my chapter, “Ethics in the Analytic Tradition”, in the Oxford Handbook of the History of Ethics (R. Crisp, ed.). That chapter gives the history of analytic ethics during roughly its first fifty years, from G. E. Moore to R. M. Hare and Stephen Toulmin. The history treats ethics as a field of philosophy many of whose movements and changes have come about as a result of movements and changes in other fields like metaphysics and the philosophy of language. For example, I explain the radical impact of Moore’s Principia Ethica on Anglo-American ethics as continuous with the revolution in British philosophy that Moore and Russell ignited through their attacks at the turn of the 20th century on British Idealism. These attacks and the positive constructions to which they gave rise constituted the beginnings of the analytic movement in philosophy. The first chapter of Moore’s Principia, I maintain, should therefore be read as one of the major contributions to the beginnings of this movement and not, contrary to current fashion, as a rhetorically powerful recycling of ideas from Sidgwick.
In this vein, I would argue that meta-ethics emerged as a distinct subfield of ethics as a result of the linguistic turn that analytic philosophy took under the influence of Wittgenstein’s Tractatus. Carnap, on whom the Tractatus had an immediate and powerful influence, proposed in Logische Syntax der Sprache to replace philosophy, as traditionally practiced, with the study of the logic of science. To conduct such a study required constructing a formal language suitable for expressing scientific knowledge as found in all branches of empirical science. One then studied the logic of science by studying the syntax of this language. The formal language one constructs is thus the object of the study and is, accordingly, known as the object language. The language in which one conducts the study is known as the meta-language. The object language contains both variables and constants, and how one interprets the constants provides the language’s semantics. To facilitate his proposal to replace philosophy with the study of the logic of science Carnap restricted the interpretation of the constants to scientific terms and terms used to express empirical observations, which is to say, to the words and phrases of a natural language that are essential to the various branches of empirical science. The study was thus Carnap’s program for logical positivism. He conceived of it as extending to philosophy the program for studying the question of consistency in arithmetic that Hilbert devised under the heading of metamathematics and the program that Tarski and his colleagues in the Warsaw school, following Hilbert’s lead, devised for studying logistic systems generally under the heading of meta-logic.
Clearly, one could reject Carnap’s program by rejecting his restriction on the interpretation of the constants of a formal language to scientific terms and terms used to express empirical observations while retaining the idea of replacing philosophy with the study of the syntax of a formal language suitable for expressing not only propositions of empirical science but also propositions or proposition-like thoughts of other disciplines. Specifically, one could set aside traditional studies in normative ethics and instead study the syntax of a formal language suitable for expressing propositions or proposition-like thoughts of normative ethics. Alternatively, one could study the logic and syntax of sentences of a natural language in which the basic terms of normative ethics, such terms as ‘good’ and ‘bad’, ‘right’ and ‘wrong’, occur. In this alternative, the questions of cognitive meaning and definability of ethical terms on which moral philosophers in the middle decades of the 20th century concentrated would correspond to questions in the Carnap-inspired study about the formation rules for sentences in which the constants that one used these terms to translate occurred together with questions about how these constants were introduced into the formal language. Similarly, questions in the alternative study about logical relations between these sentences and others would correspond in the Carnap-inspired study to questions about the formal language’s transformation rules. Such a study, in either case, would, following Hilbert and Tarski (et. al.), be aptly called meta-ethics. And while no philosopher, as far as I know, ever proposed a Carnap-inspired study of the syntax of a formal language suitable for expressing theories of normative ethics, much work in ethics at mid-century falls within the alternative. A prime example is Hare’s The Language of Morals.
The earliest use of the term ‘meta-ethics’ or a cognate that I’ve found fits this alternative. It occurs in Abraham Edel’s contribution to the volume on Moore in the Library of Living Philosophers series (The Philosophy of G. E. Moore, Schilpp, ed., 1942). Edel, in this essay, distinguishes a category of terms he labels “meta-ethical” from three other categories, which he labels “ethical,” “partly ethical,” and “non-ethical” (ibid., pp. 138-41). Meta-ethical terms, Edel says, are terms used to talk about ethical statements. (The OED cites a 1938 article in the Philosophical Review as the source of the earliest occurrence of ‘meta-ethics’ in English, but the citation is misleading. The article, “Philosophy in France 1936-37” by André Lalande, is a translation from the French, and ‘meta-ethics’ is the translator’s translation of ‘métamorales’, a French term from the writings of the anthropologist Lucien Lévy-Bruhl. If anyone knows of a philosophical use of ‘meta-ethics’ or a cognate that precedes Edel’s, I’d be grateful for the reference.)
On the Carnap-inspired study, the point would be to investigate the syntax of a formal language suitable for expressing theories of normative ethics. The study would go beyond being a study of the syntax of such a language only in examining the bearing of the formation rules and transformation rules on formulae that contained constants whose translations were moral terms. Consequently, it would be necessary in constructing the object language to identify those constants, if any, in its set of primitive constants that were to be so translated. On the alternative study, the point was to investigate the logic and syntax of the portion of natural language in which such theories were expressed. Either approach is wholly independent of the project of expounding a theory of normative ethics, and meta-ethics was therefore understood to be a distinct subfield of ethics from that of normative ethics. None of its results had consequences for normative ethics since the object of its investigations, the object language, is fixed and unaffected by conclusions reached through these investigations about the language’s syntax or the logic of its vocabulary.
Interest in meta-ethics, so conceived, receded during the latter half of the 1960s and most of the 1970s as Anglo-American moral philosophers came to focus on the large and pressing moral and political issues that arose with the social upheavals of the 1960s. In the late 1970s, however, a new problem was introduced into the study that reignited interest in meta-ethics and transformed it. The new problem was that of realism, and the renewed interest in meta-ethics it sparked was doubtlessly due to the strong interest in realism that emerged in philosophy of science with the success of certain attacks on the empiricist programs that were the heirs of logical positivism and in philosophy of language with the introduction of theories of direct reference. Two works, both published in 1977, were especially influential in bringing about this renewed interest in meta-ethics. They are Gilbert Harman’s The Nature of Morality and J. L. Mackie’s Ethics: Inventing Right and Wrong. Both offered seemingly powerful arguments against the reality of moral properties.
Harman’s argument was fresh. It relied on a contemporary argument for scientific realism conjoined with a thesis of Quine’s that observations in ethics, unlike observations in science, are not probative (see “On the Nature of Values” in Quine’s Theories and Things). Over time Harman’s argument fell victim to criticisms that undermined both of its pillars. On the one hand, realists who accepted the argument for scientific realism on which it relied forcefully disputed Quine’s thesis. On the other, realists and others who had no stake in the debates over scientific realism effectively questioned whether this argument was relevant to the question of the reality of moral properties. Mackie’s argument, by contrast, was not fresh. He had advanced a shorter version of it thirty years earlier in a 1946 article, “The Refutation of Morals”, that appeared in the forerunner to the Australasian Journal of Philosophy. At the time of its publication the argument attracted little interest and quickly disappeared into the stacks of research libraries. Thirty years later its impact was thunderous. And, unlike Harman’s argument, it has not only survived but yielded progeny.
Such is the incongruity of analytic philosophy. Harman’s argument is elegant, even if both of its pillars are open to serious objection. The position Mackie defends, what he dubs “second order moral skepticism,” is incoherent. Let me explain. Its core thesis is negative. It is the thesis that there are no objective values. To hold this thesis, Mackie observes, is to be a moral skeptic, but such skepticism is not to be confused, he cautions, with other kinds of moral skepticism. Other moral skeptics reject morality wholesale or the conventional morality of their society. They hold, Mackie says, first order views, whereas his view is a second order view. As such, his holding it is compatible, he claims, with his having moral convictions regardless of content, even deeply held ones, because those convictions represent first order views. Since first order views and second order views are “completely independent” of each other, there is no incompatibility (Ethics, p. 16). The incoherence of his position then becomes evident once one sees how he distinguishes first order and second order views.
Mackie characterizes his distinction between the two views as an extension of the traditional distinction between normative ethics and meta-ethics. He writes, “The distinctions [I’ve] drawn . . . rest not only on the well-known and generally recognized difference between first order and second order questions, but also on the more controversial claim that there are several kinds of second order moral question” (ibid., p. 19). The distinctions to which he refers in this passage are the one between first order and second order moral skepticism that we have reviewed and one between two kinds of second order moral subjectivism that he uses to enlarge the class of questions that he thinks belong to meta-ethics. Thus second order moral skepticism, he says, is also a form of moral subjectivism, but it is not to be confused with the form of second order moral subjectivism according to which a simple ethical sentence like “Lynching is evil” is equivalent in meaning to the speaker’s saying that he or she hates evil. The form of second order moral subjectivism with which Mackie identifies his position, by contrast, is different from this one because it leaves open the question of what a simple ethical sentence means. It is a form of moral subjectivism, Mackie observes, only because a consequence of one’s denying that moral values are objective is that one takes them to be “in some very broad sense subjective” (ibid., p. 19). And its being different from the first form thus shows, Mackie concludes, that the class of second order questions includes ontological ones as well as linguistic or conceptual ones.
The problem, clearly, is that ontological questions cannot be second order questions if the latter, as Mackie maintains, are completely independent of first order questions. To deny that there are any objective values is to deny that there is evil in the world. And if one denies that there is evil in the world, then one either holds that the sentence ‘Lynching is evil’ is false or falls back on some linguistic or conceptual thesis about the logic or meaning of the word ‘evil’ to explain its being true or its being neither true nor false. Mackie, who argues for understanding a sentence like ‘Lynching is evil’ as purporting to represent some fact in the world, is therefore committed by this account of what the sentence means together with his moral skepticism to holding that ‘Lynching is evil’ is false. And this is a first order view about lynching. Nor obviously is the alternative of falling back on some linguistic or conceptual thesis about the logic or meaning of the word ‘evil’ so as to avoid denying that lynching is evil open to Mackie, for it would mean that he would have to abandon his taking ontological questions as a kind of second order question that one can ask independently of linguistic and conceptual ones. Hence, Mackie’s attempt to add ontological questions to the class of second order questions while preserving the independence between first and second order questions fails. Likewise, his idea of avoiding a noncognitivist account of the meaning of ethical words by supplementing the moral skepticism he shares with noncognitivists with an error theory also fails. A noncognitivist, unlike the error theorist, can coherently deny that lynching is evil (or, what is much more likely, affirm it) consistent with being a moral skeptic in Mackie’s sense.
In making this criticism of Mackie I am assuming that when he says that his distinction between first order and second order moral skepticism “rests on the well-known and generally recognized difference between first order and second order questions”, he is talking about the common distinction between normative ethics and meta-ethics that Anglo-American philosophers had been drawing for more than a quarter century and is taking questions of normative ethics as first order questions and questions of meta-ethics as second order. Given this assumption and using the Carnap-inspired study that I described earlier as a model of this common distinction, one can see immediately where Mackie errs in placing ontological questions among the questions of meta-ethics. The study of the logical syntax of a formal language can tell you which primitive symbols of the language function as vehicles of reference and how one can introduce new symbols that are also vehicles of reference. But what it cannot tell you, with respect to any of these symbols, is whether something exists in the world to which the symbol refers. However one determines whether such a thing exists, the determination is made outside of the study of the syntax of any language that contains symbols that one could use to refer to that thing if it exists.
While Mackie failed to elevate sentences expressing propositions of normative ethics like ‘There is no evil’ and ‘Nothing is good’ to the level of meta-ethics on the once common understanding of meta-ethics as a field that is completely independent of that of normative ethics, he did initiate a debate with defenders of objective values that significantly transformed the understanding of the field and what questions it contained.. The debate, owing to its becoming for a later generation of philosophers the paradigm of a dispute in meta-ethics, effectively removed the sharp boundary between meta-ethics and normative ethics that was generally recognized at mid-century. Ironically, then, Mackie, by bringing about a significant change in the subject, succeeded in this back-handed way in getting ontological questions recognized as questions of meta-ethics.