PART I
Background Thinking
1
GOOD AND GOLD
Jordan Howard Sobel
Editors’ Note
Jordan Howard Sobel (1929–2010) was, at the time of his death, working on a book-length manuscript entitled Good and Gold: A Judgmental History of Metaethics From G. E. Moore Through J. L. Mackie. We are happy here to publish substantial excerpts. While the title “Good and Gold” is a sensible one for Sobel’s whole project, it is clearly somewhat misplaced when applied only to this selection; nevertheless, we prefer to leave it as he intended.1
1 Introduction
This chapter is mainly about J. L. Mackie’s projectivist error theory of the concepts and realities of ethics (Mackie 1977, 1980, 1982). The theory is in part that “the realities of the concepts” do not answer to them: this is the error part. It is for the rest that in moral thought and discourse we project what are, in reality, subjective sentiments to “make up” objective moral properties and conditions: this is the projectivist or objectificationist part. Ordinary moral views and statements in which these would-be properties and conditions are ascribed are in error, which is not to say that these views and statements are all false.2 Consequent to the particular error common to them all, ordinary moral views and statements are none of them true or false. They are, of course, grammatically truth-apt, and may be true in the superassertability-sense of Crispin Wright (1996). They are also truth-apt in the sense of Frank Jackson (1998): they do purport to represent things as being a certain way. But as a matter of philosophical fact, given failure of presuppositions of the reality of their subjects, ordinary moral views are none of them true, nor are any of them false. So goes Mackie’s projectivist error theory. [. . .]
This theory has three pillars. The first is that the realism common to ethical intuitionisms regarding “moral values” is true of the concepts of moral thought and talk. The second is that this realism is not true of the realities of this thought and talk. And the third is that the realities of this thought and talk are subjective sentiments and interpersonal demands, the projection and objectification of which are in the interest of humanity, in that they facilitate the enhancement of these sentiments, since it is in the interest of humanity that they be strong.
2 Intuitionist Realism Is Correct as a Theory of the Concepts of Ordinary Moral Thought and Talk
(Mackie 1977: 38)
According to Mackie, ordinary moral thought and talk is in itself largely as intuitionists say it is. They are right as far as the semantics, logic, and concepts of moral thought and talk. They say of it nothing but “platitudes” made precise and sometimes unfamiliar by being dressed out in philosophic terms (“a priori,” “synthetic”), and terms used with particular philosophic edges (“entails,” “necessary”). There are many arguments for the power of the conceptual analysis run out by the intuitionist. Mackie organizes his own support for intuitionist realism as a good theory of the concepts of morality, and the “platitudes” surrounding them, by playing off principal alternatives in modern metaethics: naturalism and noncognitivism. Major weaknesses of these approaches to the concepts and semantics of ordinary moral thought and talk are strengths of the intuitionists’ line.
According to Mackie, moral judgments presuppose that there are objective values, that there are objectively prescriptive qualities. Simple moral judgments purport to ascribe them to things, deny them of things, and so on.
(Mackie 1982: 238)
Not only philosophers think of values as “at once prescriptive and objective,” as “external, extra-mental realities” which when known and appreciated do not “merely tell men what to do but will ensure that they do it” (Mackie 1977: 23). Ideas of values “at once prescriptive and objective” (1977: 23) are to be found not only in Plato. They are implicit in, they are part and parcel of (Mackie maintains), all ordinary distinctively moral thought and talk. They are central to the meanings of special words for this thought and talk: moral concepts are of such realities, such values. [. . .3]
3 Intuitionist Realism Is Incorrect as a Theory of the Realities of Ordinary Moral Thought and Talk
The first sentence of Ethics, after its preface, expresses its manifesto: “There are no objective values” (1977: 15).
Intuitionists are wrong about the realities of morality. And so are most people, though there are some, including some “outlaws and thieves” (1977: 10), who may have seen the light. If they have not clear consciences, but no consciences, this could be because they see that there is nothing moral. But such characters are rare. Most people are mistaken and encouraged by our ordinary moral concepts and ordinary moral thought and talk to take moral matters seriously.
But all of this is mistake-ridden, and centrally: it is of the essence of moral thought and talk to say that there are objective values. It is a general presupposition of all ordinary moral judgments that there are instantiable properties of objective value—which is to say that there can be objectively valuable things, or (in equivalent theoretical terms) that there are objectively valuable things in some possible worlds. This is the error Mackie finds in ethics. When he says there are no objective values, he is not saying merely that nothing is actually morally good, bad, right, wrong, and so on (“not merely”?!). He is saying that, as a matter of metaphysical fact, nothing is possibly morally good, bad, right, wrong, and so on.
It is not merely that, for example, there are no objectively wrong actions, but that there is no such thing as objective wrongness (cf. Mackie 1982: 115). It is certainly not merely that no kinds of action are actually objectively wrong, or wrong in this world, though some are wrong in other possible worlds. For the objective wrongness of a kind of action would supervene on its nature. And though (contrary to Swinburne 1976) “the logical character of this supervenience” would not be analytic (1982: 115),4 it would be necessary, so that if a kind of action were objectively wrong in any possible world, it would be actually objectively wrong in every world, and variations in what would be objectively wrong are not so much as possible. Mackie’s nihilism runs not only against there being particular actions and things that are objectively right or wrong, or good or bad; it runs not only against there being kinds of actions and things that instantiate these properties; it runs against these properties themselves of objective rightness and wrongness (objective ought-to-be-doneness, and ought-not-to-be-doneness), of objective good and bad (ought-to-be-ness and ought-not-to-be-ness). [. . .]
4 Reading Mackie’s Argument in Bayesian Terms
Mackie has negative reasons and positive reasons against ontological intuitionism. He opposes this theory somewhat as someone might (and as Mackie does) oppose the hypothesis that there is an Intelligent Designer of enormous power who is largely responsible for the variety and details of forms of life.
Negatively, a person can be impressed by features of living things that are apparently pointless and useless, and thus puzzling on the Intelligent Design hypothesis: nothing like the quite useless human appendix will be found in a well-made watch or computer. And a person can be impressed by how different a designer of forms of life would need to be from designers of which we have uncontested experience, and how different and mysterious it would be in its work. For this designer would evidently be incorporeal, since otherwise there would be the mystery of how it has gone all these years without being seen, or involved in collisions with things seen. Given this, it would have set about its great work of shaping species simple and complex, assembling genes, “tweaking” DNA, and so on (I am out of my depth!), with “no hands,” simply by the power of thought and will.
Positively, a person opposed to that hypothesis may be comfortable in his opposition because he has in hand another hypothesis that would explain the phenomena of forms of life simple and complex, especially those details that challenge designer hypotheses. There is, he might reflect, Darwin’s hypothesis of random variation and natural selection over billions of years sans intelligent intervention and free of deep mysteries of process and agency. There is, he might think, an alternative hypothesis to Intelligent Design that would have had “in the beginning”—i.e., before assimilation of the evidence of life, its history and diversity, now in hand—greater initial plausibility (that is, in terms of Bayesian confirmation theory, a greater prior probability), and which would have made the total evidence now in hand more likely (that is, in terms of Bayesian confirmation theory again, he might think that this alternative hypothesis will have had a greater likelihood for the now-in-hand evidence, or that this evidence would have been more likely) (cf. Sobel 2004, ch. 7.)
Mackie (1946: 77) writes that he pretends not “to be advancing any particularly new ideas” but to be offering a “re-statement of them” that reveals “how they may be brought together and interrelated” in a manner “radically destructive of all common views of morality.” I am offering to contribute to this project by casting the ideas Mackie’s advances in his 1977 book in terms of Bayesian “priors” for possible explanations of evidence, and “likelihoods” of this evidence on these explanations, i.e., for a Bayesian assessment that favors error theories (particularly Mackiean) over realistic theories (particularly intuitionist) of morality. I make this offer though there is no indication that Mackie was thinking in Bayesian terms when he first formulated his argument (1946: 77–86) and very little that he had it even somewhat in mind when he reformulated it in 1977. I make it because: (i) it is clear that from the start he intended an argument from many “considerations” to an “in all probability conclusion”:
(Mackie 1946: 77)
(1946: 80)
(1946: 86)
(ii) In my opinion, bringing together and interrelating “all the evidence and considerations” of chapter 1 of Mackie’s 1977 book, in terms of Bayesian “priors and likelihoods” assessment of a projective/objectifying error theory and an intuitionist theory of objective moral values, enhances the argument of this chapter for the former “in all probability” over the latter.
Further to Bayesian priors and likelihoods of an hypothesis h in relation to evidence e: What would have been unconditional probability Pr(h), before this evidence was in hand, is the prior of h; and what would then have been the conditional probability Pr(e|h) = Pr(h & e)|Pr(h), is the likelihood of h. In this scheme, what would then have been the conditional probability Pr(h|e) is the posterior of h. It is a theorem of standard probability theory that if hypotheses h and h' are probabilistically exclusive, Pr(h & h') = 0, and probabilistically jointly exhaustive, Pr(h ∨ h') = 1, then:
and similarly (of course):
so that:
Indeed, this equivalence of inequalities obtains for evidence e and alternative hypotheses h and h' whether or not these hypotheses are probabilistically exclusive and jointly exhaustive alternatives, since it is a theorem of standard probability theory that, for any probability function Pr, and propositions p and q,
This is a consequence of the so-called definition of conditional probability, and the principle that probabilities of logically equivalent propositions are equal.
A Darwinist would find that to his mind both the priors and likelihoods, in relati...