This is the first in a series of posts on the Moral Twin Earth.
G.E. Moore’s Open Question Argument is taken by many to refute analytic moral naturalism. These naturalists take good and some natural property N, like well-being, to be analytically defined–Sam Harris, for example. The open question argument says that competent speakers will treat the question “is is true that natural property N is good” as an open question. But if it is analytic, it wouldn’t be an open question in the way “is a bachelor an unmarried male” is not an open question. So, analytic moral naturalism is refuted.
The other option for a moral naturalist is to be a synthetic moral naturalist. These naturalists want to synthetically define moral terms in the analogous way water is synthetically defined as H2O. A synthetic definition of morality is not a priori defined, and requires empirical work. Against these naturalists Horgan and Timmons construct a Moral Twin Earth argument that aims to show problems with synthetically defining moral terms.
Horgan and Timmons build a Moral Twin Earth argument upon Putnam’s famous Twin Earth thought experiment. In Putnam’s thought experiment, Earth and Twin Earth are mostly identical except that the stuff called ‘water’ on Earth is H2O, while the stuff called ‘water’ on Twin Earth is XYZ. Here, the intuition is supposed to be that when Earthings and Twin Earthlings talk of water they mean different things, so there couldn’t be disagreement about what is water; they’re merely talking past each other. (This intuition may come from joining meaning and reference on a semantic externalist view.)
Horgan and Timmons present a parallel though experiment involving moral terms instead of ‘water.’ On Earth, moral terms are causally regulated by a consequentialist theory TC; and on Twin Earth moral terms are causally regulated by a deontological theory TD;. If an Earthling said ‘lying is not always wrong’ and a Twin Earthling said ‘lying is always wrong’, we have the intuition that they’re disagreeing. But on Boyd’s causal theory of reference, they’re not disagreeing; they’re talking past each other like in the water case. Horgan and Timmons conclude that Boyd’s semantics cannot work for moral terms, and the same problem will occur for other semantic theories that express moral properties in terms of natural properties. Synthetic semantic naturalism, of any kind, is in trouble.
Copp notes that synthetic semantic naturalism is partly a semantic thesis that’s not entailed by all versions of moral naturalism. Moral naturalism is only the metaphysical thesis that moral properties are natural properties. A moral naturalist, like Nicholas Sturgeon, can hold that moral properties have the meta-property of being natural without holding to synthetic natural definitions; the MTE argument cannot be used against this version, assuming that this version doesn’t leave open the possibility of moral terms expressing different moral properties between Earth and Twin Earth.
Copp offers two replies, or two routes, that the moral naturalist can take to diffuse the MTE argument.
First reply: Best Translation
The Moral Twin Earth argument assumes moral disagreement entails that ‘wrong’ and ‘twin-wrong’ express the same property. Copp thinks that there can be moral disagreement even if ‘wrong’ and ‘twin-wrong’ express different properties, because it might be that the ‘twin-wrong’ is the best translation for ‘wrong.’ Different languages often have inexact translations for their terms. Copp says that while moral and twin-moral terms would have different meanings in the philosophical sense, there is an ordinary sense where they have the same meaning. (I found this point about meaning confusing. I think he’s saying if we reject Boyd’s semantics we can have the same meaning.)
Second reply: About errors in semantics
Copp thinks that the tools Putnam developed in explaining various errors in semantics can be used to show that ‘The resulting semantics implies that corresponding Earthling and Twin Earthling moral and twin-moral terms in Twin Earth scenarios express the same property if they express any property at all.” In other words, based on Putnam’s semantic theory, the MTE scenario is impossible: the Earthlings make a mistake, or the Twin Earthlings make a mistake, or they both do.
While Boyd used a causal theory of reference, Copp prefers to use Putnam’s semantics. First we need to know something about Putnam’s semantics. In determining the extension of a term we start with the referential intention, and, then, we figure out the “important physical properties” that the samples have in common. So in the case of the term ‘water’, the referential intention would be ‘the liquid that has the underlying nature of the local samples’, and the important physical properties that the samples have in common is H2O. In addition to natural kinds, there are artifact terms like “soda pop” and functional terms like “milk.” The important physical properties are “interest relative.” For water the molecular structure is most important, while for milk the function is the most important, since human, cow, and goat milk all have different molecular structure. So milk as a functional term will refer to something like ‘the liquid mothers feed their young.’
This semantic theory also needs to take various mistakes in to account. For example, if Putnam pointed to a glass of clear liquid and ostensively defined it as ‘water’ while it was actually gin, then Putnam does not intend for his ostensive definition to be accepted. This is because the referential intention of the speaker presupposes the liquid in the glass is the same kind as the liquid found in lakes. Or, suppose Putnam pointed to a narwhale tusk and called it a ‘unicorn horn.’ Here, the referential intention is for a horse-like animal, so, in this case, the extension of ‘unicorn horn’ is empty.
Copp says that moral semantics must allow for analogous mistakes: “if our moral views are mistaken, then we might make mistakes in calling actions right or wrong.” Copp says that this is the most important error in Horgan and Timmons’ reasoning: It could be that speaking of ‘wrong actions’ is like speaking of ‘unicorn horns’ or calling gin ‘water.’ So even if Earthlings believe that TC actions are right it doesn’t follow, given the referential intentions and interests of the speaker, that TC; actions are right.
Given these considerations about Putnam’s semantics, Copp concludes:
… any coherent moral theory would say if wrongness is instantiated on the one planet it is instantiated in the other, and vice versa, and that the kinds of actions that are wrong on the one planet are wrong on the other, and vice versa. But if the worlds are the same in all morally relevant respects, and if corresponding terms in the languages spoken on the two worlds are used with relevantly the same referential intentions, and if people’s interests are the same in using these terms, then the terms would be assigned the same meaning in Putnamian semantics.
Next up is Horgan and Timmons’ reply.