Metaethics 0007: G.E. Moore's Open Question Argument

This posting is inspired by Michael Ridge's article, "Ecumenical Expressivism: Finessing Frege," Ethics, Vol. 116, No. 2 (January 2006), pp. 302-336. However, that article is not substantially about Moore's open question argument. It discusses the argument and its relationship to non-cognitivist theories of value. Ridge's article is about the theory of value he proposes, "ecumenical expressivism". But, in doing so, he discusses the role that Moore's open question argument has in motivating non-cognitive theories.

In this post, I seek to take a look at that foundation as Ridge described it.

The open question argument aims to show that the meaning of "good" cannot be reduced to any natural property.

In all honesty, I agree with this conclusion. I argue that "good" = "is such as to fulfill the desires in question". However, it is a mistake to argue that this is what the term means in standard English. To call something good is to say that there are reasons to realize that thing. It is an a poteriori fact - a fact learned from experience - that the fulfillment of desires are the only reasons for intentional action that exist. Consequently, Consequently, "good" = "is such as to fulfill the desires in question" is not an a priori truth (true by definition).

Still, the open question argument has diverted a lot of philosophical labor-hours down some very unfruitful paths. Even if I were to offer "good" = "is such as to fulfill the desires in question" as being true by definition, Moore's open question argument would fail to show that this is mistaken.

So, what is the open question argument and how is it supposed to show that moral claims cannot be reduced to natural facts?

Let me borrow Ridge's characterization:

Take any proposed naturalistic analysis N of a moral predicate M. Moore’s open question argument maintains that it will always be possible for someone without conceptual confusion to grant that something is N but still wonder whether it is really M. If, however, N really was an accurate analysis of M then the question “I know it is N but is it M?” would not be conceptually open.

Moore used his open question argument against Sidgwick's claim that pleasure is the only good, so let us try it there. We take Sidgwick's claim, "good" = "pleasurable". We apply Moore's open question test and translate Sidgwick's hypothesis into a question: "Drug induced highs are pleasurable, but are they good?" You do not need to find an example where the conclusion is clearly "no". Any example will do. The point that Moore wants to make is that, no matter what example you take, the question, "X is pleasurable, but is it good?" has no obvious answer. The question, "X is good, but is it good?" would be taken to have an obvious yes answer. Indeed, we can ask why anybody would even ask such a question. If "good" = "pleasurable", asking the question, "X is pleasurable, but is it good?" would be like asking, "X is red, but is it red?" or "X is in Denver, but is it in Denver?" However, "X is pleasurable, but is it good?" is not like these other examples. This shows that "pleasurable" cannot be the same as "good".

The argument then takes one further step. It claims that there is nothing natural or material that you can test in this way and come up with a closed question. Pick any natural property you want, it will fail the open-question test. If all natural properties fail the open-question test, them "good" must refer to a non-natural property.

That's the argument.

The problem is that the open question argument does not work when you are dealing with ambiguous terms. Terms that refer ambiguously to two or more natural properties generate open questions when they are subject to the open question test.

So, let's take the proposal, "good" = "is such as to fulfill the desires in question."

Let us subject this to the open question test and ask, "X is such as to fulfill the desires in question, but is it good?"

Well, that depends. Are the "desires in question" the same in both cases?

The question could be asking, "X is such as to fulfill Agent1's desires, but are they such as to fulfill Agent2's desires?" This is an actual open question. The answer may very well be "no".

If we look at what desirism says about morality, we find that people quite often ask the question, "X is such as to fulfill those desires that people generally have reason to promote universally, but is it such as to fulfill Agent1's desires?" This is the way desirism understands the question, "Why should I be moral?" The answer may well be "no". There is no guarantee that doing the right thing will fulfill the desires of the agent. In fact, desirism understands most wrong action - most cases of people doing what they ought not to do - as cases where the agent does something that fulfills his desires, but does not fulfill the desires that people generally have many and strong reasons to promote universally.

In fact, things get worse for the open question argument when we introduce the idea of "conventional implicature". Conventional implicature is what happens when the perceived meaning of a term or phrase differs from the literal meaning. An example would be my wife shouting down to me as I work on a philosophy blog posting, "Honey, it's ten to six!" What she says is literally true. However, in this particular context, it means in this context is: "Quit working on that stupid blog posting and get ready for work!" We know what a term or phrase "really means" when we can imply that meaning from the conversational context.

Conventional implicature allows us to recognize rhetorical question. Rhetorical questions are questions with an obvious answer that tells the hearer to take this obvious answer and run with it. When we deal with an ambiguous question, conventional implicature typically allows us to judge when the agent is asking a rhetorical question (with an obvious answer) and when she is asking a serious question. A rhetorical question will be one that uses the same definition of the term in both parts of the sentence. A serious question uses different definitions in each part of the sentence.

Conventional implicature tells us to interpret the question, "X is such as to fulfill the desires in question, but is it good?" as a serious question. This means to interpret the question as, "X is such as to fulfill the desires in question, but is it such as to fulfill those other desires in question?"

This, indeed, is an open question - a question to which the answer is "no". This is because people do not ask questions like this if the answers are painfully obvious - and the answer would be painfully obvious if it referred to the same desires in question in both cases. Consequently, conventional implicature tells us to look for a different "desires in question" in each case. When we do, the question is a genuine open question.

So, I do not think that "good" = "is such as to fulfill the desires in question" tells us the standard English-language meaning of the term "good". However, even if it was, you would not be able to use Moore's open question test on it. Ambiguous terms generate open questions by default.