Questioning The Open Question Argument
Definitions can be substantive
It seems to me like there’s a knock-down reply to the open-question argument, even if one is an analytic naturalist. So, for defenders of the open question argument, why is what I’m about to say wrong?
What is the open-question argument? It’s an argument against analytic naturalism. Naturalism says that the moral facts are just some kind of fact about the actual, concrete, physical world. The first kind of moral naturalism says that moral claims are true by definition. For example, when you say “murder is wrong,” that might just mean “murder causes more pain than pleasure.” There’s another kind of moral naturalism called synthetic naturalism which says that moral claims are discovered through some type of natural inquiry. An example of a synthetic nonmoral claim is the following—I’m in the United States. That’s not true by definition but it is true nonetheless.
The most common objection to analytic naturalism is the open-question argument. Suppose you say that statements of the form X is wrong means X has some natural property. For example, the statement “murder is wrong,” has the same meaning as the statement “murder is not conducive to aggregate well-being.” But then you can ask “is it wrong to do things that are not conducive to aggregate well-being.” This seems like a substantive question at the very least; perhaps the answer is yes, perhaps it is no, but it’s not the kind of question that indicates being unfamiliar with the English language. But if wrong just means—by definition—what isn’t conducive to well-being, then it becomes an empty question—it’s like asking whether doing things that aren’t conducive to well-being is conducive to well-being. Of course not! So the fact that you can ask coherent questions about whether the thing that supposedly defines morality is really right shows, according to proponents of the open-question argument, that analytic naturalism is false.
But this argument seems to rest on an error. It’s perfectly possible to wonder coherently about whether a word has some particular meaning. Philosophers are, to this day, debating what knowledge means. If a person says “knowledge means justified, true belief where one would only believe it if it were true,” one can coherently wonder whether that’s true. Conceptual analysis involves picking out a term that is synonymous with some other term—but you can coherently wonder if some conceptual analysis has been done correctly.
I think pleasure means, by definition, desirable mental states. But one can coherently wonder about whether I’m right about that. People can propose other definitions of pleasure and then argue about which one is correct. Therefore, even if a word and a phrase are synonymous, one can still coherently wonder if examples of the first are necessarily examples of the second. Therefore, the central premise of the open-question argument would seem to be wrong.
Yea, I think this is right. I don't have the impression many people take the open question argument very seriously these days, and I recall encountering similar objections that struck me as fairly decisive. Personally, I think this is another instance of philosophers putting too much stock in how things seem.
Michael Smith gives basically this same response to the OQA in chapter 2 of his book The Moral Problem. In his view, the problem with the OQA is that is makes bad assumptions about the nature of conceptual analysis.
The OQA assumes, according to Smith, that if one concept is correctly analyzable in terms of another concept, then one of the concepts must be already "contained" in the other in some way, which implies that the analysis must be trivial and uninformative. However, he argues that an analysis of a given concept is correct if it articulates all and only the inferential and judgmental dispositions of those who do in fact have mastery of the concept. He believes that this is the correct account of conceptual analysis because of his account of what it is to acquire a concept. To acquire a concept is just to acquire a certain set of inferential and judgmental dispositions.
On Smith's account of conceptual analysis, a correct analysis can be non-trivial and informative. Even though someone who has mastery of a concept must *have* certain inferential and judgment dispositions, such dispositions may not be *transparent* to them. That is, whereas mastery of a concept requires knowledge*-how*, mastery of an analysis requires knowledge-*that*. In fact, basically all attempted analyses in philosophy seem to be non-trivial (e.g., knowledge, color, intentional action, etc.), with some calling this phenomenon the "Paradox of analysis".
I've seen different SEP articles mention this as well. E.g. this snippet is from the SEP article on Moral Naturalism:
> Most philosophers think that the OQA does succeed in refuting analytic descriptivism, yet analytic descriptivists have developed several responses to the OQA....Another traditional response appeals to the idea of the Paradox of Analysis: if all analytic claims are obvious, then it’s impossible for there to be an interesting, informative conceptual analysis. Since some conceptual analyses are interesting and informative (we philosophers tell ourselves, desperately), not all analytic claims are obvious. The OQA is a test of obvious analyticity, but the correct definitions of moral terms might just be non-obvious (Smith 1994, 37–39).