The “Open Question Argument” is supposed to establish something important for (meta)ethics; namely, that the property of being good (or value, or of what one ought to do, etc.) is not entailed by, and thus not identical to, any natural property like pleasure or knowledge. It goes something like this:
For all natural properties N, settling questions about whether some A is N leaves open whether A is good.
So, goodness is not entailed by any natural property N.
One way of replying to this argument is to deny the validity of the argument, by appeal to a concept/property distinction like that between water and H2O. Another way, with which I have some sympathy, is to say the argument is confused because there is no such thing as goodness simpliciter. But a third way is to deny the premise, which seems easier than it’s alleged to be:
Someone asks me whether dousing a cat in gasoline and lighting it on fire for fun is good. I reply, No, that’s not good. “Are you sure? Isn’t that an open question?” Yes, I am quite sure; no, it’s not an open question. Likewise, I am pretty certain that raping young girls in front of their parents in order to humiliate them is not good. And I am pretty certain that taking the time to help an elderly person open a heavy door, for its own sake and when nothing else is pressing, is good. When I think about cases like this, it seems like there are all sorts of areas in which goodness or its lack is not an open question at all. (Also, AFAICT, anyone who thinks the problem of evil is a serious issue for theism is committed to the same view.) Arguments to the contrary would strike me as theory-driven. In short, the OQ argument strikes me as far more powerful in the abstract, and for general properties, than for more particular, concrete cases.
What I don’t have (much of) is any kind of systematic account of how all these closed questions about goodness fit together. What does that show? That I have no comprehensive ethical theory. (This is why, presumably, the more abstract questions about goodness remain open for me.) But it seems to me that it also shows that, were I to get a theory, it would be one which equated goodness with natural properties.
There is another, anti-theoretical, option too: perhaps ‘good’ is a family-resemblance term like Wittgenstein’s ‘game’: I am quite sure in a number of cases whether some activity is a game; there are another bunch of cases in which I am not sure; and there is no over-arching theory about games because the term isn’t a rigorous one that way. This does not show that games are non-natural entities or that calling something a game is expressive of my non-cognitive attitudes. Perhaps something along these lines is the right way to understand “value pluralism.”
At any rate, however, the premise of the OQ argument seems to me just plain false for a wide range of cases. Am I missing something important? Thoughts?
Heath,
You think that “Is dousing a cat in gasoline and lighting it on fire for fun good?” is clearly a closed question. But what exactly do you take the distinction between open and closed questions to be? For instance, is the question “Is grass green?” open or closed on your view? It seems to me that you are assuming that whether a question is open or closed depends on whether its answer is obvious or certain. I don’t think that’s right, or, at least, I don’t think that what those who offer the open question argument have in mind. I think that they must think that whether a question is open or closed depends on whether or not it’s possible for a competent speaker to understand the question and yet not know its answer, or they must think that it depends on whether or not it is sufficient to know only the meanings of the words contained by the question in order to know its answer. If that’s right, then it’s not clear to me that “Is dousing a cat in gasoline and lighting it on fire for fun good?” is a closed question even though I’m quite sure the answer is “No; It’s not good.”
Heath, I think that the target sentences that are then put into the OQA are supposed to be those offered as analyses/identities. So I think fans of the OQA need not deny that some predications involving the relevant terms are closed.
Echoing MvR: If I recall correctly, in Moore’s presentation of the OQA, he explicitly says that questions about ‘the good’ (i.e., which items bear the property ‘good’) often can be closed, but all proposed analyses of ‘good’ are open.
Doug wrote:
I think that they must think that whether a question is open or closed depends on whether or not it’s possible for a competent speaker to understand the question and yet not know its answer, or they must think that it depends on whether or not it is sufficient to know only the meanings of the words contained by the question in order to know its answer.
That might be what they had in mind, but then a related worry needs to be addressed. If there is a plurality of possible ways someone might come to grasp the concepts needed to grasp the question or the ‘bar’ for counting as a competent speaker is set too low, it could be that certain questions can be given in answers that are conceptual truths even if a competent speaker does not grasp that such questions can be given answers that are conceptually true. So, the fact that a question is open for such a speaker cannot be taken as any indication that the question is open in the sense that it’s correct answer depends upon more than some mere conceptual truth.
A second related worry is that if you take the openness of a question to be a matter of whether the question can be settled by, say, any competent speaker, it seems that it is possible for a competent speaker not to know a conceptual truth that is for another self-evident. It seems odd to think that the question is open if there is someone who knows on conceptual grounds how to answer it. I mean, the following seems true:
(1) For Heath it is just obvious that dousing a cat in gasoline and lighting it on fire for fun is not good.
It _seems_ like I contradict myself if I add:
(2) It is, however, an open question as to whether dousing a cat in gasoline and lighting it on fire is good.
But, then, I’d say that a question cannot be open when there is someone who knows how to answer it on conceptual grounds alone. Given my intuitions about knowledge ascription, that there is someone who can answer such questions knowingly is consistent with there being competent speakers that cannot.
There is an interesting question here that if I recall correctly is addressed by Ross and Sidgwick. I thought that Sidgwick was an intuitionist who thought that our intuitive knowledge of certain moral truths concerned general propositions whereas Ross seemed to think that what we knew with self-evidence was something about particular situations. Someone who wants to run OQA’s, I would have thought, would want to address not just claims about general connections between goodness and natural properties, but also about goodness in particular situations described naturalistically. But, maybe if they try to run such OQA arguments, they need something to deal with Heath.
Clayton,
Could you give an example?
Not to me. Try it with Doug’s suggestion of replacing the thing that seems obvious to Heath with something else obvious, like the fact that grass is green.
Doug and maybe Jamie think the question of whether cat-immolation (for short) is good is open. Mark and Michael are willing to grant that this question is closed, though analyses of good remain open. In two parts then:
Suppose first, with Mark and Michael, that the cat-immolation question is closed but analyses of the good remain open. OK—but this, I think, shows very little. Questions that ask for analyses of knowledge or free will or [pick your topic] remain open, since we don’t know how to answer them definitively. (Note that we are often sure about individual cases of knowledge, free will, etc., at least pre-theoretically.) But that doesn’t show that there is no analysis of these concepts. Maybe it’s just hard; or maybe they’re like Wittgenstein’s games. Few people feel pushed toward non-naturalism or expressivism about these concepts (though maybe that would be interesting).
Alternatively, maybe the question about cat-immolation remains open because the answer to this question, though very clear, is not conceptually true. I think Doug puts this challenge exactly right. Here’s a tentative answer: I would claim to know that immolating cats is not good. I know this a priori, just by considering the question. If this is analytic a priori knowledge, then the answer to the question is conceptual and the question is closed. If this is synthetic a priori knowledge, then there is a relation between the properties (not the concepts) of goodness and cat-immolation. Either way, it’s naturalist moral realism.
If we rephrase Clayton’s challenge this way—“Heath knows a priori that p but it’s an open question whether p”—that strikes me as self-contradictory. You need the success term “knows” or something similar in there.
And it’s quite possible for one person to know something a priori when another doesn’t; lots of Burge’s examples would fit. For example, it may be an open question for A whether one can take out a mortgage on a houseboat. (Answer: no.) A might be in many respects a “competent speaker” with ‘mortgage.’ But it may be a closed question for B, who understands the concept mortgage more deeply. I don’t have a problem thinking that my (or anybody’s) understanding of goodness might be like that.
Of course, at this point, one might challenge my claim to knowledge about cases of goodness. But I don’t think the OQA will be any help (quite the contrary) in backing up that challenge.
Damn, I hadn’t seen Jamie around in a while. Just when I thought I could get away with something…
When I said ‘For Heath it is just obvious that…’, I was assuming that in ordinary usage that implied ‘Heath knows that…’ (I thought if Heath was mistaken, you’d correct me by saying ‘For Heath it _seems_ obvious’). Anyway, the view I was floating was that knowledge closes questions. If you do not want to say that the question as to whether grass is green is open, let’s say that apriori knowledge closes questions.
The examples that motivated anti-individualism were the examples I had in mind when I was talking about the ‘low bar’ to having sufficient grasp for entertaining the question and not being able to close it. So, for example, even if it is an open question for Gil as to whether the Pope is a bachelor, that seems consistent with someone else’s knowing apriori that all unmarried adult males including the Pope are bachelors. (If the Pope is not a bachelor, rewrite as you see fit. Obviously, this is a question I cannot close.)
Heath, I agree that the ‘openness’ of questions even to competent speakers shows very little. So even making the adjustment to the argument so that the target claims are analyses, I think the OQA doesn’t show either that the things designated by the terms are non-identical, or even that we don’t have an analysis of the concepts involved.
I tried to post this earlier today, but it doesn’t seem to have shown up. And Clayton appears to be saying something along the same lines (leave it to the epistemologists to notice this!).
Heath wrote:
For all natural properties N, settling questions about whether some A is N leaves open whether A is good.
Here is an N that some A has: I know that it is good. (Heath’s words: “any natural property like pleasure or knowledge.”) If I know that q, then the question whether q is closed, on any recognizable sense of ‘closed’.
Ethical non-naturalists might want to also be non-naturalists about epistemic properties, too, in which case they won’t be impressed by the example. If that’s a further commitment of ethical non-naturalism, then it’s worth making it explicit.
Well, ‘open question’ is a term of art, so I don’t think we can use our inuitions about its extension as premises. So I’m having trouble following the dialectic.
Heath, I just don’t understand the example about mortgaging a houseboat. I would have thought it was obviously not knowable a priori that nobody can get a mortgage on a houseboat. Clayton, I guess my own view is that Gil has a different concept in mind that he associates with ‘bachelor’.
John, even a naturalist about knowledge must surely be a non-naturalist about the property, being known by John to be good. Right? (Since it analytically entails being good.)
Jamie,
My understanding (!) is that a mortgage is by definition a lien against real estate. So you can take a loan out against your houseboat, but it won’t be a mortgage.
OED:
Intuitively obvious!
Jamie’s right: ‘open question’ is a technical term, so some of us have been a little sloppy. This goes back to something Mark van Roojen said early on in the thread. (Leave it to the Harman students to point this out.)
The open question argument is intended as a refutation of analytic reductionism, which says (roughly) that ‘good’ just means the same thing as ‘N’, where ‘N’ names (or predicates) some natural property.
Let an ‘open question’ be one that a competent language user could possibly reasonably answer either way, i.e., ‘yes’ or ‘no’ (maybe ‘maybe’ or ‘I don’t know’ should be options, too–it doesn’t really matter).
The open question argument, then, goes something like this.
(1) If ‘good’ were synonymous with ‘N’, then ‘Is an N-thing a good thing?’ would [could?] not be an open question.
(2) But that is an open question.
(3) Therefore, ‘good’ is not synonymous with ‘N’.
This is consistent with certain natural properties entailing that something is good, and thereby settling (in the ordinary, non-technical sense of ‘settling’) particular moral questions. So ‘J knows A is good’ entails that A is good, and may settle the question of whether A is good.
But even granting that the property known by J to be good is a natural property, it’s still false that ‘good’ is synonymous with ‘known by J to be good’. Consider:
(1′) If ‘known by J to be good’ were synonymous with ‘good’, then ‘Is this good thing a thing known by J to be good?’ would not be an open question.
(2′) But that is an open question (something could be good without J knowing it).
(3′) Therefore, ‘known by J to be good’ is not synonymous with ‘good’.
(See here for more.)
I’m uncomfortable with the set-up of the problem. Do any properties *entail* other properties?
(P.s. I seem to remember that Stephen Ball had a nice paper on the OQA from APQ 1988.)
Chisolm said Yes.
Nick,
Doesn’t being a mammal entail being an animal, and being a square entail being a quadrilateral? I’m guessing it’s just a terminological matter. We could dress things up so that we talked about truths regarding property instantiation entailing other truths about property instantiation, and the basic line of thought will remain essentially the same, no?
Uh.
Chisholm.
Well, we need *a* notion of dependence or determination that has nothing to do with meanings or propositions or truths or ‘settling questions’. Call those *Matters-Medieval*. Heath’s examples (in Renford Bambrough style) are of his confident moral views or confident views about what is morally right and also necessarily so. How do these ‘intuitions’ bear on Matters-Medieval? One suggestion: a la Kripke; they are an initial appearance of Matters-Medieval that should stand? However, even if veridical, such intuitions do not seem to bear on the questions about identity and distinctness of moral and natural properties that at least some of the time preoccupied Moore.
To me, the force of the OQ argument is more in the challenge it provides for justifications that are offered for claims about goodness. You claim to quite certain that lighting cats on fire is not good. (I agree but…). You gave no other justification. So now we would ask whether the question ‘are things that Heath is quite certain aren’t good in fact not good’ is an open one. Surely it is. Meaning that you need to do more to justify your goodness claim than just express your certainty. I’m sure you could do this but the OQ test will always be lurking as a challenge to whatever justification you put forth.
I know this isn’t precisely what Moore had in mind, but I think it’s in the spirit of the criticism of hedonic utilitarianism. Bentham or Mill’s justification for the non-goodness of burning cats would be that it increased net suffering in the world. Run the OQ test. Is the question ‘are acts that increase net suffering always bad (not good)’ an open one? If so, then you have do more to justify non-goodness claims than than pointing to a net increase in suffering.
In other words, the OQ argument seems most effective when it is used to challenge particular justifications of moral claims. (And this challenge applies whether the justifications come in form of a theory or not.) It is less effective, as many have noted, when it attempts to establish sweeping claims about the undefinability of goodness.
I appreciate Tammler’s comment because it gets at the heart of something important about moral epistemology. I’ll try to be brief: in non-moral epistemology, the need for some kind of foundationalism is widely acknowledged. That is, we need to have justified beliefs, which are not justified by any other belief, in order to get knowledge off the ground. Otherwise we get familiar kinds of regress arguments for skepticism. It is possible to invoke externalist conditions in our explanation of foundational justification, but by their externalist nature, those won’t (usually) be articulable by the knower. Furthermore, most non-moral epistemologists would say that what’s foundational are particular beliefs, not general epistemic principles. (Cf Chisholm on particularism and methodism.)
Now it seems to me that the OQA is, in effect, a request for a self-evident-—that is, foundational, on an internalist understanding of foundations—-moral belief. And this has generally been sought at the level of moral principles. But moral epistemology ought not be either internalist or principle-driven (methodist); our first moral knowledge is probably particular moral judgments, foundationally justified in an externalist way. So that thought, extraordinarily rough as it is, lies at the root of my willingness to stick my neck out and say that I just know that burning cats is bad.
Nick, I don’t know if any of this answers your worries…I’m a little afraid I’m not erudite enough to be on the same blog with you!
Heath, if I’m understanding you correctly, you want to pursue a strategy here similar to that which Moore pursued against skepticism. (At least as I understand how Moore was responding to the skeptic.) Skeptics, Moore claims, want to argue this way:
(1) We can know that p (for any p) only if we can rule out skeptical hypothesis H;
(2) We cannot rule out H;
(3) Therefore, we cannot know that p.
Moore, however, wants to argue this way:
(4) I know that here’s a hand;
(5) I cannot rule out skeptical hypothesis H;
(6) Therefore, it is not the case that I know that here’s a hand only if I can rule out H.
Moore then asks us whether we should accept (1) or (4).
Heath, I think, wants to pursue a similar strategy. Imagine a “methodist/generalist” opponent who holds:
(7) I am justified in believing that burning cats is bad only if I can provide further evidence in support of that belief.
Heath wants to reply:
(8) I am justified in believing that burning the cat is bad;
(9) I cannot provide further evidence in support of that belief;
(10 Therefore, it is not the case that I am justified in the believing that burning the cat is bad only if I can provide further evidence in support of that belief.
Heath now asks whether we should accept (7) or (8); that is, whether (7) or (8) is a better candidate for “self-evidence.”
Heath, if this is in fact the point (or one of the points) you want to make, and if you haven’t already, you might want to look at Scott Soames’s chapters, in *Philosophical Analysis in the Twentieth Century*), on Moore’s ethics, in which Soames raises the question of why Moore didn’t resolve a certain tension in his philosophy by utilizing the strategy you may be suggesting here.
Heath,
Thanks for the reply. A couple of quick points.
You write “in non-moral epistemology, the need for some kind of foundationalism is widely acknowledged.”
I’m not so sure. One common response to skeptical uses of the OQ argument is to employ the ‘companions in guilt’ strategy, right? Sturgeon, for example, as I understand him, argues that foundationalism isn’t tenable in scientific inquiry and therefore that we shouldn’t hold moral inquiry to foundationalist standards either. Instead, both in science and in ethics we arrive at our knowledge through something like wide reflective equilibrium (WRE). I have some sympathy with that kind of response, actually. (Dan’s point about whether (7) or (8) is more self-evident applies here as well.)
However, I think that the OQ argument or challenge could apply to the moral epistemologies themselves. Whether one appeals to straightforward foundationalist standards or externalist inarticulable conditions or wide reflective equilibrium– one can still ask whether “if after employing [your favorite moral epistemology] we determine that something is bad, is it in fact bad?” is an open question. If so, that spells trouble for the moral epistemology, foundationalist or not. (Sorry if that seems convoluted, I hope it makes sense.) So I’m not sure I agree that the OQ argument is, or has to be, a demand for a foundationalist notion of self-evident beliefs. Or if it is, the foundationalism to which it appeals is at a deeper level than just ethical inquiry. Of course, at some point, one might reply ‘no, it’s not an open question. If a particular methodology or epistemology produces the judgment that something is bad, then it is bad. Case closed.’ At that point the OQ challenge will have been met. But it is no less worthy a challenge for that.
Dan,
Your 4 – 6 makes it sound as though Moore denied closure under known obvious entailments. Dretske and Nozick make that move, but that’s not how I read Moore.
The standard interpretation has Moore accepting closure and arguing something like:
M1. I know that I have hands.
M2. I know that I have hands only if I know that I’m not dreaming [or whatever skeptical H you’d like].
M3. Therefore, I know that I’m not dreaming.
Tamler,
Whether one appeals to straightforward foundationalist standards or externalist inarticulable conditions or wide reflective equilibrium– one can still ask whether “if after employing [your favorite moral epistemology] we determine that something is bad, is it in fact bad?” is an open question. If so, that spells trouble for the moral epistemology, foundationalist or not.
Could you say why it would spell trouble for the moral epistemology?
Hi Heath,
Sorry, I didn’t mean to erudite. You open by saying that the conclusion of the OQA is about properties. But then it seems that what you say is heavily epistemological, about foundationalism etc. I couldn’t clearly see the connection, that’s all.
Cheers, Nick
Hi John,
I’m happy to defer to the “standard interpretation” here. I’m sure others know Moore’s epistemology much better than I do. But can you say why you think 4-6 make it sound like Moore denied closure under known obvious entailments? (Sorry to be dense here. I’m not seeing it.)
Dan,
I didn’t mean to sound doctrinaire, and certainly don’t want to encourage people to simple defer. Anyhow, here’s the motivation for my remark.
Substituting ‘We’re all brains in vats in a world where there are no hands’ for ‘H’, we’d get Moore saying:
(4) I know that here’s a hand.
(5′) I cannot rule out the skeptical hypothesis that we’re all brains in vats in a world where there are no hands.
(6′) Therefore, it’s not the case that I know that here’s a hand only if I can rule out that we’re all brains in vats in a world where there are no hands.
The obvious known logical entailment would be:
Hand and World (HW): If here’s a hand, then we’re not all brains in vats in a world where there are no hands.
Dan,
I hadn’t thought of it in precisely the terms you articulate – I was more worried about regress-skepticism than alternative-hypothesis skepticism – but I’m not opposed to Moorean-epistemological lines on principle. This is an area in which I need to think more deeply before committing myself too much. Thanks for the Soames reference.
Tamler,
My thought is this: we can talk about our methods of settling D-domain questions as a “D epistemology.” Now, if you think that the best efforts of your D epistemology don’t settle D questions, that’s just a lack of confidence in your D epistemology. Which may be justified or not; but then you’re just a D-skeptic. Compare:
If that’s an open question for you, then you just don’t have a lot of confidence in the relevant scientific methods, and you’re a chemistry skeptic (in the sense that you think we can’t really have any chemical knowledge). Obviously, this goes beyond simply thinking that the relevant tests are fallible.
Nick,
Erudition is no vice! I am far from sure that I’m seeing what you’re not seeing 🙂 but my idea was that property “identity” (no obvious concept that) between P and Q was just
Necessarily, (x)(Px iff Qx)
and property “entailment” of P from Q was
Necessarily, (x)(Px if Qx)
and that I was putting forward claims with the latter structure. Whether they are analytic/conceptual or not (I wouldn’t go to the wall for the analytic/synthetic distinction, btw) depends on something about the meaning or conceptual grasp of the predicates, and is, I think, irrelevant to my initial knowledge-claims and also to the epistemological views (given that I’m willing to go externalist) I half-articulated above.
Hi John,
Thanks. If I’m understanding your reply now, it’s not that (4)-(6) suggest that Moore denied the *thesis* of closure under known entailment, but that (4)-(6) suggest that Moore wasn’t utilizing an obvious entailment that he knows. Is that right? (BTW, I certainly didn’t take you to be being doctrinaire. I was just admitting that my knowledge of Moore’s epistemology is, unfortunately, not great.)
Dan,
Yes, there is that loophole. I think he’d use it if he could, though!
Heath and John,
The remark that John asked about (and Heath responded to) wasn’t clear. It sounded like I was using OQ argument to challenge the truth of a particular judgment produced by a moral epistemology. Heath’s ‘companions in guilt’ response about the chemistry skeptic works fine for that. So let me try again:
If it’s an open question whether acts judged to be good produced by [your favorite moral epistemology] are in fact good, then it would show that [your favorite moral epistemology] isn’t capturing the meaning of ‘good.’ For example, suppose someone said that holding the door open for elderly ladies was good and provided the following justification. After employing wide reflective equilibrium, I have determined that fully rational people would desire to hold the door open for elderly ladies. One response could be: Ok, I agree that fully rational people would want to hold the door open for elderly ladies, but is holding the door open for ladies really good? If that genuinely seems like an open question, then it would spell trouble for the moral epistemology– not only that it was fallible, or that it wasn’t producing true judgments of goodness in this particular instance, but that it didn’t even capture our intuitive understanding of the meaning of goodness. I’m not sure this is analogous to “chemistry skepticism” since, at the very least, the method employed in that case still captures what we mean by sodium cloride.
However, now I’m starting to think that the Open Question argument is playing less and less of a role…