Companions in Guilt

Let there be two discourses, A and B, and we're trying to figure out whether A is real or not.

Here's an abstraction of a companions in guilt argument:

(1) If A isn't real, then B isn't real. (Premise) (So it follows that B isn't real)
(2) But B is real. (Premise)
(3) A is real. (1-2)

How do we know (1) and (2) are true?

(1): Suppose that there were a reason to think that B is real that is not also a reason to believe that A is real. Then would (1) still be plausible? No, because it could be that A isn't real because it doesn't have B's good feature. So the plausibility of (1) depends on there being no reason to believe that B is real that isn't also a reason to believe that A is real.

(2): We have some reason for thinking that B is real. Suppose that this reason for thinking B is real is also a reason for thinking that A is real. Then we wouldn't need the above argument at all, rather, we would be able to argue that: B is real, and if B is real then it follows that A is real, so A is real by a simple modus ponens. So either the above argument is unnecessary, or there is a reason to believe that B is real that is not also a reason to believe that A is real.

But now (1) and (2) seem to stand in tension. What makes (1) plausible is that there's no good reason to believe that B is real that isn't also a reason to believe that A is real. But what we saw was that either the above argument isn't really necessary (since our reason for thinking B is real is just as good for thinking A is real) or the reason for thinking that B is real is not also a reason for thinking that A is real. So if the companions in guilt style argument I presented above is necessary to prove the reality of A, then it seems to be struck with internal tension.

Now, Cuneo's core argument is:

(1) If moral facts do not exist, then epistemic facts do not exist.
(2) Epistemic facts exist.
(3) Moral facts exist.

Why should we believe (1)? Only if there is no reason to think that epistemic facts exist that isn't also a reason to think that moral facts exist. Why should we believe (2)? Because we have reason to believe that epistemic facts exist. Do these reason(s) also apply to moral facts? If they do, then we don't need the companion in guilt argument, so if we do need the above argument, then there's a reason to believe that epistemic facts exist that isn't a reason to think that moral facts exist. And so we have no good reason to believe (1) anymore.

In chapter 8 Cuneo recognizes this as a possible objection to his argument, but all I'm trying to say here is that it's not just a possible interpretation of the data, but rather that the tension is an inevitable feature of these sorts of arguments.

Now, what can one respond to this? It seems clear that we have a reason to believe in epistemology that isn't a reason to believe in ethics. Is there any reason to believe (1) under these circumstances? Yes, if you think that the following is true: if the features that ethics and epistemology share are not just problematic, but they're REALLY problematic, and would overpower whatever reason we have to think that epistemology is real. In other words, having that good feature would not be sufficient to override the bad features.

So here’s a fuller version of the argument:

(A) Moral facts do not exist only if [For any discourse, if a discourse has the "problematic features" then the discourse is not real even if it is indispensable for theorizing.]
(B) Epistemic facts have these same set of “problematic features.”
(C) Epistemic facts do not exist, even if they are indispensable for theorizing.
(1) So, if moral facts do not exist, then epistemic facts do not exist. (A-C)
(2) Epistemic facts exist, because they are indispensable for theorizing.
(3) Moral facts exist.

So now we have a contradiction, so what went wrong? The problem can't be (1) or (C), and so (A) and (B) are suspect. Cuneo argues that (B) is true, so let’s give him that. And I guess what we’re supposed to see is the falsity of (A). So if (A) is false, then the following conditional is false:

(A’) If [For any discourse, if a discourse has the “problematic features” then the discourse is not real even if it is indispensable for theorizing] then moral facts do not exist.

That (A’) is false means that it could be true that in general the “problematic features” kill a discourse, but that moral facts exist nonetheless. This means that moral facts must be missing the problematic features.

I still feel really uneasy with this argument, because the contradiction doesn't seem surprising at all. If we think that being indispensable to theorizing (or whatever reasons we have for thinking that epistemic facts exist) are really knock down considerations, then why on earth would we think that the problematic features could override them? In general, I just have no good reason to believe (A), especially given the fact that the considerations in favor of believing the reality of epistemology are supposed to be so very strong. So I'm inclined to doubt (A).

So here's what I'm going to do now. First, I'm going to try to look at the rest of the arguments that support (B), specifically the argument from disagreement (an argument that Cuneo marks as being the one that perhaps favor epistemology over ethics). At this point I'm willing to concede the first four arguments. I've expressed a bit of skepticism already about the explanatory requirement being a problem equally for epistemology as it is for ethics--though I should probably make sure that I'm not failing to think of epistemology in a sufficiently normative way.

After that, though, I'll just give Cuneo (B). And I think this point is why I'm not sure that there's much of a good reason to believe the core argument.

But Cuneo's got one more argument, and that's presented rather quickly in Chapter 8 in response to this argument.

Then I've got to figure out what can be taken from this. I think that the idea there is going to be that we need to figure out how to make sense of the queerness of normativity, because Cuneo has successfully shown that epistemic facts exist and that they would be normative and so queer. Maybe we can distinguish between the normativity/queerness of ethics and epistemology, or maybe we can focus on one of these first four arguments to try and figure out why those problematic features are not problematic.

Cuneo's 6 Arguments against Realism: III

I'm a bit confused by Wright's version of the explanatory requirement, and so I think that I probably chase down "Truth and Objectivity" at some point and take a look. But Wright says, according to Cuneo, that "were moral facts to exist, at least the following must be true: Such facts would have to explain the existence of non-moral facts of a sufficiently wide array of types." This is slightly different from Harman's requirement, which (as far as I could tell) was that "such facts would have to explain some non-moral observations." For Harman the concern was that experiment and testing need to be relevant for these facts, and seemingly observation is irrelevant for ethics. Sayre-McCord had a modified version of Harman's requirement that shares the spirit of it. In a certain sense, does this mean that Wright's requirement is more restrictive than Harman's? If so, this could be cheating a bit, since on a less restrictive version of the argument epistemology might be able to get in.

Under the explanatory requirement, ethics is (seemingly) out. This is because ethical facts, if they exist, don't explain any non-moral facts, it seems. And then he quickly moves to "consider a putative epistemic fact... a fact of this sort appears..to no more explain phenomena in nature such as that which biologists and physicists study than does the putative fact "that an act is wrong."" So epistemic facts would also fail the explanatory principle, and wouldn't exist.

Now, I'm not going to hold Cuneo responsible for ignoring the possibility of a potential indispensability argument in support of epistemic realism. He brings this up in Chapter 8. I think it's a serious concern for him. The entire companion in guilt style argument seems flawed, to me, for the following reason. It also starts with two similar areas of discourse, one of which we're much more confident about it being real (for instance). Do we have a good reason to believe that this discourse is real? Yes? Does that reason exist in the other discourse? No? Then isn't this THE important disanalogy between the two areas of discourse?! Am I missing something here? So we should hardly be surprised that there is an important disanalogy that can be used to argue for the reality of one rather than the other. That important difference between ethics and epistemology is exactly the one employed in chapters 4-7, namely, the indispensability of epistemology to theorizing, and hence, to science and the project of explaining our observations. I think that this is an important concern for the defense of moral realism, and a more general concern about what we aim to get out of companion in guilt style arguments.

For now, I just want to express my genuine confusion for why this isn't brought up in Chapter 3. In Harman's chapter in "Ethics and Observation" he brings the indispensability argument for mathematics as a way to account for mathematics relationship with observation. In Quine, the indispensability argument works because mathematics is empirical knowledge since epistemic holism is the case. So evidence for any thing to which math is indispensable is indirect evidence for the truth of math. It is in this way that observation is relevant to mathematical knowledge, and fulfills the spirit of the explanatory requirement.

I wonder if the indispensability argument has been transformed in the hands of Enoch and Cuneo. Both talk more about pragmatic indispensability, as if the strength of the argument is just that we can't blame folks for doing what they have to do. As far as I can tell, that's not quite the way the argument looks in the hands of a Quine or Putnam or a Colyvan. For them, the strength of the argument comes from the fact that any part of a theory is as good as any other part, and if mathematics weren't true then observation wouldn't confirm science, but observation clearly does. Enoch is rather explicit about this, because the Q-P indispensability argument, in his view, is a particular instance of a more general valid family of indispensability arguments, where various things are indispensable to various projects. I suppose that's good as far as it goes, but we should then be rather clear about when we're invoking the particular indispensability argument and when we're talking about the more generic famliy of indispensability arguments.

It seems to me that epistemology is a good candidate for a Quine-Putnam indispensability argument. Though, I probably HAVE to be wrong because Quine himself didn't think so, because he thought that epistemology should be neutered of normativity. But our epistemic beliefs seem to be indispensable for the practice of science and the expression of our scientific theories. Could we do science, or even express our theories, without talking about what constitutes good evidence and bad evidence for a theory? How could we explain any observation at all if there aren't some features that are more relevant--that is, our understanding of what is a natural kind depends, eventually, on some sort of epistemic fact about what constitutes justification for that belief. And so why not say that the very fact that observation and science is possible is good evidence that epistemic realism is true? In other words, does the scientific realist have to accept epistemic realism?

If so, then I think that epistemology fails the explanatory requirement in the same way that math does; by being indirectly relevant for the explanation of observations.

Congrats to Cuneo!

http://www.apaonline.org/opportunities/prizes/book.aspx

Congrats to Terence Cuneo for receiving an Honorable Mention from the APA's Book Prize for "The Normative Web." I've enjoyed reading his work tremendously, and I'm glad to see that it's getting the respect it deserves!

Looking at the list of previous awardees of the book prize, it reminds me how much good work in meta-ethics has come out in the past decade or two. The field is really exciting and I can't wait to see what happens next.

Thought I'm only a college senior, I'm excited by the idea that I can participate in this exciting field in even an extremely minor, insignificant way.

Developing an argument against Cuneo

I intend to continue writing about Cuneo's specific arguments--right now I'm 1/3 of the way done evaluating his extension of arguments against moral realism to arguments against epistemic realism. But I want to jump the gun, because I'm am impatient man. I want to put down on electronic paper what I think the difficulties facing Cuneo are. As always, this is intended with a sense of modesty that comes from knowing that I'm probably wrong, but here we go anyway.

What's the deal with companion in guilt style arguments? They seem to be very difficult to make. Say that we suspect A of something bad, and we wish to vindicate it by comparing it to B, which we don't suspect of something bad. This comparison has to be based on some shared features, and so A and B share a great many features. However there is at least some features that they don't share. How do we know this? From the very fact that our belief in the goodness of B. If B isn't bad, there must be some features that are keeping it from being bad. And that means that B is different from A in some regard. So what good is the argument, then? The argument can show that the bad features aren't that bad, that they don't settle the question of whether A is bad or not. But we can't actually vindicate A, because A is lacking the features that made us secure in B in the first place.

Let's bring this back to ethics and epistemology. What is an companion in guilt style argument going to be able to show? Ethics and Epistemology, arguably, share certain features. These features are bad features, insofar as its been argued that possession of these features makes a ethics less like to be real. These features include metaphysical queerness, queer supervenience relations, lack of access (or some stronger version of that argument, following Enoch). And if Epistemology has them, then the argument goes to show that these features aren't killers, they don't settle the question of whether a discourse is real or not. This works because we're sure (let's say) that epistemology is real, and epistemology has these features.

But why are we sure that epistemology is real? We must have a good reason--otherwise, maybe epistemology has these features because it isn't real! No, we do have good reason to think that epistemology is real. This is because if epistemology isn't real, then we have no reason to believe anything. But we do have reason to believe things! Specifically, there's a big difference between the good ways of knowing things and the bad ways of knowing things. The good ways of knowing things are called science. And so epistemology is real because it has a close connection, and helps to support, science.

(And doesn't this mean that, according to Harman, epistemology should have a place in our world, because if epistemology was wrong then there wouldn't be a difference between our good methods of knowing things and our bad ones. But experiment can confirm that the good ways of knowing things can correctly predict more observations than the bad ones, and our observation that certain methods are better than others needs to be explained. Isn't the most plausible explanation that epistemology is real, and that some methods of investigation really are better than others? Am I missing something--why does Cuneo think that he is right that epistemology plays no explanatory role? I'll deal with this in a post analyzing the explanatory requirement again, but I guess it's because this is just a case of epistemology explaining an epistemic observation. But wouldn't the foundational role that epistemology plays in justifying our ordinary observations make epistemology relevant for ultimately explaining ALL of our observations? I need to think this mess of thoughts through more carefully.)

Ethics, on the other hand, doesn't have this close connection to science, it doesn't help to support science in this way. And it's not surprising to us--or at least, it shouldn't be--that there's an important difference between ethics and epistemology. In fact, the entire companion in guilt argument depends on there being an important difference that is relevant to the realness of a discourse. If there were not such an important difference, then ethics and epistemology would be on the same level.

Now, what I just said isn't necessarily true. It's possible that there could've been an argument whose aim is to reveal that ethics and epistemology are really on the same grounds, and that's because the good feature (what Cuneo calls a redeemer) isn't only present in epistemology, but also in ethics. In other words, maybe ethics also has the nice feature that epistemology does, the feature of supporting science and the possibility of some arguments or theories being better than others. But does ethics really play that role? Probably not. In which case the companion in guilt style argument starts seeming a bit misguided. What were we trying to show, anyway?

Well, here's what Cuneo definitely does show: that epistemic facts would be queer too, and that's because queerness arguments trade on the normativity of ethics, and epistemology is just as normative as ethics is. But how can he manage to show that ethics and epistemology are on the same grounds, as far as realism goes, when that argument seems to depend on there being an important difference between ethics and epistemology that stands in favor of the realism of epistemology?

Cuneo's response, if I understand him correctly, is to suggest that ethics and epistemology are really the same thing. Or that they're close enough that if you've accepted epistemic facts then you've also accepted ethical facts. I think that there are problems with this argument, a lot of them actually, and that leads me to wonder what else Cuneo could be saying, wondering if I've got him right. I need to read him more carefully, and this post is long enough. But I wonder: how exactly are companion in guilt style arguments supposed to work at all? I think I'll start reading the book "Companions in Guilt" to try and change my perspective.

Pet Peeve: Evolutionary Arguments

Whenever I read an evolutionary argument in a philosophy paper--either to defend moral realism, moral anti-realism, to defend epistemic realism or whatever, I basically roll my eyes and say "Evidence, please!" Evolutionary arguments seem to me like one of those "proofs that p":

It's plausible that P being true would confer on humans some sort of evolutionary advantage.
Therefore P.

Sometimes, if the reader is lucky, we get some sort of acknowledgment that phenomena such as genetic drift and the linking of non-adaptive phenotypes with advantageous phenotypes can lead to the former being selected for. Having acknowledged that, the author then has full intellectual permission to wildly speculate over empirical reality.

Unless I'm missing something, this seems grossly unscientific. We need evidence for claims, folks! Especially since we have evidence that just because we can imagine something being advantageous/not advantageous has no bearing on what occurs in reality. In 50 years, won't people be laughing at our evolutionary arguments the way students now laugh at Descartes' speculations about the pineal gland?

Another possibility in logical space

One could reject moral/epistemic realism while maintaining realism about a subclass of morality or epistemology.