April 17, 2004

Ecumenicism, Internalism, and Metatheory

Over in Fake Barn Country Allan Hazlett attacks the Ecumenical Solution to the epistemological internalism/externalism controversy:

(ES) There's two concepts of justification, one internalist, the other externalist. They're both useful, they're both concepts of justification, and each serves to capture some (but not all) of our intuitions about justification. Epistemologists should proceed to examine both, and end this bickering over the "real, correct" account of justification.

Asks Allan, why should we find ecumenicism any more convincing here than for Newcomb's problem? (Or ontology, but I find it convincing for ontology, so I'll pass.) We can say: "One-boxerism accounts for some of our intuitions about rationality (the "if you're so smart how come you're not rich?" intuition); two-boxerism accounts for some others (rational actions cause their intended outcomes)." But nobody's going to find that satisfying.

"Why Does Justification Matter?"* [SEE UPDATE IN EXTENDED ENTRY] is meant to give a framework for ecumenicism that doesn't simply depend on the fact that different notions of justification each imperfectly capture our intuitions. --In fact, I think intuitions about justification are cheap, because "justification" is a philosopher's term. (On the FBC thread Jamie Dreier cites Stew Cohen making this point, so I can cite a lot of authority here!) The argument concerns not so much what "justification" should mean as what epistemological properties we should care about. There's more of it below the fold.

But Allan effectively raises the question: how can I even say that there are two different properties you can care about? Mightn't the person who cares about one property be wrong; the way one-boxers are convinced two-boxers are wrong?

My first thought was that the hypothetical person in Newcomb's Problem is faced with an exclusive decision: Take one box or two. That's not true with respect to justification; you can have your cake and eat it too, by stipulating that "justification" has two different senses. Subscript them if you like.

This looked to me like a distinction between practical and theoretical reasoning. But I don't think it is; I think it's a distinction between normative theory and metatheory. Normative practical reasoning issues in commands "Do X" or "Do Y"; normative epistemology issues in commands "Believe P" or "Believe Q." Meta-practical rationality tells us "It is good for actions to have property D"; meta-epistemology tells us "It is good for beliefs to have property G."

In metatheory, there's no reason an action/belief that has one good-making property has to have them all, so we can just say that there is more than one good-making property. In normative theory, you have to fish or cut bait: do it or don't do it, believe it or don't believe it. You can't be ecumenical about it.

Now, the Newcomb problem may seem as though it's metatheoretical. We're discussing what conception of rationality to use--does rationality depend on your actions' expected utility or their expected effects? That sounds as though it's ripe for ecumenicism. But it's not--because the point of saying that a certain action is rational is to say that you should perform that action--period. It's not to say that it has one good-making property but perhaps not another.

Perhaps you want your concept of justification to work the same way--it tells you what to believe, period. No room for saying "this belief is good in this way but not good in this way." Then, I think, there's substantial pressure toward internalism. Because if you ask yourself the question "What should I believe?" you have to be able in principle to figure out the answer. That won't happen if the answer depends on things outside your experience. The commands issued by your conception of justifcation will be commands you can't understand yourself.

(This doesn't happen for one- versus two-boxing. There the subject has enough information to make her choice, whichever criterion you use.)

So my attempt at ecumenicism is just an attempt to sneak internalism in the back door. Well, my internalism (such as it is) stems from the conviction that a criterion on belief isn't much use if you can't use it at least as a regulative ideal for inquiry. Many will say that knowledge is such a regulative ideal, but I think striving to make our beliefs true is as good as striving to make them knowledge.

Here's the argument of "Why Does Justification Matter?" in brief: The idea is that, if you say property P is epistemologically important, you're endorsing P as a property that people's beliefs should have; and you're effectively advising everyone, "Make sure that all your beliefs are P, and that you have all P beliefs." Advice has two dimensions of felicity; it should guarantee achievement of your goal, but it should also be something you can do. So there are two dimensions to evaluating P's epistemological importance: (1) How likely is it that P beliefs will be true? (2) To what extent does P depend on factors independent of the subject's experience (the less, the better)?

Truth is the best property with respect to factor (1). An internalist notion of justification is the best property with respect to factor (2) (actually, I call it "experience-dependent" to avoid some baggage of internalism; I don't want to commit myself to the idea that experience is internal in any way, let alone that justifications have to be available to conscious processing). And though I don't stress this in the paper, externalist notions of justification will be intermediate with respect to (1) and (2).

*The link is to the penultimate version; I haven't put the final version online yet, partly for copyright purposes. Also, the PDF I linked doesn't work on all computers (I think not in Mac OS X); here's a link to the Word version. If you can't read the Word, either, let me know and I'll e-mail you the text!

Posted by Matt Weiner at April 17, 2004 03:02 PM

Your paper looks very interesting, and I'd like to read it, but the link you posted seems to point to a pdf file full of jibberish. Mostly boxes and H's, with a few Q's and other letters thrown in.

Posted by: Jonathan Ichikawa at April 18, 2004 02:51 PM

Hmm... either you fixed the problem, or the problem was with the computer lab computer's Acrobat Reader -- I'm getting it now from home just fine.

Posted by: Jonathan at April 18, 2004 03:57 PM

It's been pointed out to me that my PDFs don't show up on some versions of Acrobat--I think the ones that get used with OS X and later. The conversion I use (Neevia's online version) doesn't have an option for including the fonts, and I don't have access to Acrobat.

If anyone knows of an online pdf converter that lets you include fonts, I'd appreciate hearing about it.

Anyway, as a stopgap, I'm putting up a link to the word version; hope that covers most people!

Posted by: Matt Weiner at April 19, 2004 08:52 AM

Check out Ghostscript, a Unix application that's been ported to other platforms (if, as I suspect, you're still a Mac user: I believe the current Mac port is called "MacGhostscript," but I may be misremembering; just Google "Macintosh" and "Ghostscript.")

Posted by: bza at April 21, 2004 02:23 PM