In his "Reasonable Religious Disagreements" Richard Feldman uses the principle that "Evidence of evidence is evidence" (lets call it EEE) to argue that (very roughly) one can't use one's own private experience as a way to settle disagreement with an epistemic peer."
Here's a test case you and I are looking out across a field and I (honestly) claim to see a sheepdog about 10 yards off. You say you don't see any such thing. There is nothing in the vicinity with which to confuse the sheepdog, it's fairly bright day, I have (antecedent) reason to believe your eyes are functioning properly, etc. Some would say (and van Inwagen seems to say) that the fact that my view still seems to me to be true after I consider your disagreement, then that persistent belief is justified.
Rich says this can't be because I know that *your* view seems true to *you* and I have no reason to think you are more prone to error than me. He uses EEE to urge this for he says that your statement that you don't see it is evidence that you've had an experience the veridicality of which would entail that I am in error. Thus I have evidence that I am in error.
So far so good, but I don't think it follows from this that I am unreasonable to maintain my belief. Rather, I think a conclusion more like that of Christensen 2007 is correct: that I ought to revise my belief down in light of this new evidence.
I'll continue the discussion below the fold.
The reason is that I am directly acquainted with my own sensory evidence E. So let P(S/E) represent the epistemic probability that there's a sheepdog, conditional on my having the sort of sheep-dogish experience I've just had, let it be p. Then I have the experience. Since P(S/E) will be nearly 1 (there will be *some* uncertainty about the experience to begin with), P(S) should be very nearly 1. Let ~E describe the experience you are reporting, one such that P(S/~E) = 1-p. But I'm not directly acquainted with your experience, so ~E isn't evidence for me. Rather your *testimony* that you've had that experience is my evidence. Let T~E describe your testimony. So there are two ways we could formally represent my point.
First, we could say that P(S/T~E) < P(S/~E) and so < 1-p and so the balance of my evidence supports my belief. Or we could express essentially the same thing by saying that though P(S/E) = P(~S/~E) when we update (using "probability kinematics") we must weight those liklihoods by by P(E) and P(~E) respectively, but P(E) will be higher for me than P(~E) because of how I get support for them. E names or describes (it doesn't much matter how you do it, I like naming but describing is much simpler to work with) the experience I myself host whereas my probability for ~E comes from T~E. There are possible cases, perhaps (though perhaps not), where I'm more confident that your testimony is correct than I am that I've had such-and-such experience (it really depends on how you individuate and refer to experiences), but that's OK since I'm only making a possibility claim here myself.
This post is just programmatic to clear a way for something I want to say later. I'm very interested in trying to describe the sort of "incommunicable insight" as Peter van Iwagen puts it in his anti-Cliffordian paper (there's a new paper of that type forthcoming that I'm now reading as well) that we might have which gives us a reason to favor our own view in disagreements with a certain kind of not-so-uncommon epistemic peer. PvI sometimes talks as if it's a reason and sometimes doesn't. I think it clearly is a reason, just not one that can be expressed in ordinary language (besides maybe mentalese). I do think there's hope for and utility in describing this kind of reason in more detail than PvI has so far.
You say you can't see a sheepdog when I clearly see one there. What better evidence am I waiting for, to demonstrate that your vision is *not* as good as mine?
Heath, there are various notions of epistemic peerage, but I think all relevant notions will involve my having sufficient evidence that you are no more likely to be subject to dysfunction than I am. So suppose I know that *one* of us is subject to serious malfunction but I have evidence that it is no more likely to be you than me, so it would be arbitrary (and so unjustified, epistemically) to go with my appearances rather than yours.
Trent, my point was that “my having sufficient evidence that you are no more likely to be subject to dysfunction than I am” is not generally independent from how things appear to me. If an angel tells me that either Joe or I have a false belief about the sum of 2 and 2, and I think it’s 4 while Joe thinks it’s 5, then clearly Joe is the one with the problem. Likewise with the sheepdog observation case. Joe will disagree, of course, but so what?
Philosophical arguments (including those for religious conclusions) aren’t much like this, I don’t think. There, two people are considered epistemic peers because of inferences from their intellectual (-cum-moral) character as exhibited in other cases. That is, we think that because Joe is just as good a philosopher (in other areas) as I am, Joe and I are epistemic peers, so our respective opinions about an argument should carry equal weight.
One problem with this approach is that, in the current philosophy scene, intellectual virtue is measured procedurally rather than in terms of outcomes. In other words, you can think someone is an excellent philosopher while also thinking their views are completely wrong. Nobody would take this approach to mathematicians or perceivers; why we take it with philosophers is a (serious, large) question. But tentatively, I would say that this fact tends to undermine the EEE principle as applied to philosophical arguments of any kind.
I should say I think there is a real problem about religious disagreements; I just don’t think (yet) the present suggestion gets at its core.
Heath,
Appropriately, it seems we disagree. :-)~ I think the way I've described the situation is just a corollary of a truism: that a proposition is justified for one to the extent that it fits one's evidence. Our disagreeing is evidence that you are mistaken, but it's also equally good evidence that I am mistaken. If I have no evidence to break the tie then it just follows that I cannot maintain my belief with justification. Most attempts to avoid this just end up attributing more evidence to one of us.
We have to distinguish between two different ways of filling out the Angelic Math case. It can be embedded within an ordinary scenario where I remember being taught that 2 + 2 = 4 and that this is a common example of an obvious truth. But then we're not epistemic peers in the sense I intended because my background evidence makes it more likely that you are mistaken. I almost indexed to time t to make that more clear but I wanted to avoid the extra formality (I've already been criticized by the guy setting next to me that I was too formal!). If *at t* the time of the recognition of the disagreement, I have no better evidence that you are mistaken than I am then it is arbitrary and unjustified to think it's you rather than me.
Or we could embed the example in a way that brackets ordinary background evidence. Here it would be better to switch to, say, the corresponding conditional of modus tollens, seen for the first time by two individuals who have otherwise common knowledge (in the formal sense). Then I think it's more clear that it's unjustified for me to go with my intuition that it's valid over your intuition--or Joe's--that it isn't. Most attempts to circumvent this just point to features of my evidence that make us not epistemic peers.
That I bother to make this reply to Rich should not be read as an affirmation that I--or he--think that this is the locus of the trouble disagreement causes for religious--or other--beliefs. I don't think it does--nor does Rich--but it an interesting species of the problem and, we think, worthy of discussion. It is also an interesting case of a consequence of evidentialism and might help us think more clearly about the "real world" problem of religious disagreement once someone tells us just what that problem is supposed to be. It should be noted that Rich started thinking--and writing--about this primary in response to a Plantinga piece on religious disagreement where he seems to stick his opponents with the kind of view that Rich--and I--have as targets here.
I'm not sure what you have in mind by the comparison with Mathematics or perception. Clearly enough the matters that philosophers discuss are less subject to clear verification than mathematical or perceptual matters (here I'm with PvI) (and to the extent that they get esoteric--like intuitionism or constructivism generally--then I think it becomes philosophical enough to accept reasonable disagreement (in the sense I think you specify). Philosophical arguments are more like arguments in interpreting literary texts or arguments in history or forensic science or even largely theoretical physics. Those are all areas where a good bit of reasonable disagreement is expected. Philosophical disagreements are bigger than those, but that's fine because philosophy is further down the spectrum from them, but that there is a spectrum and that most philosophy--most non-formal metaphysics anyway--is on the opposite side of pure math and perception is, I take it, clear enough.
I hasten to add that I don't think there isn't a problem for philosophical disagreement. I think a lot of people are committed to more charges of unreasonableness than they are comfortable making. So my hypothesis is that in might just be a social problem. In our environment we just don't want to say that sort of thing. I've only read two or three things from David Lewis (with a bit of Kitcher) on religion but it seems to me he's not taking the arguments seriously (this might not be a good comparison because there is evidence that Lewis was not widely read on the subject and even for one of the three smartest philosophers of the latter half of the twentieth century you can't just know this stuff a priori (this is a *big* problem in the I'm-a-smart-atheist-so-I-can-pronounce-on-religion industry)). I could point to specifics if I had to. I could make arguments of the kind
(1) Discourses with features F and G are positively correlated with not having taken an argument seriously.
(2) Lewis's discourse on religion has features F and G (here in this set of sentences from this article and here' in this' set of sentences from this' article).
Therefore, probably,
(3) Lewis's discourses on religion do not take this argument seriously.
Maybe I'll even have time to do that (but I doubt it) or maybe I shouldn't take the time because I'll be maligned as intolerant or insubordinate or impertinent. Maybe if I was so charged I would mount a defense that I have an explanation about how really smart and otherwise careful philosophers can come to have terribly skewed priors.
It'll all have to remain in the abstract for now though, I have to make an exam I'm pledged to give in a Stats class tonight!
PS - I don't even have time to proof read this, so please forgive me if there are howlers in here (grammatical or philosophical!).
Trent,
I was wondering if you've read Goldman's paper on relativism and disagreement
http://fas-philosophy.rutgers.edu/goldman/Papers.htm
I personally think that the distinction between animal and reflective can be applied here (Sosa), and that if I have animal justification, even though I cannot persuade nor can I give a reason for my belief over someone's, I am still justified. This makes sense in light of Sosa's view that aptness can be fragile. Of course that means one has to be persuaded of the animal and reflective distinction and even of Sosa's view.
Apolino,
Sorry for the delay. I'm fine with a distinction between animal knowledge and reflective knowledge. I think one has (or can have) both *before* one shares one's perspective with a cognitive peer. Then, when the cognitive peer becomes an evidential peer and the disagreement persists, one's perspective now includes a defeater for the evidence which gave them the animal knowledge in the first place.
Hi Trent,
I've been thinking about this issue quite a bit lately, and I have a thought that, although different from yours, is somewhat similar and might aid your general line of thought. First, though (as you're probably aware), Feldman dealt with your move in his paper "Epistemological Puzzles about Disagreement" (p. 224). Here's what he says: "the doubts about the existence of the (apparent) insights or intuitions of the conversational partner are really extremely minimal, far too weak to make one's overall evidence have the desired characteristic [i.e. justify one's own belief]" So, the idea is that although the evidence may favor one's own belief somewhat in light of your better access to your own intuition, it won't favor it by enough to produce justified belief; the degree of justification in favor one's own position will still warrant suspension of judgment.
Second, here's my thought, and I'm not sure that it is vulnerable to Feldman's above charge. Testimony is another form of private evidence, that is, evidence that cannot be directly shared. If someone I trust says p, I can believe p, but if I tell you that someone I trust believes p, it doesn’t automatically follow that someone you trust says p. In other words, you don’t acquire the same evidence that I have – i.e. having someone you trust say that p. Suppose, then, that Steve and Joe each have a community that agrees with them. Each has extra reason from testimony for believing what they do and this evidence is not transferable. Not only does the testimony provide direct evidence for p, but it also reinforces whatever other reasons one thought one had for believing p; it gives one a reason for relying on one’s own evaluation of the evidence rather than the evaluation of someone who disagrees.
Suppose, though, that Steve and Joe trust each other on most matters. Since they trust each other, shouldn’t they also trust those whom the other trusts? To some degree, but the degree of credence you put into S’s testimony surely decreases the greater the number of testifiers that intervene between you and S. Take Steve. Steve has a community that he trusts, and through Joe, he also to some degree, though to a lesser degree, trusts Joe’s community, which disagrees with Steve’s opinion. But, he trusts (and he should) his own community more, so puts more weight in their opinion, thus outweighing the testimony of Joe’s community. His degree of confidence in his opinion should go down some, but overall he still has more evidence for his own opinion.
What do you think?
Josh
Josh, LOL it's funny you mention that because I got him to admit that he was probably wrong about that--I have witnesses. For one thing, it contradicts this:
EC Believing is the justified attitude when the person’s evidence *on balance*
supports a proposition, disbelieving is the justified attitude when the
person’s evidence on balance supports the negation of a proposition,
and suspension of judgment is the justified attitude when the person’s
evidence on balance supports neither a proposition nor its negation.
Feldman and Conee 2004, 102, emphasis added.
Regarding the testimony I'm pretty sure I can tell you what Rich would say. We are both reductionists about testimony. So who trusts whom is irrelevant, it's all about what evidence I have that the asserter is reliable. Then it just all collapses into whether we share evidence regarding the testifier. If we are epistemic peers we must, and if we do, then it's not reasonable for one of us to trust him and the other not. For the testimonial reductionist your example will work the other way round. We'll argue from peerhood and evidentialism that one of you is unreasonable, either the truster or the mistruser.
Trent,
Thanks for responding. Regarding testimony, I see your move, but it seems to render it ridiculously unlikely for two people to be epistemic peers because then they would have to share all the same evidence regarding the trustworthiness of all the same people. There are surely possible worlds where this happens, but it rarely (if ever) occurs in the actual world.
Call people who share all first-order evidence about p (i.e. evidence about p that does not include the testimony of others) First-Order Epistemic Peers. Call people who are first-order epistemic peers and who share all the same testimonial evidence Complete Epistemic Peers. Here, then, would be the response to Feldman: true enough, Complete Epistemic Peers cannot rational disagree, but none of us, in actual disagreements, are complete epistemic peers. At best we are First-Order Epistemic Peers. But, if my argument in my previous post is correct, First-Order Epistemic Peers can rationally disagree. So, people in real-world disagreements can rationally disagree.
How does that sound?
Josh
That's right Josh, the skeptical results of rational disagreement might not have that much real-world bite. Good news. Still, when you read PvI he tries to make a good case that it does regularly occur and note tha if the skeptical result follows in the identity case then partial skepticism holds in the partial identity cases which are presumably much less rare.
Kelly makes the 2nd-order move in two papers, one in about 2005/06 and one forthcoming. Right when I got your note I was reading an MS by Earl Conee where he also tries to make that move. I don't think it works because it focuses on reutting defeat rather than undercuting defeat. It also depends on the principle that if I'm rational to believe that I'm rational to believe p then I've got a reason for p. But that's false.