Wednesday, June 27, 2007

On Evidence and Irrationality

One often hears the claim that Christians are, since they do not apportion their belief to the evidence, irrational. This must be carefully distinguished from the claim that Christianity is false. Though the claim that it is false might well entail that belief in it is irrational, they are yet distinct claims. The first is epistemological in kind, the second metaphysical.

I wish now to consider the epistemological claim - that claim that, since Christians do not apportion their belief to the evidence, they are irrational.

"Consider" is perhaps not the best word. "Attack" is better. Now, one might attack the epistemological claim is various ways. One might do so as did Thomas. He would attack my means of arguments meant to establish this or that Christian truth. (For my own part, I'm suspicious that arguments such as this can be made to work against the religious skeptic.) However I won't take the Thomas path. Instead I'll take on the charge of irrationality by means of a tu quoque. (Tu quoque is a form of rebuttal in which one claims that one's opponent violates some principle that he has put forward. This leaves one's opponent with two choices: (i) take back the principle, or (ii) take back the claim or claims he's made that violates that principle.)

Call the one who wishes to defend the claim of Christian irrationality in the way I've describe "Clifford". (William Clifford famously made the charge that Christians believe where there is no evidence and are thus irrational.) I think it clear that Clifford means to endorse some such principle as this:
One ought in all cases to apportion one's belief to the evidence available.
Call this putative principle "P". Now let us ask what seems an obvious question: What is the evidence available for P? I would guess that there is none on offer. (William Clifford certainly gives us none. Nor do any of the others of whom I know give us any reason to believe it.) Indeed it is difficult to think of what might count as evidence in favor of P. Let me explain.

Evidence for the truth of some claim is a set of propositions (perhaps a unit-set) the members of which are (i) all true, and (ii) together tend to show that the claim is more likely than not. (This isn't a complete definition, but it does state two necessary conditions; and those two are all we'll need.) Now, what might a set of propositions be that would tend to show P more likely than not? They would have to be epistemological, for P is epistemological; and they would have to be, in some sense, more basic or fundamental than P. (From the more basic or fundamental we derive the less basic or fundamental.) But if P were to be true, there would be no more basic epistemological principle from which it could be derived. It is the sort of principle that, if true, is absolutely basic. There is nothing "beneath" it which has the potential to prove it true. (Or so it seems to me upon reflection. My only argument here is this: I've looked and haven't found anytying more basic than it from which it might be derived.)

So, then, here's where we're at: very likely P has no evidential support. But if this is so, P is, if true, not to be believed. For if P is true, nothing - and this includes P - is to be believed without evidence. Where does this leave us? P ought not to be believed. For P is either true, or it is false. If true, then as we've seen it's not to be believed; and if false, then (as should be obvious) it's not to be believed. Conclusion: belief in P is irrational.

This is our tu quoque. Very likely the principle brought to bear upon Christianity - the principle P - can't pass the test that it itself sets up. The upshot of all this should be clear: we must all believe something without evidence if we're to believe anything at all. (This was also argued for here in a direct way. The argument of this post is indirect and proceeds by way of refutation of the contrary opinion.) We must, as it were, all strike out into the evidential void. Does this mean we're all irrational? It does not, for as we've seen the claim that lack of evidence entails irrationality is simply not to be believed. There is of course a lacuna here. We still need it explained to us how we can rationally believe a thing when we have no evidence of its truth. But I will not attempt to fill the lacuna here. Nor will I attempt to divide propositions into those that are in need of evidential support and those that are not. These tasks must be left for another day.

We must all strike out into the evidential void and plant a flag. I've planted mine. Clifford has planted his, but refuses to admit that he's done so. He says that the evidence must dictate where one plants it, but this very claim lacks evidence. If the Christian plants his flag arbitrarily, so does Clifford.


SteveK said...

Tu quoque is a form of rebuttal in which one claims that one's opponent violates some principle that he has put forward.

I didn't know this had a name. I think I used this form of rebuttal to show doctor(logic)'s definition of "objective" rendered the laws of logic non-objective, so his definition must not be correct.

I assume you read what I wrote at Tom's blog. Was I correct in my approach to this?

Peter Lupu said...


Let us examine your argument here.

You identify the following principle as one of the premises of the "Clifford" type arguments: namely, arguments that maintain that Christians are irrational because they violate an epistemological principle:

(P) One ought in all cases to apportion one's belief to the evidence available.

You then proceed to argue that since we do not have any evidence for (P) itself, at least one of the premises of the Clifford-type arguments is false; namely, premise (P). So either we have to reject (P) itself, in which case Clifford -type arguments are invalid, or we must admit that it is irrational to accept (P) in which case Clifford-type arguments themselves are irrational.

Let us examine another principle akin to (P):

(L) One ought never to knowingly accept a logical contradiction (e.g., any proposition of the form P & not-P).

What evidence do we have for (L)? Should we look for evidence in the actual belief structures of people? Not really and for two reasons. First, the actual behavior of people cannot support a normative principle no matter how wide-spread it is. And, second, as we know quite well we frequently accept conflicting propositions, even knowingly, while at the same time accept (L).
Is there any other type of evidence that would support a normative principle such as (L)?

Well, we can identify certain indirect evidence. It takes the form of examining what happens when one violates this principle. For if one violates this principle then one is committed to believe every proposition whatsoever (because every proposition follows from a contradiction according to Classical logic). And if one believes every proposition, then the whole concept of belief becomes useless.

What should we conclude from the case of (L)? I think one thing we can conclude is that the type of evidence suitable for "normative principle" is totally different than the kind of evidence suitable for "descriptive" or "factual" propositions or principles. And since (P) is a normative principle, we ought to evaluate its merit by seeing what would happen if we were systematically fail to conform to it.
I therefore conclude that your argument in this post is inconclusive at best because it does not take into account that (P is a normative principle and therefore requires special type of evidence.


Franklin Mason said...

We seem to disagree about the nature of evidence. I'll lay out my view.

There are not multiple kinds of evidence, one appropriate for one class of belief, another appropriate for another. Rather there is one and only one evidential relation. Granted there are many classes of belief that can be grounded in evidence. But to be grounded in evidence is, in all cases, one and the same thing.

p is evidence for q means something like:
1. p is known.
2. p's truth raises the probability of q's truth.
3. 2 is known, and it is part at least of the reason why q is believed.

Thus if I were to follow your suggestion about the consequences of belief, I would ask myself whether the beneficial consequences (however this should be construed) of a belief serves to raise the probability of its truth. My reply: I doubt it. It seems quite easy to suppose that beneficial consequences might coexist quite happily with total falsehood; indeed it would seem to me a happy coincidence if beneficial consequences and probability of truth were to coincide.

Side note. A agree that Clifford's evidential principle is normative; and this makes me quite suspicious that its truth should consist in, or rest upon, good consequences. I'm a Kantian about ethical obligation; so too do I think that epistemological obligation is not a matter of consequences but of simply doing the right thing as regards belief.

Peter Lupu said...


(A) In your response you make two principal claims:

(i) There is only one form of evidence for all beliefs.

You say:
"There are not multiple kinds of evidence, one appropriate for one class of belief, another appropriate for another. Rather there is one and only one evidential relation."

(ii) You argue that the consequences of beliefs, whether beneficial or not, are not a guarantee or even a sign of its truth.

(B) I shall respond to each claim in turn.

Response to (i):

Your claim that there is only one kind of evidence for all beliefs regardless of differences among the character of such beliefs appears to be false.
Thus, consider the following list of beliefs or propositions:

(a) Mathematical.
(b) Logical.
(c) Analytical:
(e.g., "All bachelors are males")
(d) Necessary:
(e.g., certain identity statements such as "water is H20" [assuming both terms are rigid])
(e) Normative:
(e.g., Rational Principles: One ought to accept those logical consequences of one's beliefs that one recognizes as such;
Moral Principles: Ought implies can.)
(f) Empirical:
(1) Empirical Laws that are an essential part of a well established theory;
(2) Empirical Generalizations;
(e.g., "All ravens are black").

Beliefs/propositions of type (a) and (b) above typically rely upon evidence consisting of deductive proofs (this comment should be reconsidered in light of Kripke's claims about the possibility of a posteriori but necessary propositions);
(c), (d), and (e) rely upon a combination of deductive as well as certain suitable conceptual considerations. It is difficult to see what empirical evidence would support a proposition such as "All bachelors are male"?
(f) above is once again different in character from the rest and (f)1 and (f)2 are also to be carefully distinguished. Evidence for (f)2 is the only case that is clearly within the province of your account of evidence as described by your clauses (1)-(3) of your definition
"p is evidence for q means something like:"
But it is far from clear that this definition is adequate even for (f)1, for I do not think that Lawlike proposition of powerful theories are supported simply by a probabilistic relationship between the laws of the theory and its observational consequences. Your model of evidence that closely resembles the Logical Positivist account of confirmation has been shown to be in many ways inadequate and incomplete.

(C) Response to (ii):

Of course the consequences of a proposition, whether beneficial or not, do not in general coincide with truth. But this response misconstrues my point as stated in the post to which you replied. First, in that post I have not used the word "consequences", beneficial or otherwise.
And, second, my example of the normative principle L given there was not meant to illustrate anything about the consequences of this or that belief. My point was conceptual:
i.e., in order for the concept of belief to make sense, there has to be a distinction between propositions believed and those that are not believed. If the set of all meaningful propositions coincides with the set of all propositions believed, then the distinction between the concept of belief and that of a propositions is a distinction without a difference.
So I was arguing there that in order to maintain a conceptual difference between the concept of belief and the concept of proposition the principle L must be accepted. And it is this conceptual fact that supports L.