Photocrat

Well, we're debating axioms here, it's just that they're the axioms we use to derive knowledge from our observations. Of course, we then turn around and look at what those axioms imply about our observations to retune them...

In any event, I do think that our experience shows that there can be true but unprovable things, whether or not they have anything to do with Godel. A trivial example is being caught looking guilty of something, when you know that, in fact, you did not do it [but when a reasonable person would believe that you had] since it's really hard to prove innosence in some cases.

bd-from-kg

Kenny:

I’ve been waiting for your reply to my earlier posts before responding, but since it appears that this is going to take a while, I think it might be productive to look at Plantinga’s concepts of “warranted”, “rational”, and “properly basic” belief from a slightly different angle.

1. Plantinga and the Great Pumpkin redux.

Take virtually any proposition P which is not self-contradictory and does not have “sufficient defeaters”. Define D(P) as:

D(P): I have a strong predisposition to believe P as part of a set of cognitive faculties which are part of a well-designed plan aimed at the production of true beliefs in the type of cognitive environment in which they were designed to function.

For a great many P there is clearly a possible world which is evidentially indistinguishable from this one and in which P and D(P) are true. In such a world I would be warranted in believing P in a non-evidential manner. Moreover, on your showing I would be rationally justified, since warrant entails rational justification. Finally, this belief would be properly basic. In other words, my belief in P would be rationally held in that world regardless of the lack of evidence for it.

Since this is true for virtually any proposition for which there are no “sufficient defeaters”, we have:

For a great number of propositions P, a non-evidentially based belief in P may be rationally held in some possible world evidentially indistinguishable from this one.

However, if I understand you correctly, the fact that a belief might be rationally held in some world evidentially indistinguishable from this one doesn’t make it “defensible” to believe it in a “properly basic” manner. For this another criterion must be met:

(iv) If it is true, there is a high probability that it will have warrant for those who believe it in a properly basic way.

But this doesn’t help much. To see why, let’s start by defining the propositions Q(N) as follows:

Q(0): P.
Q(N+1): Q(N) and D(Q(N))

Now define S(P) as the conjunction of all of the Q(N). Now if a powerful being designed me to have a strong predisposition to believe that my belief in each of the Q(N) is part of a well-designed plan, etc, it is highly probable that it designed me to have a strong predisposition to believe the conjunction of all of the Q(N) as part of this well-designed plan. Thus, if I believe S(P) in a non-inferential manner, and if this belief is true, there is a high probability that this belief is warranted. So S(P) qualifies as a “properly basic defensible” belief.

[Note: The kind of infinite chain of beliefs represented by S(P) poses no real problems provided that we do not require a belief to be conscious in order to qualify as a belief. In fact, there are lots of such chains to which I think you would have no objection. for example: (“1 + 1 = 2”, “1 + 2 = 3”, “1 + 3 = 4”, etc. Or: “My name is Kenny”, “My belief that my name is Kenny is warranted”, “My belief that my belief that my name is Kenny is warranted is warranted”, etc. It’s generally accepted nowadays that people do have an infinite number of beliefs precisely because of such infinite chains of beliefs. And I’m not assuming that all of the Q(N) are true – only that it’s possible to believe that all of them are true. But in case you have some objection to this, you can simply terminate the string at some reasonable point (say N = 3 or 4) and observe that given Q(N), Q(N+1) is highly likely to be true, so that if Q(N) is true, it is highly likely to be warranted if believed in a properly basic manner. This doesn’t really change the argument significantly. The advantage of the infinite string is merely that the step to “highly likely to be warranted” is even more plausible.]

Now I gather that what you mean by calling a belief “properly basic defensible” is that one is “within one’s epistemic rights” or is not “violating one’s epistemic duty” in holding it in a non-inferential manner. In other words, you mean essentially what I mean by saying that it is rationally justifiable. But of course, if a belief in S is rationally justified and S entails P, belief in P is also rationally justified. And S(P) does indeed entail P.

So what we have shown is that (on your showing) it is rationally justified to hold any of a great number of incompatible beliefs in a non-inferential manner! This might be called the “Grandson of the Great Pumpkin” argument.

2. Defeaters

Your answer to all such arguments seems to be based on the concept of “defeaters”. Essentially you seem to argue that there are only a few propositions of the sort that I’ve described – i.e., ones without “sufficient defeaters”. But frankly, I haven’t been able to get any sense of what “counts” as a defeater.

For example, in the case of letters spelling “Welcome to Ohio” carved on a hill, you say that evidence that the letters weren’t formed by an intelligent agent is a defeater for the belief that they convey information. But why is this a defeater for this belief? It seems to me that at most it removes one possible reason for believing it: if you had evidence that they were formed by an intelligent agent, this might be evidence that they did convey information. In other words, the fact that they weren’t formed by an intelligent agent means, at most, that there’s no evidence that they convey information, but if that’s a “defeater”, then for any proposition P (other than those that are essential preconditions for rationality), the fact that there’s no evidence for P is a defeater for P. I agree, of course, but this is the very thing that you dispute!

The only other example you give of a “defeater” is in the context of your discussion of the “Great Pumpkin” objection, where you say:

Now this is a really weird notion of what constitutes a “defeater”. After all, you agree that widespread belief in a proposition, at best, weak evidence in its favor – certainly not enough to justify belief in it. Yet now you want to argue that the absence of widespread belief is sufficiently strong evidence against it to justify disbelief? You can’t have it both ways.

Also, the fact that belief in the Great Pumpkin is considered irrational is based on two things: it seems bizarre of the face of it and there is no evidence for it.

As for “bizarreness”, many of us nontheists consider belief in God to about as bizarre as belief in the Great Pumpkin, and belief in the Christian God to be inexpressibly bizarre. The fact that this belief is widespread is irrelevant. Belief in astrology, alien abduction, palm reading, clairvoyance, and demonic possession are widespread but bizarre. Belief that flying planes into tall buildings will get you the devoted services of 72 beautiful virgins for all eternity is widespread in spite of being incredibly bizarre.

The second point, the lack of evidence, is in any case by far the more important, since if there were significant evidence for it, it would not be regarded as bizarre. (Look at general relativity and quantum mechanics!) So once again this “defeater” seems to come down to the fact that there is no evidence for. the belief.

It seems to me that both of these examples are abusing the notion of a “defeater”. The absence of evidence for a belief is not ordinarily significant evidence against it. And anyway, as I pointed out above, when you argue that lack of evidence for a belief is a “defeater” for that belief, you’re shooting yourself in the foot.

More generally, it seems to me that what constitutes a “defeater” depends on one’s presuppositions. In other words, although it might seem to you and me that something is a defeater for belief P, it might seem to someone who believes P that it is not a defeater. Again, your second proposed “defeater” is a good example. It’s based on the premise that “our cognitive design plan seems to include within it an element of social interaction”, the point being presumably that this is evidence that our cognitive design plan does include an element of social interaction. But it would be easy to imagine a belief that entails that this “seeming” is actually an illusion, and that our CDP does not actually include any such element. In that case the sort of thing that you propose here as a “defeater” wouldn’t be a defeater for a “true believer”. In fact, it wouldn’t be all that difficult to come up with beliefs that are immune, by their very nature, to all defeaters and which are such that, if they were true, it is highly probable that they would have warrant for anyone who believed them in a properly basic way. According to Plantinga, it would be rationally justifiable to take any such belief as “properly basic”. And this would be true no matter how “bizarre” or “irrational” they seemed to everyone else.

For these reasons, among others, it seems to me that Plantinga’s beliefs about the nature of rational belief are themselves bizarre and irrational. They make a mockery of the whole concept of “rationally justified belief”.

The bottom line is that if what you mean by saying that it is rational to believe in Christianity (or theism in general) without adequate evidence is that it is a “properly basic defensible belief” in Plantinga’s sense, you might be right, but since this has nothing whatsoever to do with what I mean by saying that it is not rational to believe in Christianity or theism without adequate evidence, I just don’t find the claim interesting or significant.

Kenny

Bd-from-Kg,

Sorry it's taking me so long to reply. The truth is I haven't really even had time to work on my response lately. I do intend to reply, but please be patient as it may be quite awhile.

bd-from-kg

Kenny:

I understand. It’s just as well that you haven’t been posting very actively since I haven’t had a lot of free time myself, partly because of the incredible weather here.

But I might as well “catch up” at this point and then “give you the floor” when you find the time to reply.

In this post I’m finally going to reply directly to your last long post, except for a very general issue that I’ll reserve for the next post (which will be the last until you reply).

1. On being “aimed at the production of true beliefs” in a particular “type of environment”

One of the things that makes Plantinga’s criterion for warrant so hard to argue against is that it’s so slippery. It seems at first sight to allow clearly nonrational cognitive processing, but whenever I give an example to illustrate this problem, the reply is always that the cognitive faculties in question won’t reliably produce true beliefs in a sufficiently different possible world, so they aren’t really “aimed at the production of true beliefs” after all.

The problem here is that this is precisely what makes nonrational cognitive processes nonrational! Rational cognitive processes are precisely those that “work” in the widest possible range of possible worlds, and that do not depend on the world’s having any “special” characteristics. If you’re going to rule out all cognitive processes that fail to produce true beliefs when they’re placed in an environment that isn’t “tailored” for them on the grounds that they aren’t “aimed at the production of true beliefs” in a sufficiently wide variety of environments, you end up with my definition of “rationality”, and thus of “rational justification”. But if you don’t rule out all nonrational cognitive processes on these grounds, at some point you have to accept an example of a being who arrives at conclusions that we would say are clearly not rationally justified given the evidence available to him, and argue that nevertheless these conclusions are rationally justified.

So far your only attempt in this direction has been the “Jim/Bob example. But this example isn’t really very relevant because you arranged for both Jim and Bob to have weird “inputs” which make it very difficult to say what, if anything, the “rational strategy” would be in such a situation. (In other words, it’s hard to say just how my description of the “rational strategy” should be applied since it’s not clear what should be counted as the “memory” that I argue must be presupposed to be reasonably reliable.) So here’s a challenge: Give us an example of a being who clearly violates the “rational strategy” in the way he processes “normal” inputs and as a result arrives at conclusions that are clearly different from those that an ordinarily “rational” human being receiving the same inputs would, but whose conclusions you would nevertheless consider “rationally justified”.

2. On Jim and Bob

Your example of beings with unreliable “memories” but with a special form of “clairvoyance” that serves the same function is interesting, but it has some problems. First off, there’s the assertion that the “clairvoyance” possessed by these beings is a “form of sensory perception”. Now by definition something can only be “perceived by the senses” if it’s there. And the past isn’t “there” in the requisite sense. So this is not a form of sensory perception. So if it isn’t sensory perception, what is it? Why, it’s a form of memory, of course. So what these beings actually have is two different kinds of memory.

Even if you insist that this “clairvoyance” is a form of “sensory perception”, the fact remains that it’s a form of (apparent) “direct” access to the individual’s past. To me that’s what it means for something to be “memory” regardless of how it “works”.

Now let’s consider the question of whether your beings are justified in rejecting what you cal their “memory” (but which I call their second form of memory) in favor of what you call “clairvoyance” (but which I call their first form of memory). More generally, let’s consider a case where someone (call him Jim Bob) has two sets of “apparent memories”, M1 and M2, where each is a more or less complete “apparent record” of his past, but which conflict violently. Let’s also assume that there is no objective basis (such as internal consistency) for choosing between them, and that no matter what we do to “test” which memory is “accurate” (i.e., most congruent with our present experiences), either both will “pass” the test equally well, or the two sets of memories will immediately disagree regarding the results of the “test”.

In that case I’d say that Jim Bob has no non-arbitrary basis for trusting M1 over M2 or vice-versa. In fact, he has no basis for trusting either M1 or M2. He certainly cannot follow the “rational strategy” of presupposing that his memory (consisting of M1 and M2) is generally reliable since it is massively inconsistent. Since his basic “pre-cognitive” faculties (i.e., the faculties that produce the “inputs” for cognitive processing) are hopelessly out of whack, I’d say that he is incapable of forming any rationally justified beliefs at all about the “real world” (although he might still be capable of forming RJ beliefs about tautologies).

The same kind of situation would occur, of course, if one were to experience two completely disjoint streams of “apparent perceptions”: visual images, sounds, etc. (The movie “A Beautiful Mind” gives us a rough idea of how a rational being might handle this situation: he could examine each sensory stream for internal consistency. In the movie the victim of this situation had an additional advantage: he hadn’t always been like this, so he had a large “database” of information that was not “at issue” which he could use to help judge the “veracity” of any given input by judging how well it correlated to what he “already knew”.) But if, after analyzing the inputs as carefully as possible in this way one could find no way to distinguish between “real” and “false” inputs, one would be up a creek, at least so far as forming RJ beliefs is concerned. And one would be up that creek regardless of whether one realized it.

You argue against this resolution of the problem as follows:

But there is a perfectly plausible reason: unlike Jim’s, our memories are pretty consistent by and large. So we do not have a “defeater” for the belief that they are reasonably reliable.

You say:

On the contrary, it is different. Our memories are largely consistent. When we “correct” some of them, we’re creating an ontology that can account (consistently) for a very large percentage (very nearly 100% in fact) of our “apparent” memories and which is consistent with the basic presupposition that our memories are generally reliable. Your beings can’t do this. Any consistent ontology that they create will assign a very large percentage of their memories to the “false” category, contradicting the basic presupposition that their memories are generally reliable. This is much like the situation that Plantinga argues metaphysical naturalists are in: their basic presuppositions lead to conclusions that contradict (or at least render highly doubtful) those very presuppositions.

Anyway, regardless of whether you agree with this analysis, if Jim and Bob were indeed in exactly the same “epistemic situation” (which seems to be the point of your example) and both arbitrarily chose to believe the same set of memories over the other, the only difference being that in Jim’s case they were reliable and in Bob’s case they weren’t (but neither had any way of knowing this), either both sets of beliefs would be rationally justified or neither of them would. Neither this case nor any other can possibly “violate” the FPRJ (at least as far as I’m concerned) because the FPRJ is part of (or is immediately entailed by) what “rational justification” means.

3. On Matt and Jack

I think you have a point here: it might be said that Jack’s cognitive faculties are not designed to produce true beliefs, but rather that his environment is designed to make any belief that he forms (about the future) true. So let me modify the example to meet this objection.

Matt (the designer) doesn’t design the environment to “sense” Jack’s beliefs and “respond” by making them true; rather, it is designed to be “congruent” with Jack’s weird mode of reasoning. Jack still thinks that Susie will be the 19th to appear on stage because “S” is the 19th letter of the alphabet, but the environment has been designed so that Susie will be the 19th to appear on stage.

We can even take this a step further and imagine that Matt has invented a “random cognitive process generator” which will, on demand, create a new “pseudo-random” cognitive process for Jack that produces weird beliefs (within a suitably limited range) and an environment which is congruent with it.

To us (or as I would say, to any rational being) it’s obvious that Jack’s cognitive processes are completely nonrational. (In fact, to those who are not intimately familiar with the algorithm that Jack uses for his cognitive processing, his beliefs seem to be quite random, though in an amusing way.) So we would naturally say that his beliefs are not rationally justified. But on your account they’re perfectly justified, as we can see by simply running through your criteria for RJ. Thus are Jack’s beliefs “formed through the proper functioning of his cognitive faculties which are part of a well designed plan aimed at the production of true beliefs in the type of environment in which those cognitive faculties were designed to function in the absence of sufficient defeaters”? Certainly. His cognitive faculties are functioning perfectly. Matt’s plan is very well designed indeed. It is aimed at the production of true beliefs in the type of environment in which those faculties were designed to function. And, of course, there are no defeaters.

As a further check, let’s consider your test:

The answer to this question is clearly “no”. Matt would not have given Jack those particular cognitive faculties if the resulting beliefs would have turned out to be false. That would have spoiled the effect and ruined the show. So again we see that (on your account) Jack’s beliefs are warranted and hence rationally justified.

If you take Plantinga’s criterion of warrant seriously and maintain that warrant entails rational justification, you are simply going to have to accept some such example as exemplifying this conception of rationally justified beliefs. At some point you’re just going have to accept the implications of the fact that your conception of rational justification is inherently relativistic.

4. On Fred

Not so. Earlier you said:

This entails, of course, that given the same evidence, one being may derive one set of rationally justified beliefs from it and another may derive a different set. So if the evidence in question is “sufficient knowledge of Fred’s total circumstances”, one being might rationally conclude that Fred’s beliefs are rationally justified and another might equally rationally conclude that they aren’t. That’s what I mean by saying that your definition of rational justification is not objective: on your account, rational justification is “in the eye of the beholder”.

Exactly. On your account, if Jack’s design plan tends to reliably produce true beliefs in the type of cognitive environment in which it was designed to function, his beliefs are rationally justified. If by some wild chance Fred’s design plan tends to reliably produce true beliefs in the type of cognitive environment in which it was designed to function, his beliefs are rationally justified. This is the point which is at the heart of our disagreement. I think that “rational justification” can be defined objectively; you don’t. You think that the only way to determine whether a belief is rationally justified is by looking at the overall results of the cognitive process that produced them; I think that it can be determined by analyzing the process itself.

First off, you must have dramatically different notions of what’s plausible than I do, or it wouldn’t even occur to you to say anything like this. Second, if it were true (that is, if there were a large number of at least equally plausible, but substantially different, hypotheses that could account for the experiences you have in mind just as well as the beliefs in question, these beliefs would indeed be irrational.

If you seriously believe that there’s a more parsimonious hypothesis than the “standard model” built up meticulously by millions of competent scientists (many of them incredibly brilliant) over many centuries and based on billions of observations and experiments, go ahead: present it. If you’re right you’ll be world famous. As for me, I find the idea, shall we say, implausible.

bd-from-kg

Kenny:

Now I want to zero in on what I think is at the heart of this dispute, and why I think you’re clearly on the wrong side of it.

1. On “internal justification”

Although internalism (in the strict sense) with respect to knowledge has indeed been pretty much discredited post-Gettier, internalism with respect to justification is very much alive and kicking. In fact, many “J-internalists” argue, as I do, that J-externalism is not so much a position as a changing of the subject. The most worthwhile thing that I can do here is to explain what we mean by this. If nothing else, this will shed some important light on the difference between what I mean by saying that something is rationally justified and what you mean.

The basic idea is that “justification” is essentially a deontological concept: to say that someone is “justified” in doing something is to say that he is “within his rights” or that he is not “violating his duty” in doing it.

Let’s look at what this implies when the “rights” or the “duty” involved are moral in the usual sense.

Say that you’re on a jury charged with deciding whether Smith murdered Jones. Suppose that given the evidence presented to you Smith appears to be guilty beyond a reasonable doubt, but that in fact he’s innocent. Are you justified in voting to convict him? Of course. In fact, it’s your duty to vote to convict under these conditions. It simply doesn’t make sense to say that you are not justified in convicting him because unknown to you he’s innocent. There might even be facts known to the D.A. such that, if you knew them, you would acquit him, but that doesn’t matter, because you don’t know them. Whether you’re justified in convicting Smith is entirely a matter of how you “processed” the evidence presented to you; whether this evidence has the “right relationship” to the state of affairs that makes him guilty or innocent, or even whether there is any such relationship, is irrelevant to the question of justification. In other words, moral justification is an intrinsically internal concept.

In the same way, “rational justification” is also a deontological concept: to say that one is rationally justified in believing something is to say that one is “within his epistemic rights” or “not violating his epistemic duty” in believing it. This is an intrinsically internal concept for the same reason that moral justification is intrinsically internal. Whether you’re rationally justified in believing something is entirely a matter of how you “processed” the evidence available to you; whether this evidence has the “right relationship” to the state of affairs that makes it true or false, or even whether there is any such relationship, is irrelevant to the question of rational justification.

2. On Plantinga and moral justification

In fact, the relationship between rational (i.e., epistemic) justification and moral justification is not merely that they are analogous; they are closely connected. To return to our jury example, you’re morally justified in voting to send Smith to prison because you’re rationally justified in believing that he murdered Jones. If you weren’t rationally justified, you wouldn’t be morally justified. By neglecting your epistemic duty you would be violating your moral duty. This is the kind of thing that Clifford had in mind when he wrote in The Ethics of Belief: “So closely are our duties knit together, that whoso shall keep the whole law, and yet offend in one point, he is guilty of all.”

And this is why the question of rational justification is important. We want to be able to say that people who fly planes into tall buildings for the purpose of killing thousands of innocent people are morally culpable. The fact that they believe that they’re morally justified because God wants them to it doesn’t get them off the hook, morally speaking, because this belief is not rationally justified; in forming it they violated their epistemic duty not to believe anything on insufficient evidence.

But on Plantinga’s account they are off the hook, because it’s possible that their belief in Allah and His intentions and purposes was formed through the proper functioning of their cognitive faculties which are part of a well designed plan aimed at the production of true beliefs in the type of environment in which those cognitive faculties were designed to function in the absence of sufficient defeaters for said belief. And in case this happens to be true, it is highly probable that an Islamic version of the sensus divinitatis exists and has been designed by Allah to produce accurate beliefs about Allah and His intentions and purposes through interaction with the world around us. Therefore their beliefs are “properly basic defensible”, which means that they did not violate their epistemic duty by forming them, which means that they are not morally culpable when they kill thousands (or for that matter, hundreds of thousands, or millions, or billions) of innocent people in accordance with their religious beliefs.

This is the true nature of the ideas that Plantinga is peddling. I’m sure that he means well, but the reality is that his doctrine is radically subversive of moral order. It can be used to justify any evil, any atrocity, so long as it is committed in the name of an appropriate metaphysical doctrine. And this is why there is no way on earth that I’m ever going to buy it.