bd-from-kg

Clutch:

Hate to be a pest, but this subject is pretty close to the heart of the point of this thread, so it seems to me to be worth pursuing. And no one else seems to be doing it.

[By the way, to avoid any misunderstanding on the point, I am not supporting theism. The fact that principles for evaluating empirical claims are not themselves empirical claims doesn’t imply anything at all about the existence of God, and it certainly doesn’t support presuppositionalism, which I consider to be out there in la-la land.]

Part 1:

I might as well go ahead and make what few comments seem appropriate on the last part of your June 24 post.

No, I didn’t have that in mind. Whatever gave you the idea that I might?

Do you really not understand what’s wrong with this kind of circularity?

As if the circularity weren’t bad enough, the “principle” in this case is a value judgment, so it doesn’t make sense to try to “justify” it anyway.

Here’s where you lose me completely. I don’t see the point of this weird nomenclature. I think the notion of a “self-testing” principle is logically incoherent. And it seems obvious to me that (A') is not answerable empirically. (But it’s equally obvious that the answer is “yes”.)

Part 2:

Now on to your June 25 post.

You’re taking this business of analyzing the logical structure of Eval way too seriously. The idea was simply to give a collective name to the principles for evaluating empirical claims, to make it easier to talk about them. In other words, the main purpose was conciseness. I could have said what I wanted to say without ever introducing Eval. It never occurred to me that someone would seize on this simple word-saving device and beat it to death as a way to counter my argument. But since you have, I’m going to stop talking about Eval; given your tactics it’s just getting in the way of making the point.

How can this be unclear? It’s about as plain as anything can be. In order to evaluate any principle for evaluating empirical claims, you have to have facts. And to get them you have to apply principles for evaluating empirical claims, in order to determine whether each supposed “fact” qualifies as a fact. Sooner or later (if we try to test all principles for evaluating empirical claims in this way) this must lead to a circularity.

I gather that your point is that if this is logically possible, this would show that the principles for evaluating empirical claims are falsifiable in principle, and thus are in fact making empirical claims. This is an important point, so I’ll give it a few paragraphs.

But I don’t see the relevance of whether some such principles fail strictly “by the lights of other such principles”. If the principles for evaluating empirical claims (taken as a whole) should fail such an empirical test, they fail; the details of how they failed wouldn’t matter. What matters is: what should we conclude from such a failure?

Let’s take the Principle of Induction as an example (since it’s simple and basic, and has been analyzed to death). Suppose that it’s possible to show (on some possible world) that the Principle of Induction fails. What follows?

I’d say that what would follow is that rational evaluation of empirical claims is impossible in that world. If a world is so chaotic that induction consistently gives results no better than chance, rational creature could not exist there, and if they somehow did, they would be completely helpless. There would be no way for them to learn anything about their world (beyond their current, immediate perceptions) or to function effectively (i.e., to pursue goals, to influence events in ways they consider desirable).

The same thing would follow from the failure of any other principle for evaluating empirical claims that might conceivably fail in this way. (Many, like Ockham’s Razor and Sagan’s dictum, are such that it’s logically impossible for them to “fail”.)

Thus in saying that these principles are valid, we are not saying that they will “work” in all possible worlds, but that they will work in any possible world which is capable of being understood rationally. And moreover, that they represent the “best strategy” in the absence of any information about the world (which is necessarily the situation we all have to start from). Thus following these principles is the only viable option for a rational being; if they don’t work, no rational strategy will work.

But none of this changes the fact that the principles for evaluating empirical claims cannot be empirically validated. They represent (taken as a whole) the rational strategy for understanding the world, and can be justified as logically necessary components of that strategy. That’s not an empirical justification; it’s an a priori justification.

Sorry, you’ve lost me here. I don’t know what you mean by “this positivistic sort of empiricism”. I have no idea whether, or why, “virtually everyone since 1960” rejects it. And I have no idea how this relates to my argument.

Please don’t assume that everyone here is as much of a scholar as you. This isn’t a symposium of professional philosophers. The general rule around here is that if you’re going to appeal to authority (which is what you’re doing here) you need to give at least a general outline of the main arguments that the authority used to support the conclusion in question. Otherwise the “Evolution” forum would shut down very quickly after it was pointed out that “virtually everyone” (i.e., virtually all competent biologists) believes in it.

Um, if your knowledge of the real world isn’t based ultimately on your perceptions, what is it based on? Intuition? Revelation? Other people’s perceptions? Or is it the term “perception” that you object to?

It’s true that in practice we demand more evidence for claims that really matter than for those that don’t – at least if we have time to decide, if getting more evidence is reasonably feasible, etc. But I don’t see that these kinds of things are relevant to the question of when belief is justified. It’s more a matter of when it’s justified to act on a hypothesis. To put it another way, belief isn’t an on/off switch; you don’t simply believe something or not believe it. There are degrees of belief. You can think that something is more or less probable.

In other words, we can separate the questions:

(1) How probable is it that H is true?

(2) Shall I act as though H is true?

Sometimes you can hedge your bets; you don’t have to act either in a way that will only further your ends if H is true or in another way that will only do so if H is false. But sometimes you are faced with this stark choice, although you may have some room to maneuver: you can postpone the decision for a while until more information comes in. All of these things can be very important to get right. But they all relate to (2), not to (1).

OK, let’s try “By appealing to the principles for evaluating empirical claims” The point is that ultimately if you’re going to empirically evaluate a principle for evaluating empirical claims as if it were itself an empirical claim, you can only do so by applying the principles for evaluating empirical claims. That’s circular (at least if we try to do this for all such principles); hence it’s illegitimate; hence there’s no legitimate way to empirically validate the principles for evaluating empirical claims. You might be able to do it for some of them, but not for all.

Look. My point is that Sagan’s dictum is not an empirical claim. It says nothing about the “real world”. Since it’s clearly meant to apply only to empirical claims, the question of whether it “meets its own standards” is irrelevant. To support my position that it’s not an empirical claim I’ve shown that (1) trying to test it empirically just leads one in a circle; it’s not possible in principle to test it empirically because nothing would count as evidence against it, and (2) it’s impossible in general to empirically evaluate principles for evaluating empirical claims without getting into a circle.

Now you say that we can at least test the dictum for “consistency”. But if it’s not an empirical claim this doesn’t make sense; it simply doesn’t matter what the result is.

By way of analogy, suppose that the Supreme Court says, “No one is above the law; we all have to abide by the law”. Someone objects, “But in the final analysis the law is what the Supreme Court has said it is. And the Supreme Court itself doesn’t always abide by what the Supreme Court has said; it sometimes reverses its earlier decisions. So the Supreme Court is violating its own dictum.” The answer to this, of course, is that the statement was not intended to refer to the decisions of the Supreme Court itself. The fact that the Supreme Court doesn’t abide by its own previous decisions may look like an inconsistency at first sight, but it’s not.

In fact, if Sagan’s dictum were an empirical claim, theophilus’s objection could be disposed of very quickly: we would simply test it empirically. The only reason the objection is even superficially plausible is that this hasn’t been done. And it hasn’t been done because it can’t be done, because it’s not an empirical claim.

No. I just think that it’s what we were talking about. What do transparency and honesty have to do with this discussion? They’re highly desirable (in fact indispensable) to scientific progress in practice, but they have nothing to do with whether principles for evaluating empirical claims are themselves empirical claims. On the other hand, reproducibility is an integral part of testability (at least when we’re testing proposed scientific laws) and we were talking (at some length, in fact) about testability.

I’m quite familiar with the distinction between what you call "categorical" and "hypothetical" value judgments. If it matters, I think that a commitment to the relevant value judgments (“It’s desirable that our theories work”, etc.) is part of what it means to be rational. But we seem to be getting pretty far off-subject here. I am not making the trivial point that value judgments cannot be tested empirically.

Clutch

I never believed otherwise.

I said, the self-testing version of a principle; that is, a principle tested by its own standard. You may think this incoherent all day long -- but without a cogent argument to that effect, I decline to think it along with you.
The "tactics" of looking at the truth-conditions of your argument?

"Does the prosecution claim that Smith was at the scene and committed the murder, or that Smith was at the scene or committed the murder?"

"Aww, Judge, you're taking this logical structure business way too seriously!"

For pete's sake, bd. You introduced a big, complex statement, to use it in a refutation. Whether it's a big, complex "and" statement, or a big, complex "or" statement, is sort of important.
And what we could conclude is entirely determined by what you mean when you say "as a whole". All of them conjoined, or all of them disjoined?
It's the description of perceptions as a "pitifully tiny" collection of evidence that is related to theory-level data as premises to conclusions. That's not how perception works, and, in reverse, it's not how justification works.
Yes, that's what you've been saying. But I haven't seen a cogent argument yet. By contrast, my argument is pretty straightforward: Sagan's dictum regards the pursuit of science; science is part of the real world; Sagan's claim is about the real world.So far as I can tell, you have not done this. Neither argument appears to work, for reasons I have explained at length.

bd-from-kg

Clutch:

This is a reply to your June 26 post.

As for truly untestable claims that actually affect the predictions, there's nothing to talk about, because there's no such thing; claims that affect predictions are by definition testable. But we can talk about untestable hypotheses that are embedded in testable theories, as I do below.

Correct, except that I don’t know what you mean by “higher-order”. (What would a “lower-order” test look like?) But also crucial to the argument is the point that it’s logically impossible that all research programs of the sort you describe could “work”.

Well, that would be odd, because I don’t agree. This hypothesis seems eminently testable to me. It yields testably different predictions than the “null hypothesis”. What’s the problem?

Well, yes. In the same way, measurements and observations can show only that massive bodies tend to accelerate toward one another at a rate given by the inverse-square law. They do not show that this acceleration is produced by an unimaginable, invisible, and undetectable “force”. But that doesn’t make the hypothesis that such a force produces such acceleration untestable.

If “testing” a hypothesis meant showing directly that any postulated entities really exist in addition to showing that the predictions it generates are correct, no scientific hypothesis could be tested, because this type of test is impossible in principle. All we can ever do is to compare predictions to actual outcomes.

That’s the nature of scientific theories. They typically postulate various entities (the electroweak force, quarks) whose supposed existence is manifested only in various observable phenomena; they are not themselves observable. Such theories (like all scientific theories) are considered to be confirmed when their predictions match observations and disconfirmed when they don’t.

Yes, just as the Newtonian theory of gravity could have dispensed with the “superfluous” notion that the observed accelerations are caused by a mysterious “force” acting at a distance, and contented itself with saying that bodies do accelerate toward one another in accordance with the inverse-square law. (In fact, it’s worth noting that according to GR there is no such force; there’s just a warping of the space-time-continuum which manifests itself in ways that make it look to our human intuitions as though a force is acting.) But that’s just not the way science is done. It’s perfectly normal for physical theories to try to provide a “picture” or “model” of what’s really going on, even though this model, strictly speaking, is superfluous. The important thing is that we recognize that the “picture” or “model” is just a way of helping us to grasp the theory intuitively, to figure out more easily what it predicts. The human mind really seems to need images or models before it can get comfortable with a theory. (That’s why QM is so unsettling. No one has been able to come up with such a “picture” or “model” for it. All we really have is the equations and rules for how to use them to get predictions.)

But as I hope I’ve convinced you by now, a great many of our successful theories are based on (or at any rate incorporate) just such untestable claims. And such things are useful in helping scientists to come up with fruitful new hypotheses.

Yes. And we do have such evidence.

Not really. It’s important to keep in mind that we have no way of knowing whether the physical world is “really”, in some sense, anything like the pictures or models utilized in our scientific theories. More importantly, the “pictures” or "models" of what's going on embedded in most of our physical theoried don’t affect the predictions that issue from these theories in the least, even though they may make it easier (for us humans) to see what the theories predict. In both of these senses they are "veridically worthless".

Of course, I suspect that Sagan was really referring to “dragon-in-the-garage”-type untestable claims, which are not associated with any physical theory and are worthless in all respects. The only reason I got into the other kind is to show that your “u,i,u One” example doesn’t refute my argument. It’s a perfectly good, testable hypothesis by any criterion that I know of, and the success of predictions based on it supports it, just as it would any other testable hypothesis. Any hypothesis that helps to produce testable predictions is, by definition, part of (or associated with) a physical theory, and is testable simply by checking the predictions.

In short, your argument fails. Your supposed counterexample isn’t a counterexample, and once one understands why it’s not a counterexample it should be clear why there cannot be any counterexamples.

bd-from-kg

Clutch:

Rather than replying to your last post point by point, I think it would be best to take your last statement as a springboard for recapping my argument.

So far as (1) is concerned, that’s what my last post was about. To summarize: you can’t empirically test the dictum that “it’s irrational to believe something about the real world without evidence” because nothing can count as evidence against it, so nothing counts as evidence for it either. The latter statement is logically entailed by the former, so it’s sufficient to show that nothing can count as evidence against it. The demonstration consists of observing that as soon as we know that something is evidence against it - i.e., that it’s an untested hypothesis that turns out to yield true predictions, or at least plays a crucial role in yielding such predictions - we have evidence for it, which means that it’s not untested. And it’s logically impossible that all untested hypotheses could yield true predictions. (Actually it’s quite obvious that the vast majority of them must yield false predictions, but I nevertheless provided an argument showing that for every such hypothesis that yields true predictions there’s at least one that yields false ones.)

As for (2), the principle involved is extremely simple: it’s the principle of “You can’t find out whether a person is truthful by asking him.” Now of course, in a sense you can sometimes tell whether a person is truthful by asking him – namely if he replies, “No, I’m not.” But if he replies, “Yes, I am” you learn nothing beyond the fact that he doesn’t contradict himself.

Now let’s ignore for the moment the possibility that a principle for evaluating empirical claims is self-contradictory, or that given the evidence it “declares itself” to be false, and consider what it would mean to say that it’s confirmed by the evidence. Surely this can only mean that the rules for evaluating empirical evidence, taken as a whole, instruct us to follow such-and-such a procedure, and that the principle is confirmed if the results satisfy a certain description and disconfirmed if they don’t.

If we follow this procedure for every principle for evaluating empirical claims, the best possible result is that all of them are strongly confirmed. But what does this mean really? It means that the principles for evaluating empirical claims (taken as a whole) tell us that they’re true. And this “evidence” is just as good as the evidence that a man provides of his truthfulness by assuring us that he’s truthful. Which to say that it’s no good at all. But this means that it’s impossible in principle to verify such principles empirically.

Note that we’re not talking here about proving empirically that these principles are valid; nothing can be proved empirically. We’re asking only for evidence sufficient to justify rational belief. The fact that they "declare themselves" to be valid isn’t even close to being sufficient to justify rational belief, any more than that fact that someone assures you that he’s truthful is sufficient in itself to justify rational belief that he is.

The only possible conclusion from all of this, it seems to me, is that our confidence in the validity of these principles is not based on empirical evidence. And that means that they are not empirical claims.

Now for a couple of side points:

(1) It should be clear from the above why the “logical structure of Eval” is irrelevant. What matters is that by definition the principles for evaluating empirical claims yield a procedure for evaluating any given empirical claim. In fact, without a procedure for testing a claim, it cannot be called an empirical claim at all. So if these principles do not yield a procedure for evaluating a claim the only possible conclusions are that the claim is too vague to constitute an empirical claim, or that it simply isn’t, by it’s very nature, an empirical claim.

(2) A couple of posts back I presented this argument:

So far you’ve answered this by saying that this argument represents a “positivistic empiricism” that practically no one accepts any more. This is nonsense. The idea that one’s picture of reality has to derive ultimately from one’s own perceptions is very widely accepted; it’s hardly limited to positivists. The only other logically possible sources of information about the external world (aside from quibbles about whether “perceptions” should be replaced by “sense data” or “sensory inputs” or whatever) are innate knowledge and revelation, and it’s safe to say that a great many professional philosophers reject both of these as sources of information about the external world. And obviously the only way to get from perceptions to propositions about the external world is by applying principles for evaluating empirical claims. That’s the function of principles of this kind; it’s what they’re for.

Your latest post explains what you think is wrong with this argument as follows:

I can’t believe that you really object to the characterization of your own perceptions as a “pitifully tiny” set of facts relative to the vast pool of facts that you have at your disposal. As for the relationship of one’s own perceptions to all of those other facts, I don’t know what exactly you mean by saying that I describe it as a relationship between “premises and conclusions”. I certainly didn’t use those words. All that I said is that you have to get somehow from the former to the latter, and that your beliefs in all those “non-perceptual facts” are only rationally justified if you can do so by means of valid principles for evaluating empirical claims. What part of this do you not agree with?

Clutch

bd,

You are right to say that you have simply repeated your arguments. For reasons I've already given, they are not compelling.

I've tried several times now to see where they rise above question-begging or simple non-sequitur, but I can't.

So I'm afraid I have it to leave it at the explanations I've already given, and allow that the fault may be mine for not finding the cogency you evidently find in your reasoning.