So I'm not going to put forth counter-arguments to show that the things you're putting forth are wrong, because they're mostly not. — Pfhorrest

OK, so you agree that confirmationism and falsificationism are, as I have been arguing, two sides of the one coin with both being in play in science and everyday empirical matters? I have to say it hasn't seemed like you have agreed to that. Also I never accepted that the invalid syllogism "If P, then Q, Q therefore P" represents verificationist logic. I doubt the members of the Vienna School would have, either. And yet despite my objection to that you have kept trotting it out to use it as a strawman to support your contention that it's all about falsification. And you accuse me of poor reading simply because I don't agree with you.

Then when I told you the correct confirmationist formulation is "If Q, then P", you tried to claim that this is essentially falsificationist which it isn't at all. Its equivalence with with "If P then Q, not-Q therefor not-P" show that both confirmation and disconfirmation are in play in empirical investigations, and yet for some reason you don't want to admit that. You seem like a dog with a bone that it won't let go of. If you don't agree then tell me what exactly it is that you've been claiming that you think I have been disagreeing with all along.

Yeah I already addressed all of that in my last post, so if you're not going to read the whole thing, maybe just don't bother responding to it.

I'll just repost the most important bit that I think cleared everything up.

--------

the correct formulation for confirmationism is "if Q then P", — Janus

I just went to look up a quote about confirmationism in a reliable source, the Stanford Encyclopedia of Philosophy, to settle this merely nominal dispute once and for all, and I found something interesting: there are two mutually contradictory things both called "confirmationism".

I learned confirmationism in school as synonymous with the hypothetico-deductive method, which to quote SEP means:

e HD-confirms h relative to k if and only if h ∧ k ⊨ e and k ⊭ e; — SEP

Where e is some evidence, h is some hypothesis, k is some set of background assumptions, and ⊨ is a symbol for entailment, i.e. necessary implication.

So on that hypothetico-deductivist confirmationist account, if the hypothesis (and background assumptions) entail the evidence (but the background assumptions alone don't), then seeing that evidence confirms the hypothesis. That is exactly what I have been saying confirmation claims, except using "P" for the hypothesis+assumptions together, and "Q" for the evidence: that if P implies Q, and Q is the case, then P is the case.

But, it seems, Hempel's model, which I learned in school as something against confirmationism and more a step toward falsificationism -- because it says exactly the opposite of that hypothetico-deductivism above, just like falsification does -- is apparently also called "confirmationism", and I'm guessing that that's where you're coming from, Janus. Hempel's account says:

e confirms h relative to k if and only if e disconfirms ¬h relative to k. — SEP

Which is pretty much the same thing falsification says, except it doesn't call that "confirming" h, because falsificationism uses the term "confirmation" to refer to hypothetico-deductive confirmation. It just says that that's falsifying ¬h, which of course entails that h as well.

Yeah, I read all that before, but you stopped short of admitting that your identification of confirmationist thought with the invalid syllogism was incorrect. And you also haven't admitted that falsification/verification are inseparable and that both are at work in empirical investigations. It appears this is all we have been arguing about.

but you stopped short of admitting that your identification of confirmationist thought with the invalid syllogism was incorrect — Janus

Because it's not incorrect, it's just a different sense of the word "confirmationism" than you're using. Hypothetico-deductivist confirmationism is just like I've been saying confirmationism is, because that's the thing I've been arguing against since the very beginning.

Which is as I've said many times before, I'm not strawmanning you're position, you're identifying yourself with the position I'm arguing against and then saying that that position isn't actually like that but is instead like something I never disagreed with.

Do you really believe there have been any philosophically significant confirmationists stupid enough to base their whole system of thought on an invalid syllogism, though?

Do you really believe there have been any philosophically significant confirmationists stupid enough to base their whole system of thought on an invalid syllogism, though? — Janus

Apparently, since there's a whole name for that methodology and it was taught as one of the several views discussed in my university philosophy of science class.

OK I'll take your word for it; but it does seem incredible! Can you cite any texts by, for example, Ayer or the Logical Positivists that explicitly identify their thinking with the syllogism in question? Perhaps the verificationists have simply been misunderstood, or even tendentiously misrepresented, by their critics.

Can you cite any texts by, for example, Ayer or the Logical Positivists that explicitly identify their thinking with the syllogism in question? — Janus

I already clarified that verificationism and confirmationism don’t mean the same thing here.

OK, well then just who were the confirmationists you have in mind?

I don’t recall any supporters’ names being taught, and my search on SEP suggests that it’s merely a traditional view:

Originally, Glymour presented his sophisticated neo-Hempelian approach in stark contrast with the competing traditional view of so-called hypothetico-deductivism (HD).

...

For one thing, the very idea of hypothetico-deductivism has often been said to stem from the origins of Western science.

Etc.

:ok:

This from the Wiki page on verificationism:

Although Karl Popper's falsificationism has been widely criticized by philosophers,[19] Popper has been the only philosopher of science often praised by many scientists.[12] Verificationists, in contrast, have been likened to economists of the 19th century who took circuitous, protracted measures to refuse refutation of their preconceived principles.[20] Still, logical positivists practiced Popper's principles—conjecturing and refuting—until they ran their course, catapulting Popper, initially a contentious misfit, to carry the richest philosophy out of interwar Vienna.[11] And his falsificationism, as did verificationism, poses a criterion, falsifiability, to ensure that empiricism anchors scientific theory.[2]

In a 1979 TV interview, A. J. Ayer, who had introduced logical positivism to the English-speaking world in the 1930s, was asked what he saw as its main defects, and answered that "nearly all of it was false".[18] However, he soon admitted to still holding "the same general approach".[18] The "general approach" of empiricism and reductionism—whereby mental phenomena resolve to the material or physical, and philosophical questions largely resolve to ones of language and meaning—has run through Western philosophy since the 17th century and lived beyond logical positivism's fall.[18]

In 1977, Ayer had noted, "The verification principle is seldom mentioned and when it is mentioned it is usually scorned; it continues, however, to be put to work. The attitude of many philosophers reminds me of the relationship between Pip and Magwitch in Dickens's Great Expectations. They have lived on the money, but are ashamed to acknowledge its source".[2] In the late 20th and early 21st centuries, the general concept of verification criteria—in forms that differed from those of the logical positivists—was defended by Bas van Fraassen, Michael Dummett, Crispin Wright, Christopher Peacocke, David Wiggins, Richard Rorty, and others.[2]

It seems that the nub of the problem is that universal statements cannot be definitively verified for obvious reasons; we can never observe every case, or even if we have observed every case, know that we have. So universal statements can only be falsified. Apparently some of the logical positivists were not happy with this because according to their own criterion universal statements would have to be thought to be meaningless. It's surprising that something so obviously wrong would be clung to by highly intelligent thinkers.

Interestingly, the situation is reversed when it comes to statements that are not in the universal form; they cannot be falsified but only verified. So, to return to a previous example "All swans are white" is a universal statement that can never be definitively verified, but can be definitively falsified. Conversely "Some swans are purple" can never be definitively falsified, but could be definitively verified.

Mostly just wanted to give you a :up: in return, but one thing:

Conversely "Some swans are purple" can never be definitively falsified, but could be definitively verified. — Janus

Only given certain background assumptions which theory-laden that observation, which assumptions may themselves be false (and so that conclusion as well). One can be certain of having had some experience that seems to them to have been of a purple swan, but one can never be definitively certain that “there exists at least one purple swan” is in fact the correct interpretation of that experience, e.g. maybe you have been somehow deceived and in fact there are no purple swans despite this convincing appearance of one.

Of course that also applies to falsifying particular universal hypotheses — maybe your falsifying observation isn’t genuine somehow — but in that case you’ve still falsified the conjunction of that particular hypothesis and the rest of the background theory that ladens your observations such that the hypothesis seems falsified.

Your caveat here is noted, and it applies to all sensory experience, hence no absolute certainty to be found anywhere. If I were to find a purple swan, and bring it to a lab to be tested to see if it really is purple, and it was found to be such, then the proposition would be verified to the same degree as anything observable can be.

:up:

universal statements cannot be definitively verified for obvious reasons; we can never observe every case, or even if we have observed every case, know that we have. So universal statements can only be falsified. Apparently some of the logical positivists were not happy with this because according to their own criterion universal statements would have to be thought to be meaningless. It's surprising that something so obviously wrong would be clung to by highly intelligent thinkers. — Janus

The problem is only with the definition of the set. If you say all Xs just happen to have property Y, you're expressing a probability function between two independent variables, this can be dealt with using Bayes. If, on the other hand, you're proposing some relation between X and Y, then the strength of your claim depends on the mechanism describing the function of Y given X.

In neither case do we need to observe all Xs to verify the universal statement. The only situation in which we would might be a frequentist probability function of independent variables, but there's no reason why we'd need such a thing.

So "all swans are white" does not need observation of all swans. If 'swaness' and 'whiteness' are independent, then verification is just the approaching to 1 of P(white|swan), but if we know 'swaness' and 'whiteness' are independent we must already have prior beliefs about both. In the more normal case 'swaness' and 'whiteness' are not independent, which means that "all swans are white" can be verified by the definition of 'swaness' - say, for example, some gene which codes both for 'whiteness' and some other defining characteristics such that nothing not white could possibly be a swan by definition.

Also, if we have reason to believe that P(A|G)=1 where G is some set of variables (x,y,z...) then in the case that P(y),P(z)...etc are non zero (ie, not yet falsified), an increase in P(x) does indeed lead to an increase in P(A). So where z implies A, z does increase the likelihood of A given a set of prior beliefs about the set of variables conditional for A, of which z is part.

In short, your instinct was right, they weren't that stupid. The confusion comes from a naïve treatment of our beliefs as if they were a half-dozen independent logical correlations rather than a complex network of hundreds of thousands of interconnected implications.

In short, your instinct was right, they weren't that stupid. The confusion comes from a naïve treatment of our beliefs as if they were a half-dozen independent logical correlations rather than a complex network of hundreds of thousands of interconnected implications. — Isaac

I'm not familiar, other than by hearsay, with Bayesian thought, but what you are saying sounds like what I have been groping towards. My main point has been that in order for one thing to be falsified, a whole interconnected range of others things must be counted as being confirmed. The example I gave of the proposition "some swans are purple" would not be taken seriously unless there were some good reasons within the coherent network of our beliefs, to think that a swan could be naturally purple. So, I don't see how inductive and confirmationist thought can be dispensed with, or that falsification, to quote Pfhorrest, "does all the heavy lifting".

I never said inductive thought generally was to be dispensed with entirely. I've repeatedly said induction is a fine way to come up with beliefs in the first place.

Where falsification does all the heavy lifting is in deciding between competing beliefs that both fit some pattern that induces us to believe them. In that case, observing something that is predicted by one of them doesn't help at all (contra what hypothetico-deductive confirmationism would have us think), unless it's also against the predictions of the other (i.e. falsifying it).

The point about the interconnected beliefs was discussed at length earlier with regard to confirmation holism and theory-laden observation etc. If you observe something that agrees with all of your beliefs, you've learned nothing, as described above. If you observe something that's contrary to any of your beliefs, then you've learned that that combination of beliefs is not possible. It may not be that you have to reject the one belief you thought you were testing -- you could reject some other beliefs instead -- but still you've learned that you have to change something about your beliefs.

This was what I was bringing up with the purple swan earlier. You can't be sure that you've observed a purple swan, because the observation of what seems like a purple swan could always be interpreted as not really a purple swan if you change some of your background beliefs instead. So in the strictest sense, you haven't confirmed that there exists at least that one purple swan, even though in a colloquial sense we can often be sure enough for practical purposes.

That does means likewise that you can't be sure that you've falsified that all swans are white. But you can be sure that you've falsified something, because you can't both keep all of your background beliefs that lead you to interpret that observation as a purple swan, and also keep your belief that all swans are white. If you see something that seems to be a purple swan according to your background beliefs, you've got to reject either some of those background beliefs, or the belief that all swans are white. In either case, the full set of beliefs you had before are now known to be for sure false, even if you don't know exactly what change to your beliefs is the best one to make yet.

This still accords with Bayesian reasoning, because you could reason along the same lines, but probabilistically instead of in those absolute statements. If the thing you're observing is very likely to be a real purple swan given your background beliefs, and yet it's very likely that all swans are white given what you believe about swans, then what you're observing must be very improbable. Contrapositively, some or other of the beliefs that lead you to believe you're observing that must be very improbable: either your background beliefs, or your beliefs about swans. But it's very unlikely that you're both probably right about all swans being white, and probably really seeing a purple swan, so you're probably wrong about at least one of those things.

I never said inductive thought generally was to be dispensed with entirely. I've repeatedly said induction is a fine way to come up with beliefs in the first place. — Pfhorrest

I had thought that you said early on in this thread and repeatedly thereafter that it doesn't matter what we believe (that is falsifiable) as long as it hasn't been falsified.

So in the strictest sense, you haven't confirmed that there exists at least that one purple swan, even though in a colloquial sense we can often be sure enough for practical purposes. — Pfhorrest

Yes, but that "strictest sense" as I already said applies to everything; it is merely the fact that there is not any absolute certainty; no deductive strength proof of anything, when it comes to empirical matters. But that is really a useless wasteland of weeds we don't need to get into.

You could say the same about black swans; that it could never be absolutely proven that they are in fact swans.

But it's very unlikely that you're both probably right about all swans being white, and probably really seeing a purple swan, so you're probably wrong about at least one of those things. — Pfhorrest

Yes, inconsistencies do spell trouble for belief systems to be sure.

I had thought that you said early on in this thread and repeatedly thereafter that it doesn't matter what we believe (that is falsifiable) as long as it hasn't been falsified. — Janus

Yes, which means that induction is a perfectly fine way of coming up with beliefs. There's nothing wrong with using induction to get to something or another to believe. It just can't tell you that your beliefs are more right than some other beliefs.

But it's also possible that different processes for coming up with beliefs will be more or less productive in coming up with beliefs that are unlikely to be falsified. Believing that patterns you've observed are likely to continue (i.e. induction) could very well be one of those safer methods (and I'm intuitively inclined to say it probably is, but I don't have any arguments to that effect).

Yes, but that "strictest sense" as I already said applies to everything; it is merely the fact that there is not any absolute certainty; no deductive strength proof of anything, when it comes to empirical matters. But that is really a useless wasteland of weeds we don't need to get into.

You could say the same about black swans; that it could never be absolutely proven that they are in fact swans. — Janus

Yes, that's correct. But it is that "strictest sense" that we're talking about here. It's in that sense that although we can never be sure any particular set of beliefs is the correct, we can be sure that some particular set of beliefs is an incorrect one, if e.g. that particular set of beliefs says both that all swans are white and that the thing we're seeing right now is a purple swan. We can't be sure which of those (or which part of which of them) is incorrect, but we can be sure that the world definitely isn't exactly like that, it's different than we thought in some way or another.

It's in that sense that although we can never be sure any particular set of beliefs is the correct, we can be sure that some particular set of beliefs is an incorrect one, if e.g. that particular set of beliefs says both that all swans are white and that the thing we're seeing right now is a purple swan. We can't be sure which of those (or which part of which of them) is incorrect, but we can be sure that the world definitely isn't exactly like that, it's different than we thought in some way or another. — Pfhorrest

So what belief has been falsified here? Not the belief that there are only white swans. Not the belief that I haven't seen any purple swans (I'm presuming I was deceived). Not the belief that my observations are always accurate and unambiguous (I never believed that). Not the belief that I haven't ever seen anything which even looks like a purple swan (That belief was true at the time - all time-dependent beliefs change as time changes - I've never seen 7:45 on the 20th November 2020...until now).

So exactly what belief has now been falsified which anyone ever actually had prior to this observation?

Not the belief that I haven't seen any purple swans (I'm presuming I was deceived). — Isaac

Other way around: you change the beliefs which initially lead you to construe your experience as genuinely seeing a real purple swan, if you instead conclude that you must have been deceived.

If you already believed that you were being deceived into seeing a purple swan, then none of your beliefs would change, but then your experience wouldn't be contrary to your prior beliefs either (you would be expecting to see what appears to be a purple swan), so there would be no prompt to change any beliefs, and so no falsification.

Other way around: you change the beliefs which initially lead you to construe your experience as genuinely seeing a real purple swan, if you instead conclude that you must have been deceived. — Pfhorrest

Which beliefs? I never believed that all my observations are accurate and unambiguous. What I believed prior to seeing the purple swan was that some observations turn out to be true and others don't.

This still accords with Bayesian reasoning, because you could reason along the same lines, but probabilistically instead of in those absolute statements. If the thing you're observing is very likely to be a real purple swan given your background beliefs, and yet it's very likely that all swans are white given what you believe about swans, then what you're observing must be very improbable — Pfhorrest

...which equates to your error above (not that this hasn't already been pointed out). P(A|B) is not the same as P(A and B). You're misunderstanding Bayesian probability - which is fine if what you're doing is inquiring, but when you're declaring some theory to be consistent with it you really ought to do more than just glance at a Wiki page on the topic.

Which beliefs? I never believed that all my observations are accurate and unambiguous. What I believed prior to seeing the purple swan was that some observations turn out to be true and others don't. — Isaac

So you were not surprised by the apparently purple swan, and it was consistent with your prior beliefs? Then you have no contradictory observations to falsify anything. You just saw something consistent with your expectations.

P(A|B) is not the same as P(A and B). — Isaac

I never said it was. I say that the probabilistic equivalent of a conditional statement is a conditional probability: the probabilistic equivalent of "B if A" is "P(B|A)". (I did misleadingly say that "P(B|A)" was equivalent to "P(B if A)", but that was for a natural-language reading of "B if A", and that equivalency is only problematic when "B if A" is taken as a strict material implication).

So you were not surprised by the apparently purple swan, and it was consistent with your prior beliefs? Then you have no contradictory observations to falsify anything. You just saw something consistent with your expectations. — Pfhorrest

Surprise has nothing to do with it. I might be surprised by a purple swan because I wasn't expecting one. This is the part about time-dependant beliefs I mentioned earlier. I don't believe I can predict the future, yet things still surprise me about it. I have expectations about future events despite not believing that I can predict them accurately. I believe the set of observations which seem to me to be accurate up to this very moment in time, but I cannot believe now those that I will see in future.

This is where probabilistic belief becomes crucially important to our understanding of belief. I believe almost nothing 100%, I believe different things with different strengths. The strength with which I believe future events is inevitably altered as the time becomes the past, again no-one doesn't already believe this.

Neither confirmationists, nor fideists, nor nihilists, nor any of your stated targets believe they can predict the future with 100% percent accuracy. So none of them are disproven by your statement that observing something surprising contradicts a belief (an expectation) that we wouldn't.

I never said it was. I say that the probabilistic equivalent of a conditional statement is a conditional probability: the probabilistic equivalent of "B if A" is "P(B|A)". (I did misleadingly say that "P(B|A)" was equivalent to "P(B if A)", but that was for a natural-language reading of "B if A", and that equivalency is only problematic when "B if A" is taken as a strict material implication). — Pfhorrest

Your claim above is

This still accords with Bayesian reasoning, because you could reason along the same lines, but probabilistically instead of in those absolute statements. If the thing you're observing is very likely to be a real purple swan given your background beliefs, and yet it's very likely that all swans are white given what you believe about swans, then what you're observing must be very improbable. — Pfhorrest

So let A be "the thing you're observing is very likely to be a real purple swan given your background beliefs" and B be "all swans are white given what you believe about swans". You're saying that the probability of A and B ("then what you're observing must be very improbable") is P(A and B), but it is not - under Bayes - it's P(A|B) which is a different calculation.

Neither confirmationists, nor fideists, nor nihilists, nor any of your stated targets believe they can predict the future with 100% percent accuracy. So none of them are disproven by your statement that observing something surprising contradicts a belief (an expectation) that we wouldn't. — Isaac

You’re the only one bringing up any belief about being able to predict the future perfectly. That’s not anything I’m talking about at all.

I’m saying that if you see something and think “whoa a black swan, I didn’t think those were possible...” and then either “I guess they are possible after all” or “it must be fake somehow”, you’ve revised the beliefs you initially had. (Switching to black for this example to illustrate how either is a plausible option).

If instead you see the same thing and think “oh look, somebody painted that swan black...” then you don’t have to revise any beliefs because a fake black swan is what your background beliefs initially lead you to perceive and that doesn’t contradict any other beliefs such as that all swans are white.

You're saying that the probability of A and B ("then what you're observing must be very improbable") is P(A and B) — Isaac

No, I’m saying that if P(A) is small and P(B) is small then P(A)*P(B) is small. “P(A)*P(B)” is the probabilistic equivalent of “A and B” in the same way that “P(A|B)” is the probabilistic equivalent of “A if B”.

But it's also possible that different processes for coming up with beliefs will be more or less productive in coming up with beliefs that are unlikely to be falsified. Believing that patterns you've observed are likely to continue (i.e. induction) could very well be one of those safer methods (and I'm intuitively inclined to say it probably is, but I don't have any arguments to that effect). — Pfhorrest

For me the fundamental presupposition of an invariant nature, without which neither confirmation nor falsification could gain any traction, or even have any meaning, is indispensable to all our investigations and conjectures. That is the basic issue I have with the idea that we can believe whatever (in principle falsifiable) things we like as long as they haven't been falsified. I don't think we actually do, quite apart from whether we ought to, believe, or even entertain, any such things in any case; usually our hypotheses are extrapolations from, and both coherent and consistent with, the vast store of what we take to be observationally and experimentally confirmed scientific knowledge.

I’m saying that if you see something and think “whoa a black swan, I didn’t think those were possible...” and then either “I guess they are possible after all” or “it must be fake somehow”, you’ve revised the beliefs you initially had. — Pfhorrest

So, here's a great example: you cite just two possibilities; the impossibility that the laws of nature could have suddenly changed and things might just morph at random from white to black to purple or whatever is implicit in the way you arrive at what you consider possible. This is entirely on account of inductive expectation; it's not logically prescribed that the laws of nature cannot change.

So, here's a great example: you cite just two possibilities; the impossibility that the laws of nature could have suddenly changed and things might just morph at random from white to black to purple or whatever is implicit in the way you arrive at what you consider possible. This is entirely on account of inductive expectation; it's not logically prescribed that the laws of nature cannot change. — Janus

If the laws of nature had just changed such that swans could now change color willy-nilly, or even if they had always been such that that was possible, then that would make “all swans are white” no longer (or maybe never have been) true, so changing your belief that all swans are white would cover that.

More generally though, on this topic of the laws of nature not changing: that is not something we believe because of induction, but something we must believe to do induction. If we don’t assume that that’s the case, then there is no reason to expect patterns to continue as we have seen them do thus far.

On my account, that is one of the background assumptions not about the content of reality per se but which we use to structure our experience of reality whatever its contents should be. It’s up there with assuming there are physical substances that bind together all of the attributes of things and not just improbably coincidental constant conjunctions of the same attributes moving through space in unison.

These are things like the assumption of objectivity about which we could not possibly know one way or the other whether they are true, but which we cannot help but assume one way or the other through our actions, and without the assumption of which we could not possibly hope to ever know anything, thus pragmatically requiring us to always act as though they are true or else give up all hope of knowledge.

More generally though, on this topic of the laws of nature not changing: that is not something we believe because of induction, but something we must believe to do induction. If we don’t assume that that’s the case, then there is no reason to expect patterns to continue as we have seen them do thus far. — Pfhorrest

I disagree: I think we believe in an invariant nature because that is all we have ever experienced; that's induction in a nutshell. That's not even an arguable point as far as I am concerned. If you disagree with that, then I can only wonder what planet you've been living on.