Talk:Problem of induction/Archive 3

Archive 1

Archive 2

Archive 3

The historical considerations

I consider the sentence:

The most famous formulation of the problem was proposed by David Hume in the mid-18th century, although versions of the problem date back to the Pyrrhonist school of Hellenistic philosophy and the Cārvāka school of ancient Indian philosophy.

There might be an anachronism here when we refer to Indian philosophy and perhaps also the Hellenistic philosophy as presenting the problem of induction. No evidence is given that it was seen as a problem at the time. Especially in the case of the ancient Indian philosophy, it is very likely that they simply knew that inductive justifications were not valid and that is it, no problem. Unless sources are provided that show that these ancient philosophies supported the inductive view, at the least during a period of time, I propose that we simply mention that they rejected inductive justifications without suggesting that it was seen as a problem. This means that to our knowledge the most ancient formulation of the problem was given by Hume. This is, for example, the position taken in SEP. I suggest the following

The original source of what is known as the problem today was proposed by David Hume in the mid-18th century, although inductive justifications were already argued against by the Pyrrhonist school of Hellenistic philosophy and the Cārvāka school of ancient Indian philosophy in a way that shed light on the problem of induction.

Dominic Mayers (talk) 18:01, 9 December 2020 (UTC)

"Problem" appears here to have some slipperiness. The bulk of the surviving works of Sextus Empiricus are about flaws in the justifications for knowledge provided by the dogmatists. "Invalid" is a little slippery too. The Pyrrhonists were explicitly fine with using induction to infer that if there was smoke there was fire or that if there was a scar there had been a wound. Such things in Pyrrhonism are "commemorative signs" involving "evident" matters and are fine (albeit prone to occasional error) for dealing with practical matters. The problem the Pyrrhonists were pointing out that induction from the evident to the non-evident could not provide adequate justification for knowledge (e.g., lightning is a sign that Zeus is displeased). This was just one of many Pyrrhonist arguments regarding the lack of adequate justification for knowledge, e.g., regress and circularity, found in the Five Modes of Agrippa. The one major fragment we have from Pyrrho, about 500 years before Sextus, appears to be substantively about the problem of induction.

In addition, plenty of people were aware of the issues with induction before Hume. "Original source" seems to be an overstatement of the case. Consequently, the sentence proposed for replacement seems to be more accurate than the proposed replacement. Teishin (talk) 19:52, 9 December 2020 (UTC)

You might be right. I am just saying that until we have the sources to verify this view, we need to restrict the article to what is verifiable. Right now, what we have is SEP that says "The original source of what has become known as the “problem of induction” is in Book 1, part iii, section 6 of A Treatise of Human Nature by David Hume." Please help us by providing precise references with sections and even pages. Ideally, even point to key sentences. Also, ideally in recent sources. Dominic Mayers (talk) 20:29, 9 December 2020 (UTC)

"Verifiable" is sort of a matter of having a source, and of course there are sources for all sorts of things that cannot be verified.... I tend towards caution with such claims. Of course, Sextus did not put the problem in the same way Hume did, hence the slipperiness of "problem", but it seems difficult to exclude Sextus from having addressed the "problem of induction". I have tracked down a very fresh 2020 book that addressed the topic, Epistemology After Sextus Empiricus edited by Katja Maria Vogt, Justin Vlasits. Here's a bit from it:

I want to isolate and explore what is puzzling about induction. Only with a clear sense of the problem can we begin to understand what a satisfactory solution would be. The normal way into the puzzle is through Hume’s justly famous argument against induction. In section 2, I argue that, interesting as Hume’s argument is, it fails to isolate what is puzzling about induction and, in fact, crucially presupposes that induction is puzzling. In section 3, I analyze an argument in Sextus Empiricus against induction—as far as I know, the very first in the history of philosophy. This argument does, I claim, get at what is interesting and puzzling about induction. Sextus’ argument has received far less attention than Hume’s.

— Justin Vlasits, chap. 11: The First Riddle of Induction

Teishin (talk) 21:14, 9 December 2020 (UTC)

I read the chapter up to the passage and a few pages after. The chapter seems to agree with both the inductivist view and the arguments against it, not a comfortable position to hold, but that's not a relevant concern here. A relevant point is that it does not appear from Vlasits's citation of Sextus that the latter saw a problem with the lack of justification for induction. I see instead that Vlasits found that Sextus's argument against inductive justification sheds more light on the problem of induction than Hume's argument. So, it is Vlasits, not Sextus, that sees the problem of induction here. Sextus only provided a nice argument against inductive justification. In my understanding, it is very unlikely that these people saw a problem. They were not in a context for that. In any case, even if Sextus expressed a support for the inductivist view, which is a requirement to feel a problem, it would still remain that it is Hume's view that it is at the origin of the problem as we know it in the following sense that modern philosophers learned it from Hume and Hume did not learn it from Sextus. You know that it's not that important as far as conveying the key idea is concerned. My only concern here is that I do not want to support an anachronism by assuming that the inductivist view was sufficiently strong at the time to be at the basis of a problem of induction. Dominic Mayers (talk) 00:36, 10 December 2020 (UTC)

I modified the sentence in view of this discussion. Dominic Mayers (talk) 00:54, 10 December 2020 (UTC)

The modified sentence seems fine. (As to whether Hume learned it from Sextus, it's awfully fishy that Hume went to some effort to mischaracterize and dismiss Pyrrhonism and then in his own writings to largely re-invent Pyrrhonism.) Teishin (talk) 19:48, 10 December 2020 (UTC)

Correct or replace?

Continuing to try to correct present article is recognition that something must change. Can we agree on what needs change? Is anyone making classification errors? Dom seems to want to ignore the whole social science effort to deal with the "problem of induction." I find that rank philosophical bias in favour of traditional classifications. If many social scientists say they are dealing with the problem, how can WP editors judge them wrong?

Allow me to explain how I came to describe and judge the present article biased and obsolete.

I read S&L “Problem of induction” after reading WP “Problem of induction.” I quickly recognized the bias of the WP article because it ignored the range of information in S&L. I classified it as violating WP NPOV rule. This classification is at once a descriptive fact and a valuation. See columns 1&2 of coverage table.

I thought the bias could not be eliminated by adding S&L info to present article--WP standard operating procedure--because S&L content and definitions framed problem differently. I was delighted to learn that SANDBOX provided means for me to attempt revision. See column 3 of coverage table. Editors Dominic Meyers & Biogeographist, unaware of my evidence of bias, logically classified my sandbox revision as violating NPOV & NOR rules and reading like an essay—unencyclopedic. SK2242, equally unaware of the evidence in my table, accepts these classifications.

My proposed revision does not violate NPOV because it describes modern alternatives to present WP article. It partially corrects existing bias.

My proposed revision does not violate NOR because it reports existing scholarship in S&L. Reporting existing scholarship requires comparing modern classifications with traditional, regardless of whether or not it appears “encyclopedic.”TBR-qed (talk) 15:42, 10 December 2020 (UTC)

I believe that there is a good consensus on what is the subject "problem of induction". It is the same old problem of induction that was raised by Hume, the one that involves justifications for laws given observations, the kind of justifications that were considered in ancient India and ancient Greece. This subject is not obsolete at all. One rule in Wikipedia is that the subject of the article must be well defined so that we can evaluate accordingly the pertinence of information to include. If you do not accept the subject, because you think it is obsolete, then you are against the consensus. This being said, it does not mean that nothing should be said about how some recent views on induction is related to this subject. It will be very helpful that you accept the subject as it is and explain how the information that you want to include is pertinent given that subject. Note also that any controversial information to include would require more than a single source. You will have to convince us through a few sources in philosophy of science that this information reflects the opinion of at the least a minority of experts on the subject "the problem of induction" as defined here so that the link with this subject is pertinent. I am not writing this with the belief that you cannot succeed. I think that there might be something there that would improve the article, but we need the sources and of course, a specific information to include. One idea would be that you write one or two sentences that could be added to the lead as it is now to show the link with the subject as it is defined now. Dominic Mayers (talk) 16:42, 10 December 2020 (UTC)

TBR-qed described how he first read this article, then read Sloman & Lagnado, and then decided that information from Sloman & Lagnado needed to be in this article. His conclusion may be partially correct: it may be true that some of the information that he encountered in Sloman & Lagnado belongs in this article. But what is missing is larger contextualization of Sloman & Lagnado's chapter: Where does Sloman & Lagnado, and the information they present, fit in the larger body of knowledge that this article is supposed to summarize? As Dominic said, we need the sources. TBR-qed's arguments have been repetitive so far because he is just comparing three sources: the present article, Sloman & Lagnado, and his draft. There needs to be comparison among a larger body of secondary and tertiary sources, not just Sloman & Lagnado. Biogeographist (talk) 16:04, 11 December 2020 (UTC)

My personal view, in case it might help thinking about the article.

I know that no original research is allowed in the article, but sharing our understanding in the talk page can help the discussion. Besides, what I am going to say is not really original research in the following sense that many others have noticed the same thing. I see that before Popper, induction was strongly associated with justification, especially in the Vienna Circle. It was important at the time to separate science from non science and the hoped difference was that science is justified knowledge and that every thing else was not even meaningful. I think that after Popper it has becomes generally accepted that justified knowledge was the wrong objective, though there is still a remain of justificationism for knowledge that is uncertain. But, even among those that stopped to fight for justificationism in any of its forms, induction was not abandoned. This new trend is that induction does not have to be justified, because it is an observed fact. This is a complete shift in comparison with the pre-Popper notion of induction. This notion of induction is not at all opposed to Popper's philosophy. In fact, it perfectly goes along with it, because it starts with the premise that justification is not needed. They would even say that, for that reason, the problem of induction does not really exist, which is exactly Popper's point. I find it weird that those who support this notion of induction pooh-pooh Popper's philosophy. The explanation, I believe, is that the true meaning of justificationism is lost. In artificial intelligence, they create machines that can infer laws from observations. The principles used by these machines are considered as some kind of justifications for the laws and they are called inductive principles. Similar notions have spread in social science. However, these inductive principles are not universal. They have nothing to do with the kind of justifications that were hoped for in pre-Popper time. There was a very good reason for the pre-Popper time justificationism. It was puzzling in ancient Greece, for Hume and in Popper's time that no true justifications can be found and it is still puzzling today. It's not obsolete at all. Dominic Mayers (talk) 18:58, 10 December 2020 (UTC)

This is just a response to Dominic's personal view, not so relevant to the article: One reason why some people are not so impressed by Popper's account of "justificationism" may be because they are already epistemological fallibilists with no help from Popper. Epistemology cannot be simply divided into infallibilist/foundationalist "justificationism" before Popper and fallibilist/nonfoundationalist "non-justifications" after Popper. There was fallibilist epistemology before Popper (e.g., C. S. Peirce), and there has been non-Popperian fallibilist epistemology during and after Popper's career (e.g., Neurath's boat). One need not be a Popperian to appreciate the absence of absolute foundations of knowledge. Biogeographist (talk) 16:04, 11 December 2020 (UTC)

Part of the lack of agreement might be due to definitions. For me, fallibilism and anti-justificationism are different in at the least two ways. First anti-justificationism rejects even the justification of uncertain inference. So, in that sense, it is stronger. Second, anti-justificationism is by definition only concerned with (rational) justification at the level of objective knowledge and therefore does not concern itself with ways to gain knowledge that operate between subjective knowledge (expectation, predisposition, etc.) and objective knowledge. Popper's view is that non objective ways of gaining knowledge are fallible, involve chance, a lot of it, but they are nevertheless not a part of anti-justificationism. Anti-justificationism per se does not exclude the possibility of some infallible aspects in subjective ways to gain knowledge. So, in that other sense, it is weaker. I disagree that it's irrelevant to the article. It's not directly relevant, but it's indirectly relevant, because if we do not agree on this kind of notions in the talk page, then it is very difficult to understand each others when we discuss the article. Dominic Mayers (talk) 18:33, 11 December 2020 (UTC)

I think I understand what you're saying; I don't see any misunderstanding here on my part. You're basically summarizing Popper's philosophy. My point is that not all epistemology before Popper was "justificationism". This is not a refutation of anything you said, since you didn't explicitly claim that all epistemology before Popper was "justificationism". I am just pointing out something you omitted, since you only mentioned Popper and "pre-Popper-time justificationism". Biogeographist (talk) 19:03, 11 December 2020 (UTC)

Another way of phrasing my point: Someone who is not a justificationist has little use for anti-justificationism. Biogeographist (talk) 19:09, 11 December 2020 (UTC)

I suspect that this is the situation with many modern "inductivists". They are inductivists that are not justificationnists. They have little concern for this obsolete view, they would say. But, this is a big source of confusion, because in the problem of induction, as defined here, the inductivist view is a justificationist view. If it was not, then indeed, there is no problem of induction. I am sorry if I referred to Popper only, but Popper was not in the same context as Peirce. He came at a time where the justificationism view was at its extreme value. So, he directly addressed the issue. Peirce might have done better than Popper had he been in a similar context. Popper had a lot of appreciation for Peirce. Nowadays, (classical) justificationism is replaced with what I would call synthetic or non universal justificationism, the one that emerges in artificial intelligence in particular. There is no problem (of induction) with this kind of justificationism. I personnally see it as a natural continuation of the death of classical justificationism and as such it does not make the original problem obsolete. In particular, it does not remove, though it might contribute to hide, the puzzling aspect of a lack of classical justification. My point is that we need primary and secondary sources to include this kind of considerations in the article. Until we have secondary sources that make the connection with the original problem of induction, the article cannot cover that. Dominic Mayers (talk) 19:58, 11 December 2020 (UTC)

I agree with all that. Biogeographist (talk) 20:31, 11 December 2020 (UTC)

coverage table

TBR-qed reports that the table below was not properly formatted and will be replaced.--Quisqualis (talk) 20:15, 12 December 2020 (UTC)

I mistakenly thought my coverage table was posted here. Here it is.TBR-qed (talk) 17:45, 9 December 2020 (UTC)

TABLE OF SCHOLARS IN 3 VERSIONS OF “Problem of induction.”

|- column 1 = EWP; column 2 = S&L; column 3 = SBox

Philosophers	Hume	Hume	Hume
	Popper		Popper
	Goodman	Goodman	Goodman
	Pyrrhonists	Logical empiric	Pyrrhonists
	Carvaka	Hempel	Carvaka
	Quine	Quine	Dewey
	Stove & Williams	Miller & Lipton
	Campbell & Costa	Carnap
	al-Ghazali & Ockham	Hacking
	Scotus	Nagel
	Salmon	Kuhn
Hard scientists	0	0	Duhem
Social scientists	0	Rosch	Rosch
		Tversky
		Kahneman	Kahneman
		Shepard
		Rips
		Carey
		Sloman
		Osherson
		Lopez
		Hampton
		Gelman & Coley
		Mandler & McDonough
		Nisbett
		Gopnik & Meltzoff
		and more.

Draft:Problem of induction

Draft:Problem of induction is a proposed rewrite of Problem of induction. It was declined by User:Theroadislong for two reasons, both because it reads like an essay, and because it is a rewrite of an article, which is not how Articles for Creation works.

There has been extended discussion of the content of this article. Three years ago, this might have been a good case for formal mediation by a member of the Mediation Committee, or even longer ago for informal mediation by the Mediation Cabal. Two years ago, in a remarkably stupid action, the community disbanded the Mediation Committee. This content dispute is too large to be handled by the dispute resolution noticeboard. Without studying the details of the content dispute, it appears to me that a mediator is needed, and that the mediator should largely assist in the formulation of multiple Requests for Comments.

This draft was resubmitted, and I removed the resubmission tag while moving the commentary to the talk page. I restored the resubmission in order to decline it again with comments. I recommend a combination of mediation and Requests for Comments. Robert McClenon (talk) 00:59, 14 December 2020 (UTC)

Improving this article is going to be a long process which will require a lot of proposals by editors, each one for some step of progress, not to completely change the article. Unless we have proposals, I don't see what would be the specific issue that a rfc will be able to decide. What will be useful is more people with some expertise on the subject that want to put the time needed to make specific proposals or discuss proposals made by others. This cannot be a one shot process. There is a consensus that a coverage of the cognitive science + AI perspective should be added, but we need specific proposals that are based on a few secondary sources and a good will to discuss them with an open mind. The draft was declined formally, but the main issue is that it rejected totally the current view of the article as being obsolete. The draft might contain elements that are useful for the article, but TBR-qed should make a specific proposal for some step of progress without totally rejecting the article. I have not edited this article until recently and I think it was good even before my edits. I suspect that it reflects a lot of consensus to which I add my support and I believe that other participants here do the same. It does not mean that it cannot be improved, but throwing it away as being obsolete makes no sense. Dominic Mayers (talk) 01:54, 14 December 2020 (UTC)

Well, in a second thought, a rfc might be useful to get some general opinions or guidelines from the community. Dominic Mayers (talk) 03:33, 14 December 2020 (UTC)

Lead

The leading sentence is too complicated, and without reason, because the "knowledge understood in the classic philosophical sense" is the same thing as "justified knowledge". Therefore it does not need to mention both the justification and the "classic philosophical sense". My suggestion was "The problem of induction is the philosophical question of whether inductive reasoning leads to knowledge understood in the classic philosophical sense...", but there are other possibilities, how to solve it. In any case, the current lead should be changed because it lacks the required conciseness.Ioannes Pragensis (talk) 07:31, 14 February 2021 (UTC)

For many readers "knowledge understood in the classic philosophical sense" does not mean much more than "knowledge in the usual sense". I don't think it conveys the idea that a justification (other than a self justification) is searched, thus is missing and that it might not even be possible. For other readers, if the notion of justification is seen in this expression, it is in the form of a belief that normal accepted knowledge in practice is justified knowledge. With such a belief, any description of a method that leads to that justified knowledge is the solution to the problem of induction, that is, the method is confused with the justification. Dominic Mayers (talk) 09:30, 14 February 2021 (UTC)

Rethinking the first sentence of the lead.

@Ioannes Pragensis, Biogeographist, and Teishin: what is your opinion on the following proposal, an attempt to make the first sentence in the lead simpler. The current sentence is:

The problem of induction is the philosophical question of what are the justifications, if any, for any growth of knowledge understood in the classic philosophical sense—knowledge that goes beyond a mere collection of observations—highlighting the apparent lack of justification in particular for:

1. Generalizing about the properties of a class of objects based on some number of observations of particular instances of that class (e.g., the inference that "all swans we have seen are white, and, therefore, all swans are white", before the discovery of black swans) or
2. Presupposing that a sequence of events in the future will occur as it always has in the past (e.g., that the laws of physics will hold as they have always been observed to hold). Hume called this the principle of uniformity of nature.

I don't think it is so bad, but I propose to change it to

In philosophy, the problem of induction is the apparent or true absence of justifications for a growth of knowledge that goes beyond a mere collection of observations as understood in inductive reasoning, in particular for:

1. Generalizing about the properties of a class of objects based on some number of observations of particular instances of that class (e.g., the inference that "all swans we have seen are white, and, therefore, all swans are white", before the discovery of black swans) or
2. Presupposing that a sequence of events in the future will occur as it always has in the past (e.g., that the laws of physics will hold as they have always been observed to hold). Hume called this the principle of uniformity of nature.

I removed the reference to "knowledge understood in the classic philosophical sense", which is linked to justified true belief, a concept that gained approval during the Enlightenment, "justified" standing in contrast to "revealed". The concept of justified true belief in relation with the Gettier problem, etc. is still discussed in epistemology in general, but not in the article. For our purpose, the concept of inductive reasoning does a better job in distinguishing revealed knowledge from scientific knowledge. So I replaced it with "as understood in inductive reasoning". To simplify further, I also used "apparent lack of justification" from the start instead of highlighting it later. Also, I made the sentence more neutral by adding that the lack of justifications might not be apparent only, but be actually a real absence of justifications.

Dominic Mayers (talk) 19:40, 17 February 2021 (UTC)

I'm fixin to rewrite the lead

Right now the lead, as Harry Belafonte sang, is clear as mud, but it covers the ground. It confuses and discourage anyone who doesn't already know what it is talking about. It's important to think of the principle of least astonishment WP:ASTONISH, as well as the MOS:LEAD. I will be building it around this formulation of the problem: "Therefore, for Hume, the problem remains of how to explain why we form any conclusions that go beyond the past instances of which we have had experience" from the Stanford Encyclopedia, as well as "On what grounds we can expect the future to resemble the past". I know there are a million reasons why this doesn't cover the ground, What about thinkers before Hume? What about.... The answer is, we can't cover everything in the lead. We owe the average reader a comprehensible, not a comprehensive, explanation. The rest of the article can cover all the rest of the ground. DolyaIskrina (talk) 22:41, 27 July 2022 (UTC)

Here's a first whack at it:

First formulated in Western analytic philosophy by David Hume, the problem of induction questions our reasons for believing that the future will resemble the past, or more broadly it questions predictions about unobserved things based on previous observations. This inference from the observed to the unobserved is known as “inductive inferences”, and Hume, while acknowledging that everyone does and must make such inferences, argued that there is no non-circular way to justify them, thereby undermining one of the Enlightenment pillars of rationality.

DolyaIskrina (talk) 23:24, 27 July 2022 (UTC)