Talk:Problem of induction/Archive 2

Archive 1

Archive 2

Archive 3

Plan suggested by TBR-qed

These are explanations of TBR-qed for his plan:

S&L are indeed psychologists who focus on the “psychology of categorical induction.” But their first 2 sentences—framing of the problem with Hume and the “fact” of sunrise— connects them directly to the WP article example of sunrise & its broader concern for justification. That is why I suggest relating 4 more scholars to S&L’s sample to show how modern scholars superficially adopt Goodman’s formalized knowledge while ignoring philosophical concerns—with negative results. I add Duhem & Dewey to show long-standing rejection of enumerative induction. I add Popper (who accepted the validity of sunrise) to support Hume & to critique S&L “induction as scientific methodology,” that conflates method of Duhem & Dewey with modern social science methods, including economist Thaler. I would leave [Inductive reasoning] & [Inductivism] unchanged to report more traditional treatments of induction.

I have no opinion, because I don't have yet a clear understanding of the plan. So, I will only mention points that come to my mind.

Regarding S&L are indeed psychologists who focus on the “psychology of categorical induction” I am guessing that the key word here is "psychology", because I don't see what would be an example of a "non categorical induction", except the cases where we have difficulty to make the induction because we are uncertain about the category. Therefore, in a way, it's always categorical. Collections in which the similar elements are distinguished by space-time coordinates are special kind of categories, but I don't see that the arguments to reject induction are specific to this kind of categories.

Regarding But their first 2 sentences—framing of the problem with Hume and the “fact” of sunrise— connects them directly to the WP article example of sunrise & its broader concern for justification I should point out that the example of sunrise in the article is biased. There is an implicit assumption that there is an observed induction and, moreover, that this observed induction is explained by the use of an ampliative rule, which thus need justification. As editors, we do not have to start the article by making a big assumption with no basis. No matter what the sources say, our terminology and our way to organize the article is our choice as editors. A source makes this assumption or it does not make it and it must be used accordingly in our article. But, I don't think that you had that aspect in mind when you wrote that sentence. Your point, I believe, is that you agree that we should have a link with the philosophical concern for justification.

Regarding That is why I suggest relating 4 more scholars to S&L’s sample to show how modern scholars superficially adopt Goodman’s formalized knowledge while ignoring philosophical concerns—with negative results I did not know much about Goodman, but I just read [1] and I found it very interesting theoretically, but also a very good example of a source that makes the big assumption that induction exists as a set of ampliative inference rules, not only as observed processes. So, he goes into a very interesting (but useless) argument how these inference rules can be "justified", just as we "justify" deductive rules. But, there is no evidence at all that ampliative inference rules are used. We only have evidence that inductive processes occur, which is completely different. These observed inductive processes do not need to be justified, as many pointed out. They only need to be explained and the simplest explanation is the use of an hidden background knowledge together with normal non ampliative rules.

Regarding I add Duhem & Dewey to show long-standing rejection of enumerative induction this seems to suggest that only this particular form of ampliative inductive reasoning is ruled out. That is a big claim. Of course, (observed) inductive reasoning does not need justification, because it is observed. However, I would say that any form of ampliative rule is ruled out and I would be very surprised that no argument was provided in the literature to reject the extraordinary claim that an ampliative rule, i.e., a fallacy, can be useful.

Regarding I add Popper (who accepted the validity of sunrise) to support Hume & to critique S&L “induction as scientific methodology,” that conflates method of Duhem & Dewey with modern social science methods, including economist Thaler I think the main contribution of Popper on the problem of induction is his very natural proposal that the use of ampliative rules is a myth and that the observed inductive processes are more naturally explained by an evolutionary process that created the required hidden background knowledge. He used a different terminology. What I call "ampliative rules" here, Popper called it "induction" and, what I call "observed induction processes" here, Popper called it "growth of knowledge".

Regarding I would leave [Inductive reasoning] & [Inductivism] unchanged to report more traditional treatments of induction, yes let us focus on this article.

Dominic Mayers (talk) 01:52, 9 May 2020 (UTC)

I hope you will be able to access my sandbox which outlines my plan. Your first paragraph above comes close to my main concern: uncertainty about the category. I find endless disagreement about operations called inductive and about their consequences. Your next paragraph says the sunrise example is biased, which is precisely why I discuss its widespread use as demonstrating the inseparability of description and justification. I think I agree with your point about editor choice, but do not understand "an ampliative rule, which thus need justification." Does your next paragraph saying "induction exists as a set of ampliative inference rules" identify operations, or conclusions of operations, or explanation of operations? TBR-qed (talk) 14:53, 9 May 2020 (UTC)

Subsection response to sandbox for easier editing

Please relate your answers to Popper's practice and to my section on sunrise. Thanks.TBR-qed (talk) 14:53, 9 May 2020 (UTC)

After having read your sandbox and considered your questions here, I think my paragraphs could have been clearer, but the main issue is with the terminology. We must first clarify the terminology before I can discuss Popper's practice, the section on sunrise and answer your questions about my paragraphs.

There is an important difference between the concept of an observed process in which some conclusion is drawn given some premises and the concept of rules of inference that can be used to explain this process. I could replace "explain this process" with "justify this process", which is what we say when inference rules explain a process, but I prefer the term "explain" because it brings out the notion that these processes are like other natural processes that need to be explained.

extra explanation

Consider the following incorrect argument from the observed q="here is a white swan" to the universal p'="all swans are white". Seeing this argument, one might conclude that the fallacy ∃x α(x) ⊧ ∀x α(x) was used (I made up that fallacy, but it more or less corresponds to what is some times called Hasty_generalization). However, the person who draw the incorrect conclusion p' might say, no I used the rule p → q, q ⊧ p, which is the fallacy affirming the consequent. Indeed, if we take p= p'∧q, then we have p → q. So, from the observed q, using the fallacy, we can infer p and then logically infer p' (because p= p'∧q). The details are not important, we could have many other examples, to illustrate that there is an important difference between an observed process in which some conclusion was drawn given some premises and the actual rules of inference that were used. We cannot know what were the rules, unless the person writes in the margin the rules that were used. Even this can be misleading, as illustrated by Newton's famous claim that he used inductive inference rules to infer his theory from observations alone, which we know cannot work. (On purpose, I have not respected the terminology used by Newton, because there would be a cyclic issue if I tried to explain a terminology by using a different and non consistent terminology.)

There are two distinct concepts: the processes that need to be explained and the rules that can explain them. A confusion between these two concepts might arise from the use of the expression rules of induction by S&L in sentences such as:

... normative rules of inductive inference that are generally accepted in the scientific community. One such rule is that properties that do not vary much across category instances are more projectible across the whole category than properties that vary more.

These are rules to classify observed processes, not to explain them. They are not basic inference rules similar to the deductive rules of inference. They exist at a completely different level.

extra explanation

These rules have not much to do with the deductive or inductive explanatory rules that could explain them, just as we have illustrated before. The explanatory inference rules are hidden and they cannot be justified by these observed rules, because there are a lot of things behind these observed rules, a lot of background knowledge, etc. To put it in another way, the inductive method of reasoning considered in S&L are emerging methods. They emerge from a background knowledge, which includes some fundamental empirical or perhaps metaphysical laws that are assumed. They are not basic inference rules similar to the deductive rules of inference. They exist at a completely different level. Some of these rules could be theorem based on some properties assigned to the categories, but that would be almost the same thing: we would need laws to verify that these properties hold.

I did not prove my previous claim and I cannot formally rule out the possibility that inductive rules might constitute a different logic for truth in science, but if it was the case it would be a very very important new branch of logic. In any case, S&L cannot prove that they have such a logic and it is a big assumption. The terminology should make clear from the start that there is another option, which do not require a new logic. In this other option, the so called inductive rules in S&L are not (explanatory) inference rules, but just rules used to classify observed processes. This other option uses only the standard, deductive, inference rules together with the assumption of a complex background knowledge provided by the evolutionary process. Because this other option does not require a new logic, it is in my view much easier to believe.

extra explanation

The point here is that, if someone has discovered a new inference rule that is ampliative, that is a big big deal. Let's keep in mind that we are talking about the "truth" of scientific laws, after corroborations. If we have a new logic with ampliative rules for this kind of truth, it is a big big deal. So big that it should totally change the face of logic. In fact, historically, logic has always been the logic of the practical truth used in science. Mathematics itself is only a tool for science. So it should lead to a new branch of mathematics based on this new logic of science. So, I am sorry, but S&L have not proposed that. I am aware that Goodman proposed a very interesting way to "justify" such a new logic using a notion of equilibrium, but this does not mean that we have found this new logic and, even if we have found such a new logic, it does not mean that it is the one used in science. Again, it would be a very big deal. In any case, our article cannot be written under this assumption. So, it should be only an option and the terminology should make it clear, from the start, that there is another option.

Given that the explanatory rules are abstract, not observed and not uniquely determined, should we ignore them as possible explanation for what is observed? Of course not. We have the same problem with the question what law is used when an apple falls from the branch of an apple tree to the ground. We cannot say that it is Newton's law, because one could rightfully reply that it could also be Einstein's law. Nevertheless, we do not abandon the question what law explains an observed phenomena. In the same way, inference rules are like laws that can explain observed reasoning processes. So, in our terminology, it is important to distinguish between observed inductive reasoning and (explanatory) inductive rules. Otherwise, the problem of induction cannot even be formulated correctly.

In contradistinction, if we use the correct terminology, then the problem can be formulated in one short sentence. The problem is how to explain observed inductive processes without using (explanatory) inductive rules or else how to justify these (explanatory) inductive rules. It can even be shorter : how to rationally explain the observed inductive processes. This is because justifying the (explanatory) inductive rules simply means arguing that they can be rationally used to explain the observed reasoning processes. Here, we see a closely related terminological distinction that is also important: explaining observed reasoning processes and justifying hidden inference rules are two different things, simply because hidden inference rules and observed reasoning processes are different things.

Here is the first paragraph in your sanbox:

The problem of induction is the philosophical question of whether knowledge humans gain by inductive reasoning can be as warranted as knowledge they gain by deductive reasoning—the alternative method of human reasoning.[1] How can one infer that observing a unique sample of an object or event is evidence of an unobservable universal kind or class of object or event?

I don't see in this paragraph where (ampliative) inductive rules enter into play. I don't see the two options: justifying these inductive rules or explaining the observed inductive processes without using these inductive rules. My personal opinion is that Popper is right that the first option is pointless, because there is a much simpler explanation that do not require the inductive rules. In any case, the problem of induction is certainly not how to do the first option only.

extra explanation

I am guessing that the paradigm in your first paragraph is simply that we can ignore or at the least not refer to inference rules as a way to explain observed reasoning processes. The suggested paradigm is that we have different kind of observed processes that we classify in terms of rules, the inductive ones and the deductive ones, but these rules are only ways to classify processes, not some kind of hidden laws to explain them. The question then becomes how the (descriptive) inductive rules can be as warranted as the (descriptive) deductive rules. This is confusing. It hides the possibility that the inductive rules (as description of observed processes) can be explained without using inductive inference rules (as laws used to explain observed processes). A better paradigm is that we simply want to explain what is observed using rules of inference and we accept that these rules interact and must be considered as one set of rules. It is fine to consider the option of adding inductive rules to this set, but it must be only one option. Another option is that we don't need to do that, because there is another way to explain what is observed.

I am not attach to any particular terminology as long as we use a terminology that does not conflate different concepts and that we use it consistently from the start. Dominic Mayers (talk) 17:57, 9 May 2020 (UTC)

I think we are now down to cases. I don't understand "(ampliative) inductive rules." I think it involves your effort to distinguish between explaining processes and explaining rules, which are both distinct from observing processes. But I think the inductive method means the following: 1)WE have a conjectural proposition--the sun [object assumed known in unspecified conditions] rises. 2)We ask whether observation [test statement] is true or false. Will it rise again [a conjectural universal predicate]? 3) Popper claimed we can't answer TRUE, but may be able to answer FALSE if it does not rise. 4)Absence of failure to rise temporarily confirms both conjectures as best tested theories.Objective Knowledge:10,20,26.

If determining what predicate is to be observed be excluded from inductive method, then Popper's purely logical conclusion follows. Duhem & Dewey show the error of this method, and sunrise instantiates it. My sandbox tries to set this scene. Do I fail, or am I mistaken?TBR-qed (talk) 13:39, 10 May 2020 (UTC)

Subsection perspectives on the problem of induction for easier editing

It seems that my attempt to explain my understanding of the perspective of Popper and many other philosophers on the problem of induction failed. You seem to say, I paraphrase, "I don't understand because here is my perspective and I cannot see how your perspective match with my perspective." It is surprising that you don't see a match, because you seem to attribute your perspective to Popper (Objective Knowledge:10,20,26). So, instead of trying again to explain "(ampliative) inductive rules", etc., I will try to discuss your understanding of Popper, your understanding of the inductive method and above all your understanding of the problem of induction. For a start, can you give one or more excerpts from Objective knowledge that correspond to the inductive method that you attribute to Popper, because I don't see where Popper has described the inductive method as you suggest that he did. If it is not from Popper (and the reference Objective Knowledge 10, 10, 26 was a mistake), please provide the correct reference and some excerpts, so that I can be certain about your sources. Dominic Mayers (talk) 19:21, 10 May 2020 (UTC)

Let me explain how I got from Popper’s Objective Knowledge to my second sentence. OK starts with Hume’s and the common sense definition of method of induction: “reasoning from [repeated] instances of which we have experience to other instances [conclusions] of which we have no experience.”(4). Popper renames Hume’s experienced instances “test statements” and his potential instances “explanatory universal theories.”(7)

S&L open their article defining the method as reasoning from “observed fact” to unobserved “conclusion.”(95) They then define a “rule of induction by enumeration” as “a universal hypothesis H being confirmed by positive instances E”(97) and the “propensity to group kindred entities and [to] project [Goodman’s name for potential universal theories] them into the future.(99) S&L eventually name the method “categorical induction,” defined as “how people generate degrees of confidence that a predicate [Goodman’s name for any conjectured test statement] applies to a stated category …”(111)

I find all these names identifying a single method: induction means counting empirical instances to confirm theoretical categories. Hume and Popper explicitly and legitimately condemn this method, which condemnation defines the logical problem—the assumption/belief that certain test statements are true, a conclusion NOT justified by repeated test statements as evidence.(7) Induction is not able to justify what it does.

Returning to Popper, he uses sunrise 4 times in OK as an example of the method and the problem. (Hume and S&L use same example). p.3 commonsense definition of method “repeated observations made in the past: we believe that the sun will rise tomorrow because …” This usage does not affirm Popper’s endorsement, but the fourth usage does, as example of psychological problem. p.26 assurance of truth of “best-tested theories, such as that the sun rises every day.” Sunrise has never been falsified. His second usage p.10 is explicitly falsified by universal theory/conjecture including time and location predicates. His third usage p.20 finds sunrise theory “preferred” because supported by background knowledge (71; Goodman called background “entrenchment by projectability” S&L p. 97). Here Popper endorses the very theory he rejects. How is it possible that knowledge of earth’s rotation does’t falsify all examples? Comments?TBR-qed (talk) 20:16, 11 May 2020 (UTC)

Two questions. First, in the sentence Here Popper endorses the very theory he rejects could you be more explicit about the theory that he endorses in p.20, but elsewhere rejects. It seems that this theory should be implicit in his third usage p.20, but I would like that you spell out explicitly what is this (general) theory that you think Popper inconsistently endorses in p 20. Second, in the sentence How is it possible that knowledge of earth’s rotation does[n]’t falsify all examples?, you speak of "examples", but examples of what? It's not only that I don't know the examples that you refer to. I also don't know what is exemplified. Dominic Mayers (talk) 01:01, 12 May 2020 (UTC)

p20. He finds "Sun rises [for earth-bound observer] in Rome every 24 hours," supported by Newton & 24 hour rotation. The experience is constant despite its false meaning. This statement differs from p10, which he interprets to mean 24 hours for every earth-bound observer in any location, an experience falsified at poles, ergo rejected. P10 logically false, p20 logically true. P3 theory is "sun will rise tomorrow because of" inductive observation interpreted as causation. P26 theory that sun rises every day" is "best tested" and must be pragmatically accepted, even though it cannot be eternally true. It is contradictory to accept a conjectural principle of transference p6 but to argue that the pragmatic problem of induction is independent of the logical problem BOTH EXEMPLIFIED BY A COMMON PATTERN OF BEHAVIOR--ENUMERATED SUNRISES. Imputing incompatible meanings to a common pattern of behavior is arbitrary metaphysics.TBR-qed (talk) 21:27, 12 May 2020 (UTC)

We have to put these four usages in their context. What he wrote in p.20 is not about induction. An argument that Popper uses in p.10 against inductivism can be turned against him in p. 20, that is, Newton + 24 hour rotation does not formally imply that the sun will rise every day, because it can explode, etc., but this would not affect much his main point in p. 20, which, anyway, is not about induction.

Similarly, p.10 is not about a method of induction proposed by Popper, but about a criticism of the method proposed by "inductivists": he basically says that even the standard examples of laws supposedly obtained by this so called method are not true universal laws. Inductivists might say that they did not claim that the method leads directly to universal laws, but only that the method exists and is used to obtain some laws. Popper discusses that also. In any case, p.10 is not at all about a method of induction proposed by Popper.

Similarly, in p.26, Popper only says that the "feeling of assurance" that inductivists claim comes from repetition is actually better explained in a different manner, more related to our pragmatic nature. He does not say it in Objective Knowledge, but elsewhere he says that it is the expression in our psychology of the natural mechanism of evolution. Popper had an organic view of science. He cites our assurance that the sun will rise every day in that context. He says that it is not the outcome of induction.

Let me add an important point in regard to p.26. One might think that the claim that no induction is used is equivalent to the claim that our experiences have no impact on our beliefs, but that would be a mistake. There are many ways other than induction through which experiences can affect our beliefs in some laws. One of them, but not the only one, is simply that our experiences can eliminate alternative theories such as the sun rises only twice a week. This is perhaps the point that interests us the most in our consideration of S&L, because it says that we must not conclude that a correlation between observations of premises and a resulting belief in a conclusion (that cannot be logically deduced from the premises) is necessarily the result of induction. This might be shocking for some people who see such a correlation as the very definition of induction, but this is only a problem of terminology. In the terminology where an inductive rule is analogous to a deductive rule in logic, we have to make sure that we can eliminate confounding factors such as some extra hidden knowledge, a correlation only is not sufficient.

Similarly, in p. 3, Popper clearly describes the standard method of induction without endorsing that it exists and is used. The whole thing is perfectly consistent. Dominic Mayers (talk) 01:32, 13 May 2020 (UTC)

I appreciate your questions, but can't accept these answers. Every reference to sunrise is an example of approval or rejection of an inductive theory of the cause of such motion. I can't follow you any farther into Popper's metaphysical third world. But I trust I have demonstrated that I am familiar with his work, and that discussing such a concrete example can promote useful discussion of that work and insight into the problem of induction by future encyclopedia readers.TBR-qed (talk) 19:40, 13 May 2020 (UTC)

My purpose was only to help you understand the literature, because a minimal understanding of the literature is necessary to edit this article. Of course, I do not want to patronize you. I sincerely thought you wanted my help. Dominic Mayers (talk) 20:30, 13 May 2020 (UTC)

Avoiding making implicit assumptions.

I just want to summarize what are my concerns. The article should not in the voice of Wikipedia implicitly conflate two notions of induction:

• any move from observations to a conclusion, in which the conclusion cannot be deductively obtained from the observations and
• the use of inductive inference rules to execute such a move.

I think the first notion is the standard concept of induction in philosophy or may be it depends of the author. For example, I think Goodman refers to rules of inferences when he explains how induction can be justified. The same terminology "induction" for totally different concepts might create some confusion. In addition, the article must distinguish between

• a process of induction with a clear causal relation from observations to a belief in a conclusion and
• a presumed process of induction where this causal relation is not so obvious.

For example, Popper argues that the causal relation between observations of sunrises and the belief in the law "the sun rises every day" is nonexistent. I am only saying that the article should not have Wikipedia implicitly adopt one of these two views, when there is no consensus about it in the literature.

However, I must say that most of the article is fine with regard to these concerns. There are the images on the left side, which incorrectly assume an inference from observations of sunrises to the corresponding law. There might be nothing else. I would have to read it again.

Dominic Mayers (talk) 21:57, 13 May 2020 (UTC)

Apology

I feel bad a bit, because we have cluttered this talk page with a long discussion that apparently was pointless. I hope that others such as Biogeographist have not been moved away by this. For what it's worth, I retract any support that I had in favor of TBR-qed plan, because I suspect that it is based on an incorrect interpretation of texts, which are taken out of context. However, my opinion about the value of S&L did not change. Dominic Mayers (talk) 20:30, 13 May 2020 (UTC)

No problem. In fact, I had stopped watching the page, so I didn't notice the discussion. I just looked at TBR-qed/sandbox, and I find it to be just as idiosyncratic as the suggestion to rewrite this article based on Sloman and Lagnado. In addition, TBR-qed/sandbox smells strongly of original research to me. What do you think of the article on the problem of induction at the Stanford Encyclopedia of Philosophy? I skimmed through it, and I think it would be a better model of what Wikipedia's article should be aiming for, although its style is not Wikipedia's style; the Wikipedia Manual of Style recommends against addressing the reader as much as SEP's article, which uses the word "we" 140 times. Biogeographist (talk) 22:31, 13 May 2020 (UTC)

I certainly enjoyed reading Henderson's article on the problem of Induction. He misinterprets Popper, but this is because most people don't realize that Popper was very close to Kant as far as a priori are concerned, especially in the years 1989 to 1994, the last years of his life. Of course, unlike Kant, he never accepted that individuals had a priori valid knowledge, but he replaced that by an apriori at the level the evolutionary process in nature. So the apriori of Kant was replaced by the acknowledgment that if the laws of nature were not supporting progress in science, then scientific knowledge would not grow. One might argue that this a priori was still a conjecture for Popper. Sure, but this conjecture plays the same role with respect to the problem of induction as Kant's a priori: if the problem of induction exists, for Poppper, the conjecture must be true, because, for Popper, if the conjecture was not true, there would be no growth of knowledge to explain and thus no problem of induction. Henderson should have classified Popper's solution in section 3.1 "Synthetic a priori", not in section 6 "Living with Inductive Skepticism".

In accepting this a priori, Popper was not so different than many other great scientists who are or have been (if they passed away) in bewildering admiration in front of the fact that the laws of nature are such that we can learn, we can be there and admire and understand the universe, more and more. The difference is that he also used that fact as a solution to the problem of induction. His argument, which is made clear in "All life if problem solving" and "The myth of the framework", is simply that whatever was there to support evolution before the rational aspect of the mind existed is still there and there is no reason why this evolutionary process should now require induction, a process at the level of the human mind. He starts by considering the "knowledge" in a dog and he goes back to unicellular life. At every level, he says, there are problems to solve and the evolutionary aspect of nature guesses solutions in a successful manner, with the need of trials and errors, but he sees no induction at all. This quality of the laws of nature is a big a priori. The fact that it is a big apriori, which we simply have to posit if science is to have any hope to succeed, is clearly mentioned by Popper in Objective Knowledge. Not only Henderson incorrectly classify Popper's solution, it also so obviously misses an important aspect of the solution. So much that it would be inadequate to give too much weight to his critic of Popper's solution.

But that's OK, the remainder of the article is very enlightening. His analysis of Hume's argument is very precise. He brings out the premises and the conclusions and consider how we can evade the argument by rejecting the premises. It's excellent. I am not opposed to this approach.

I would also keep the part on Pyrrhonism and Indian philosophy that we currently have in the article, even though the article confuses these rejections of induction with a discussion on the problem of induction. The problem of induction only arises when we also ask the question how knowledge grows, i.e., how induction as growth of knowledge can be explained. What is written about Pyrrhonism and Indian philosophy does not ask this question. It only rejects induction as a formal method of inference. Again, the confusion lies on the fact that we do not distinguish between induction as growth of knowledge and induction as a method of inference. These are completely different things. This confusion does not exist in Henderson's article for the simple reason that induction only refers to growth of knowledge and instead of using the term induction to refer to rules of inference, he speaks of arguments "having an empirical premise", which are the other arguments besides the arguments "deductively valid with a priori premises". The key point is because we assume a priori premises (such as Popper's assumption about the role of the evolutionary aspect of nature), we don't need induction ... oups, may be I should have written "we don't need arguments having an empirical premise".

Dominic Mayers (talk) 05:00, 14 May 2020 (UTC)

Notes

Is this article obsolete? (continued)

I’m glad Biogeo has returned to the discussion, so now I can ask both of you a question that has dogged me since beginning to edit WP. Please humor me, in the interest of improving the encyclopedia. As Biogeo gently put it, I have been judged “idiosyncratic” from the first, engaging in original research, and violating NPV. I reject these charges for the following reasons.

My first post revised the existing article on Instrumentalism. Its flaw was to ignore John Dewey who, for decades, was the philosopher spokesman for that “school.” Excluding Dewey completely violated NPV.

My revision described Dewey and Popper as representative supporter and critic of the school, followed by numerous social science examples of current debate on their positions. Critics reverted my post, arguing that I should just have added a reference to Dewey to the existing conventional article.

I found that impossible, because Dewey had “reconstructed” the debate. This required me to reframe the topic neutrally to include both positions. I find that same imperative in Problem of induction. I am not being original or violating NPV to try to define both the activity and the problem of induction in precise terms applicable to objective discussion of all positions taken. My sandbox draft is the current result.

I hope you will not dismiss my effort for being idiosyncratic. I hope you will critique my effort if it fails to achieve my purpose within WP parameters. That should be possible without entering into philosophical minutia, if I demonstrate familiarity with the representative positions. I think you might even enjoy seeing Popper confronted by a worthy adversary.TBR-qed (talk) 20:25, 14 May 2020 (UTC)

The reason why I don't reply is that I don't know what to do when an issue is not about the article or about (our understanding of) a source or an eventual source for the article. Dominic Mayers (talk) 16:43, 15 May 2020 (UTC)

@TBR-qed: I can appreciate that you don't like my judgment of your draft as idiosyncratic, but after looking at it again that's still my verdict.

Above, your response described (if I understand you correctly) an analogous "reconstruction" and "reframing" that you attempted in another article. I agree that you have done some reconstructing and reframing of the problem of induction in your draft, but it seems narrowly done, which is what I mean by idiosyncratic. If you were aiming in your draft for a more comprehensive and neutral perspective, as you claim above, and not merely a change of perspective to your own point of view, then you haven't yet succeeded. Your draft reads like your own original argument, and not a very clear one.

The following quotation is an example of a more comprehensive and neutral perspective, from the lead to the article on the problem of induction at the Stanford Encyclopedia of Philosophy: "A number of philosophers have attempted solutions to the problem, but a significant number have embraced his conclusion that it is insoluble. There is also a wide spectrum of opinion on the significance of the problem. Some have argued that Hume's argument does not establish any far-reaching skeptical conclusion, either because it was never intended to, or because the argument is in some way misformulated. Yet many have regarded it as one of the most profound philosophical challenges imaginable..."

That your draft is original research is clear enough from the lack of citations, although one could also point to other indicators of original research. To be clear, no original research means: don't make claims or arguments that are not attributable to a reliable source, and don't use statements from a source to reach a conclusion that is not in the source. You cite Sloman and Lagnado's article, of course, but I don't think Sloman and Lagnado's article is a good basis for radically reconstructing this article, as explained above. Sloman and Lagnado could be used for some details, but despite its title it is not appropriate as a model for massively reconstructing this article. When you look up Sloman and Lagnado in Google Scholar and look at the list of publications that cite it, you see that almost none of them are about the problem of induction! Scholars who have found Sloman and Lagnado's article useful enough to cite are not generally writing about the problem of induction. There is an interesting exception: Jackson, Alexander (June 2019). "How to solve Hume's problem of induction". Episteme. 16 (2): 157–174. doi:10.1017/epi.2017.32. {{cite journal}}: Invalid |ref=harv (help) Jackson's citation of Sloman and Lagnado's article is a good example of the minor role Sloman and Lagnado would have in a philosophical discussion of the problem of induction, although Jackson's article is not a good model for reconstructing this article either, since Jackson was making an original argument and was not attempting to follow Wikipedia's WP:NOR and WP:NPOV policies.

Perhaps a better strategy for adding the information that you want to add to this article would be: Find a highly cited source that contains the arguments that you think are important and unrepresented in the article, and then accurately describe that source in a new subsection in Problem of induction § Notable interpretations. Or two subsections, if there are different arguments in two different sources. Biogeographist (talk) 21:03, 15 May 2020 (UTC)

Thanks. Let me digest your remarks.TBR-qed (talk) 14:25, 16 May 2020 (UTC)

I sincerely thank Dominic Mayers & Biogeographist for answering my question about the obsolescence of the existing problem of induction article. Everything you wrote revealed some element of that problem.

DM (23 April) found the entire article biased because it assumes that this reasoning method starts with observation: an empirical claim. This claim ignores Popper’s thesis that reasoning starts with a theoretical conjecture about an observed-&-known phenomenon. The bias is revealed by the author’s treating repeatedly observed sunrise as an effect of unobserved solar motion, and by the Broad quote: "induction is the glory of science and the scandal of philosophy.” Popper replaced enumerative induction with falsification, making sunrises & Broad irrelevant.

I find the existing article unbiased but incomplete. It does not do justice to Popper, but correctly reflects the traditional interpretation that induction is illogical but sometimes successful in identifying natural kinds of phenomena—such as sunrises. I intend to restore the Broad quote, retain the sunrise photo, and expand coverage of Popper, Duhem, & Dewey.

Biog (15 May) makes 3 points. 1) Possible bias in existing article and my idiosyncratic definition of induction might be “merely a change of perspective to your own point of view.” 2) My draft “is original research … clear enough from the lack of citations,…”. 3) “Sloman and Lagnado's article is [not] a good basis for radically reconstructing this article.”

All fair and constructive observations. 1) is an example of the problem of induction: establishing that a single sample is, in fact, representative of a natural kind. My unfamiliar definition might be a unique idiosyncratic conjecture or a genuine continuing factor. Only subsequent observation will establish whether it fits each of the 8 scholars on whom I report. 2) Lede doesn’t need sources. It sets the stage for well-documented content to follow. 3) I find S&L discussion of kinds of induction extremely useful, but I use only 4 of their scholars as samples. I hope to have my revision ready for further critique by September.TBR-qed (talk) 14:08, 20 May 2020 (UTC)

Are you suggesting that any work to improve the article, perhaps reorganize it, should wait until after you have your revision ready for further critique by September? In addition, you stopped any effort to have a minimal required shared understanding when you wrote I can't follow you any farther into Popper's metaphysical third world. That's not the way it works in Wikipedia. There is no way editors can work together on an article without sharing a minimal understanding of the literature.

By the way, I did not even mention "third world" once in all my comments here. In fact, I never been much interested about what Popper wrote on this subject. I recently read about it because other editors mentioned it and what I found was not much. It is just the standard separation that many philosophers do. In Popper's view, the first world is the real world. From this first world emerges the world of mental activities, the second world. From this second world, emerges the objective knowledge that can be written in books, etc., the third world. It's not a big deal and it's not even necessary to use this terminology, because most of the times Popper does not use it. Instead, he refers to divisions as seen by other philosophers. For example, Hume made a distinction between the rational and the psychological in human thinking, which more or less corresponds to the distinction between the second and third worlds, but Popper uses Hume's terminology when he discusses Hume.

Any way, I think it was very unfair from you to say that you cannot discuss further with me, because [you] can't follow [me] any farther into Popper's metaphysical third world and then add that your revision [will be] ready for further critique by September.

Dominic Mayers (talk) 15:44, 20 May 2020 (UTC)

@TBR-qed: A technical detail. You said: Lede doesn't need sources. More precisely, statements in the lead don't need sources when they summarize content that appears later in the article with citations, but the lead often includes other statements that are not repeated again in the article: these statements would need sources. Biogeographist (talk) 22:07, 20 May 2020 (UTC)

Biogeographist, I do not understand. Do you mean that you encourage TBR-qed to work on his own on this article for three to four months, without any attempt from him to get a consensus on his ideas, because this is how I interpret his plan?

I am even more concerned when I see that the little he wrote above is full of misinterpretations. He wrote DM found the entire article biased because it assumes that this reasoning method starts with observation: an empirical claim. This claim ignores Popper’s thesis that reasoning starts with a theoretical conjecture about an observed-&-known phenomenon. I did not mention this. What I mentioned on April 23rd is that the article misses a key ingredient in Popper's solution, namely, his view that the problems in explaining the growth of scientific knowledge are part of the problems in the theory of evolution in general. I also mentioned a bias in the article, but finally, after looking at it, it turned out to be only Broad quote and the captions for the pictures, two points that TBR-qed disagree with, which is fine. But he also does not intend to discuss these points or any other aspects of his understanding further. This, on the other hand, is problematic.

My main concern was at a different level and I think that, if we could clarify this main concern, the other points will be easier to address. As we see clearly in the Henderson's article on the problem of induction that Biographist suggested, there is a fundamental distinction between (reasoning) arguments and the inductive processes that we try to justify using these arguments. One needs to understand this distinction to see that Sloman and Lagnado only study the inductive processes themselves and thus ignore the philosophical problem how to justify them using reasoning arguments, a point that was made by Biographist from the start.

This is related to a very important problem of terminology. Nowadays, as can also be seen in Henderson's article, we refer to these reasoning arguments as inductive and deductive. There is no confusion in Henderson's article. Nevertheless, this terminology issue can create a big confusion, because now we have very similar terminology for two different concepts: reasoning arguments and inductive processes. To make things worst, Sloman and Lagnado classify the inductive processes using rules, which they call induction rules, which make them look like rules of argumentation. I would not be surprised that there is an implicit claim in Sloman and Lagnado that there is no fundamental distinction between these rules, which only classify the observed induction processes, and the inductive rules of argumentation, also called inference rules. I see a very very big problem in this implicit claim, because these observed processes can not be under question, because they are observed, and we cannot also complains about a classification of these processes. In contradistinction, the inductive rules of inference are certainly under question and, in fact, it is nowadays widely accepted that they don't really exist.

I suspect that TBR-qed also rejects this fundamental distinction, but I am now asking TBR-qed about that. TBR-qed, when you wrote I don't understand "(ampliative) inductive rules." I think it involves your effort to distinguish between explaining processes and explaining rules, which are both distinct from observing processes, were you implicitly rejecting that there is a fundamental distinction between the so called "Induction Rules" in Sloman and Lagnado and what Hume calls arguments? This is very important, because this is definitively part of a minimal understanding that editors (except those that only look at grammar and things like that) must share before they can work on this article.

Dominic Mayers (talk) 01:11, 21 May 2020 (UTC)

@Dominic Mayers: Sorry for the terseness of my previous reply; no, I was not endorsing TBR-qed's plan. But if he wants to come back and present his revision after three months, he is free to do that, and we are free to reject it if we find good reason to do so (and if we are still around). In the meantime, editing of this article will continue as usual just like the rest of Wikipedia, of course.

You mentioned evolution a number of times in the preceding discussion and it occurred to me to mention that the general term for the study of the relation of knowledge and biological evolution is evolutionary epistemology, a term coined by Donald T. Campbell and then adopted by Popper and many other thinkers. Biogeographist (talk) 04:01, 21 May 2020 (UTC)

Yes, regarding TBR-qed's plan, I expected that you would say that he is free to do that. I do not disagree at all. In fact, I also had the same thought. However, I feel that I have to offer an alternative approach, which is to try to achieve a consensus on some minimal shared understanding.

Regarding evolutionary epistemology, Popper had this concept even before LSD was published. Here is a quote from Popper of a text originally written around 1952. This quote refers to LSD which was being translated at the time.

Sometimes I have also described it as the degree to which the theory in question ‘has been able to prove its fitness to survive by standing up to tests’. But as the context of such passages shows, I meant by this no more than a report about the past fitness of the theory to survive severe tests: like Darwin, I did not assume that something (whether an animal or a theory) that has shown its fitness to survive tests by surviving them has shown its fitness to survive all, or most, or any, future tests.

— Popper, Karl, Realism and the Aim of Science (p. 64).

But of course, the notion might have existed way before Popper. That's not important. What is important is who has seen the significance of this connection and used it. Popper understood that it was a fundamental ingredient in the solution to the problem of induction.

Dominic Mayers (talk) 10:40, 21 May 2020 (UTC)

Sure, there's no doubt that Popper was the major inspiration for Campbell's coining of the term evolutionary epistemology. I wouldn't know whether Popper was the first to connect Darwin and the problem of induction. I just searched on Google Scholar for Darwin + "problem of induction" before 1952, and discovered some fascinating stuff. I had heard of F. C. S. Schiller before, but I had never paid attention to his work. Consider this passage from his 1912 book Formal Logic, a Scientific and Social Problem (page 242, boldface added):

If in order to know anything we must first know that our principles are eternal a priori truths, and our facts absolute and immutable, it is painfully clear that knowledge is impossible, because we have not even now such facts and principles. But then we do not need them. All the equipment that we need to start upon the discovery of truth is a willingness to experiment and a willingness to learn. Granting these (and they are by no means common qualities), our experience will supply us with abundance of material. Indeed the chief difficulty will be to select the best and most workable from among the multitudinous suggestions and analogies with which the world bombards an actively inquiring mind. Fortunately there is abundant time for such selection. It has been going on for ages, and even the lowest organisms are to some extent selective in their reactions to stimulation. The selectiveness of man is enormous and all-pervading, and he is also conscious of it. Is it a wonder, then, that the results of this whole history should have crystallized into 'axioms' which now seem self-evident, and into 'facts' which now seem solid? Yet the logical value of the products which our ordinary thinking now takes for granted is not original but acquired. Nor are they, even now, immutable, or worthy of superstitious reverence; but they are secured against frivolous attacks.

Of course, merely to say "fortunately there is abundant time for such selection; it has been going on for ages" is not a fully developed evolutionary epistemology, but it seems clear enough that Schiller was thinking of evolution, given his abundant references to Darwin in this book—there is a section titled "Darwin v. formal logic"! And a section titled "The problem of induction". The basic ingredients are there, which is impressive for 1912. I searched for Schiller + "evolutionary epistemology" and noticed that Campbell mentioned Schiller in a paper published in 1989, "Fragments of the fragile history of psychological epistemology and theory of science" (page 28): "Finally, such thinkers as Dewey, Schiller, and Rignano have seemed to think that logical principles have some sort of psychological justification or else foundations in some kind of natural history."

The search for Darwin + "problem of induction" before 1952 also returned: Ritchie, A. D. (July 1926). "Induction and probability". Mind. 35 (139): 301–318. JSTOR 2249587. {{cite journal}}: Invalid |ref=harv (help) This article does not propose anything like an evolutionary epistemology (it just happens to mention Darwin) but the similarity to Popper in some other aspects is striking. In fact, looking at the list of publications that have cited Ritchie's paper (not many), one sees that Lakatos cited Ritchie and even combined Ritchie's paper and Popper's LSD into one term: "the Ritchie-Popper argument"! Very interesting—I did not know this. Biogeographist (talk) 14:13, 21 May 2020 (UTC)

TBR-qed, just letting you know that none of what follows make sense if one does not accept the fundamental distinction between (reasoning) arguments and observed scientific knowledge (i.e. the existence of universal laws in science, in particular, the fact that we can learn new laws). The philosophical problem of induction cannot be formulated without accepting this distinction. The problem is how arguments + observations can be used to obtain the laws, the current laws or new laws. A key point is that the arguments must be universal formal rules that stand alone such as A ⇒ B, A ⊧ B, not categories of observed processes, in which there can be some hidden informal knowledge. The universality of the arguments is fundamental.

Biogeographist, your comments are interesting ! But, of course, you cannot expect a book written in 1912 to get even close to Popper's view. I quickly went over the chapters, reading only some excerpts, and I think the most relevant is the section "Concessions to Psychologic" at page 392. It's the traditional "solution". People are forced to admit that there is no formal inductive argument, because they see that it cannot be valid, but then, because they assume that knowledge grows from observations within human reason, they conclude it must be in the "psychologic".

The mistake, if we accept Popper, is to assume that our connection with nature is only through observations that can be translated in observation statements. Popper's view is that the conjectures themselves must be "explained" as the result of evolution. So, in the same way as, obviously, the laws of nature created large molecules without the help of any human reason, then created life without the help of human reason, the same laws of nature have created in some form in our brain the ability to make useful conjectures. This is NOT psychological, not entirely anyway, because it happened in parts even before our psychology existed. If we had to formalize the process, unless we personify nature, that would not be the description of some reasoning process. Such formal process would somehow have to consider that the knowledge existed before it becomes human knowledge. For Popper, this is a tautology: no formal machine can know what is not known.

My way to explain this is simply that the problem of induction is an artificial problem that arises when we separate the universe in two parts: the part that tries to know and the part that contains what must be known. Given this artificial separation, we ask the question how the laws of nature in one part can discover the laws of nature in the other part using only a formal process (a formal and limited view of itself) and observations of what is going on in the other part? The answer that Hume and others correctly obtain is that the laws of nature cannot do that and we have the problem of induction. Popper's view, the way I see it, is that this separation is artificial, because the conjectures also, not only the observations, arise from natural processes. That is amazingly simple and natural. It does not break any special thesis. It certainly does not break the Church–Turing thesis, unless we consider that Nature itself (not working in two parts, where one part is restricted to be a known formal process and the other part is considered unkwown) breaks this thesis. It is only common sense. It's very similar to Kant's solution and Popper acknowledged that. The difference is that Popper says that at any given time the conjectures formulated in human brains are only guesses. They are apriori, but not valid apriori. The valid apriori of Kant are replaced by the unknown (but by definition correct) laws of nature that, luckily, allows us to make useful conjectures. Popper believes in Darwin theory and therefore said that trying to explain how it happened is difficult because it happened as a result of many accidents.

Dominic Mayers (talk) 15:54, 21 May 2020 (UTC)

Right, Schiller was writing in an era when "computers" were people. Popper in 1952 was writing in the same milieu that gave birth to artificial intelligence and later to generalized cognitive science. Still, it's interesting that in a chapter on the problem of induction Schiller clearly refers to an evolutionary epistemological process, even if he is not clearly generalizing it from individual organisms to nature in general, much less connecting it to computation. Biogeographist (talk) 17:01, 21 May 2020 (UTC)

That's it. Popper would agree. He benefited a lot from Godel's and Tarski's results in logic and foundations of computer science. Of course logic existed before. In particular, Hume considered two kind of arguments: “demonstrative” and “probable”. But the formalization of logic with a clear notion of semantic based on separate formal interpretations must have been very helpful for Popper. Perhaps even more important was the discovery of general relativity, which was just beginning in 1912. Schiller could not benefit from the fundamental philosophical consequences of this. Dominic Mayers (talk) 17:43, 21 May 2020 (UTC)

Reconsidering progressively the article

The revert of TBR-qed major edits should not be interpreted to mean that the article cannot be improved, of course. In my view, the article mostly contain excellent well sourced content. Something bothers me a little though. It seems to fail to distinguish between different descriptions of induction, which may or may not be equivalent and thus should not be considered a priori as one and the same concept of induction in Wikipedia. The challenge here is that not only different descriptions of induction seems to exist in the literature, but also different ways to distinguish between different descriptions seem to exist. It becomes easily a mess if one wants to consider all of this. On the bright side, I believe there is a classical notion of induction and it is not the notion of induction in Sloman & Lagnado, for example. I am not sure how we should manage this situation.

However, at the least, we should make sure that the article does not mess up different notions as if they were the same. I believe the article suffers from this kind of problems at few occasions. My main criteria is that the traditional notion of induction, even the probabilistic version, has clearly been rejected in philosophy, starting from ancient Greek and Indian philosophy and more recently with Hume. There has been many attempts to find a valid classical notion of induction, but no success. My favorite example is the entire work of Lakatos. He criticized Popper because he did not propose an inductive process and took for himself the task of finding inductive rules, but at the end he only obtained descriptive rules. He could not find rules that could be used by scientists and he acknowledged that. He argued that it was good enough, but many others philosophers noticed that it was not the original goal. For example, Feyerabend wrote in "Against Method" that Lakatos' methodology of scientific research programmes is epistemological anarchism in disguise and Musgrave made a similar comment. Anyway, if the article speaks of induction as inference rules that are used in practice and differ from deduction by the fact that they are uncertain or probabilistic, then a big red signal should be raised, because most likely if we look carefully at the source, we will see that it's not the classical notion of induction. Dominic Mayers (talk) 03:23, 16 November 2020 (UTC)

The recent big edit of TBR-qed

I am not sure this big edit is an improvement. The problem of induction is one of the most challenging, if not the most challenging, subject in the philosophy of science and I think, we need the contribution of other editors here. It's going to be a challenge, if we want to make major modifications to this article. I intend to revert the last edits of TBR-qed. Dominic Mayers (talk) 15:26, 15 November 2020 (UTC)

Mcc1789 (talk) on 18:12, 15 November 2020 (UTC) wrote It seems fine.

I am glad to have this opinion, but still think that a more substantial discussion is needed. Dominic Mayers (talk) 18:20, 15 November 2020 (UTC)

@Dominic Mayers II: Please revert the edits by TBR-qed. If you don't revert, I will. The edits by TBR-qed suffer from the same issues that we discussed above regarding the draft at User:TBR-qed/sandbox, etc. For example, it over-relies on Sloman & Lagnado. The first paragraph is insufficiently clear for Wikipedia. Other parts of the lead section are unencyclopedic in tone. Dominic noted other problems below. I appreciate TBR-qed's enthusiasm for this topic, but what TBR-qed is trying to do in these edits may be better suited to an original journal article, not to Wikipedia. I suggest further discussion here with TBR-qed about what he is trying to do, if he wishes. Biogeographist (talk) 19:18, 15 November 2020 (UTC)

Yes, I am looking forward toward useful discussions with TBR-qed. I don't feel his points are wrong and disconnected from the literature, not at all, but they need to be attributed and balanced with other POV. Dominic Mayers (talk) 19:48, 15 November 2020 (UTC)

And yes, in view of the controversial nature of the subject, these points might be too recent and better suited in an original journal article. Yet, I think some balance is needed. Editors should work together to include as much recent and established knowledge as possible, but in respect of WP:RSUW. Dominic Mayers (talk) 20:09, 15 November 2020 (UTC)

What method of reasoning?

The first sentence in the lead is

The problem of induction is that this method of human reasoning, unlike its mate deduction, mysteriously and paradoxically creates illusions of knowledge of universal patterns out of fragments of experience.

It assumes from the start that there is an actual method of reasoning called induction and do not attribute this view to any philosopher. It is as if the encyclopedia states this as a fact. All recent edits of TBR-qed are based on this bias toward one point of view. Given that this is at the center of most debates on induction, this is not acceptable in Wikipedia. Dominic Mayers (talk) 18:59, 15 November 2020 (UTC)

Yes, I mostly agree. I'm not totally sold on the current wording but it's significantly better than the previous one. GliderMaven (talk) 22:00, 15 November 2020 (UTC)

Given that we have Inductive reasoning it would not appear to require further sourcing. What concerns me about the sentence is 'mysteriously and paradoxically creates illusions of knowledge" which seems to me to be editorializing.Teishin (talk) 18:11, 16 November 2020 (UTC)

You are right. The issue is not that the method of inductive reasoning is not defined and attributed to an author. The issue is the indirect assertion that it is actually used in practice and "mysteriously and paradoxically creates ...". The current version is more neutral, more adequate for a lead. Though, I am also not totally sold on the current wording either. Dominic Mayers (talk) 20:26, 16 November 2020 (UTC)

Rebutting my reverters

Dom’s and Bio’s reasons for reverting me are so flimsy that I cannot take them seriously.

Dom starts by asserting—without authoritative support—that the problem of induction is so large in the philosophy of science, that a single editor or a pair of psychologists can’t revise it. There must be a multitude of editors who presumably will agree on what the problem is. This is precisely why the existing article is obsolete. It is futile to keep repeating old inconclusive arguments

He goes on to deny the problem exists, asserting that my post is biased because I assume “that there is an actual method of reasoning called induction …” Ignoring centuries of philosophical debate, and the 9 scholars I discuss in the body of my post (which he has not seen), he seems to endorse Popper’s denial that inductive reasoning exists in the formal logical sense of deductive rules. I do NOT assume the existence of that which I analyze under the label of “problem of induction” Having judged the problem and the method to be illusions, he denies I have a right to examine ancient and modern debates about it. He is the biased one.

He condemns my reliance on two psychologists whose recent publication studies philosophers AND psychologists who practice and debate what they call induction.

Bio condemns my effort to reframe traditional debates in the terms currently employed by practitioners, labeling my definitions “unencyclopedic in tone.”

I will not permit these two to forbid my use of the lede to introduce the vocabulary currently used by practitioners of induction, nor to dictate which professional publications may be permitted to argue the illusions and facts of that kind of rationality.

I am going to restore and complete my revision, bringing the problem of induction into the 21st century. This not an uncyclopedic or biased enterprise.TBR-qed (talk) 13:41, 16 November 2020 (UTC)

A 15k+ character revision of an article is almost certain to be contentious. In the case where such large revisions are needed, they should be done piecemeal, over an extended period of time.Teishin (talk) 14:39, 16 November 2020 (UTC)

@Teishin: your point is important. @TBR-qed: you should consider it. You should also apply a similar principle when you write points that expect an answer. Take into account that each single point that you make that expects an answer, even if the point is short, can require a long answer. Please focus on one point that you want to be answered, so that we can actually have a discussion. I am waiting. Dominic Mayers (talk) 16:26, 16 November 2020 (UTC)

I too agree with Teishin's comment. See also WP:Consensus. If TBR-qed does not want to do the recommended small, gradual edits, then an alternative is to do a more large-scale revision in User:TBR-qed/sandbox and ask for a review by other editors. Then other editors could comment on what, if any, of the sandbox draft would be acceptable for merging into the article. Given our objections to what has already been reverted, it's unlikely that the other editors here would accept a whole replacement of this article by one written entirely by TBR-qed.

Above, in May, I said: What do you think of the article on the problem of induction at the Stanford Encyclopedia of Philosophy? I skimmed through it, and I think it would be a better model of what Wikipedia's article should be aiming for, although its style is not Wikipedia's style. I would like to hear TBR-qed's opinion of the SEP article (setting aside its interpretation of Popper, to which Dominic objected), and an explanation of how and why TBR-qed's approach would be different from the SEP article's general approach. Biogeographist (talk) 23:31, 16 November 2020 (UTC)

Just to quickly say, regarding Popper, that it's not my personal view versus SEP. The SEP article presents Salmon's view on how Popper cannot answer the strong interpretation of the Duhem-Quine thesis (see Lakatos, The methodology of scientific research programmes). It says nothing about Popper's evolutionary epistemology and how it relates to this thesis. Not so surprisingly, because many editors contributed to it, the current article does a better job on this respect, though I think it can still be improved. Dominic Mayers (talk) 23:53, 16 November 2020 (UTC)

I didn't mean that it was Dominic's personal view, but he happened to be the one who voiced that view here. Interestingly, an earlier version of the SEP article (Fall 2011) mentions evolutionary epistemology briefly, and in fact the last section of the article is devoted to it. But that section disappeared in the subsequent version (Winter 2012), and then several more substantial changes were made to the article and eventually a new author wrote the current version. Biogeographist (talk) 02:36, 17 November 2020 (UTC)

I also would like to hear TBR-qed's opinion on the SEP article. I feel that it will help us to understand what he is trying to do. Dominic Mayers (talk) 15:50, 17 November 2020 (UTC)

Thanks for correcting my ignorance.

Your replies have finally penetrated my ignorance of how I can use a sandbox to present a proposal rather than making a drastic revision. I shall do that. I shall review and comment on the SEP article. But first, let me make 2 assertions to show where I am coming from--which is doubtless as controversial as my content:

An encyclopedia is a living tool, not a crypt or mausoleum. Would it not be more useful to users of Wikipedia to learn how Daniel Kahneman’s Thinking Fast and Slow perpetuates the problem of induction than to learn what Sextus Empiricus wrote about it?TBR-qed (talk) 17:31, 17 November 2020 (UTC)

Wikipedia serves a wide variety of users who have a wide variety of information needs. Our concern as editors is not so much about whether X is more useful than Y, but whether X is useful enough to include. Teishin (talk) 18:17, 17 November 2020 (UTC)

Teishin's point is essential. I would actually be interested in reading about how Daniel Kahneman's Thinking Fast and Slow perpetuates the problem of induction. But keep in mind that such an account would need to be based on at least one secondary source to be included in Wikipedia. We can't make original arguments about primary sources (WP:NOR), so if you're making original arguments about Kahneman and the other authors that you listed above at § Induction & deduction, Wikipedia is not the place to do it. WP:PSTS says: "Secondary or tertiary sources are needed to establish the topic's notability and to avoid novel interpretations of primary sources. All analyses and interpretive or synthetic claims about primary sources must be referenced to a secondary or tertiary source, and must not be an original analysis of the primary-source material by Wikipedia editors." Biogeographist (talk) 18:39, 17 November 2020 (UTC)

Your points are well taken. I am making progress on my sandbox proposal which you can judge. I report my reaction to the SEP article.

RANSON ON SEP “The Problem of Induction.” First published Wed Mar 21, 2018 by Leah Henderson

Henderson’s discussion of Baysian probability (section 3.2.2) accurately states the problem of induction: “We want to infer not what the sample will be like, with a known hypothesis, rather we want to infer a hypothesis about the general situation or population, based on the observation of a limited sample.” I find this completely compatible with my opening statement: “The problem of induction is that this method of human reasoning, unlike its mate deduction, mysteriously and paradoxically creates illusions of knowledge of universal patterns [hypotheses] out of fragments of experience [limited but representative sample].”

Unfortunately, her opening paragraph does not state the problem. Instead it gives an example of the method of enumerative induction as the prototype of inductive inference. She follows with a common sense inference “from the observed to the unobserved, or to general laws… as ‘inductive inferences,’” further enforcing the belief that induction means counting observed samples as evidence of unobserved kinds, denying that other methods can legitimately be called inductive. This denial ignores the existence of a well-established scientific alternative—a terrible oversight which Sloman and Lagnado and I correct.

I present Duhem and Dewey as practitioners of scientific induction, which neither Henderson nor S&L do. The closest she comes is to identify Reichenbach’s “pragmatic vindication” of induction. Why not Dewey’s? To exclude Dewey’s Logic: the Theory of Inquiry is rank bias.

Henderson does not name Hume’s method enumerative induction. She attaches names recognizable only by philosophers to empirical-evidence-joined-to-conceptual-generalizations. “For convenience, we will refer to this claim of similarity or resemblance between observed and unobserved regularities as the ‘Uniformity Principle (UP)’. Sometimes it is also called the ‘Resemblance Principle’, or the ‘Principle of Uniformity of Nature’. S&L use the name common among psychologists: “similarity-based induction.” Both accept without debate that observable similarity is an empirical universal attribute of every sample of inductive inference.

Henderson and S&L accept without question Goodman’s premise that the attribute green is attached to emeralds by enumeration, and might change at any time. All ignore its scientific definition which cannot change metaphysically. None of its attributes is observable. All are predictable once kind is defined scientifically. “Emeralds are formed when chromium, vanadium, and iron are present in the mineral beryl. The varying presence of these three elements gives emerald its range of color.” Seeing Green: All About Emeralds - GIA www.gia.edu › seeing-green

Henderson’s “meta-induction (section 7.3) seems to encompass S&L’s “categorical induction,” but she is unaware of the psychologists’ idea, much less its similarity to the philosopher’s idea.

My conclusion is that Henderson’s piece does more to conceal the nature of the problem of induction than to clarify and resolve it. — Preceding unsigned comment added by TBR-qed (talk • contribs) 18:32, 19 November 2020 (UTC)

TBR-qed: Thanks for your commentary on the SEP article. A quick response to one point you made: You said, comparing Reichenbach and Dewey: To exclude Dewey's Logic: the Theory of Inquiry is rank bias. I'm not sure exactly what you mean by "rank bias", so I may just be restating your point, but I think Henderson's inclusion of Reichenbach and not Dewey reflects the fact that Reichenbach's book has been much more commonly discussed in scholarly accounts of the problem of induction than Dewey's book has been. This can be seen quickly, for example, on Google Scholar by searching for "problem of induction" in publications that cite Reichenbach's and Dewey's books; Reichenbach comes out far ahead of Dewey. Nevertheless, I will be interested to see what you say about Dewey and which secondary sources about him and the problem of induction you are summarizing. Biogeographist (talk) 19:43, 19 November 2020 (UTC)

reforming writing style

I am trying to relate secondary sources to primary sources to be more encyclopaedic. I hope you will help me correct my lapses rather than just counting them.TBR-qed (talk) 17:15, 22 November 2020 (UTC)

Proposed revision in User:TBR-qed/sandbox

@TBR-qed: I just looked at this latest revision of your proposed changes to Problem of induction. I understand that it may not be finished yet, but overall there is a major problem with your method: You are just commenting on primary sources. This is the same method that you used to rewrite two previous articles, Instrumental and value rationality and Instrumental and value-rational action, and your rewritten versions of those articles have been appropriately tagged by Omnipaedista with the {{Essay-like}} cleanup tag. Your method is not an appropriate method for Wikipedia, as it produces an original argumentative essay that violates the WP:NOR policy. If you want to produce content that is appropriate for inclusion in this article, you will need to adopt a different method: Instead of commenting on quotations from primary sources, find secondary sources about how those primary sources relate to the problem of induction, and then summarize the secondary sources. One can cite primary sources, but the content should be based mostly on secondary sources and should not make original arguments. This has already been explained above, and has probably been explained to you on other talk pages as well. Thanks, Biogeographist (talk) 15:02, 21 November 2020 (UTC)

I second what Biogeographist said. TBR-qed's method of writing articles is unacceptable for Wikipedia. --Omnipaedista (talk) 15:34, 21 November 2020 (UTC)

These comments finally make me recognise how you distinguish an essay from an encyclopedia article, and I grant some of my earlier posts violated this position. I think the present proposal is minimally guilty of the charge, and I ask you to suspend judgment until you see the final product. Over 1/2 is directly reporting on Sloman and Lagnado.TBR-qed (talk) 15:43, 21 November 2020 (UTC)

Just to say that the very structures of both the present article and the proposed revision are somewhat inappropriate. This type of structure might be problematic even if secondary sources have been used. Wikipedia articles about major concepts in analytic philosophy are rarely mere anthologies of summaries of philosophical works. Wikipedians tend to avoid this style of essay-like article structure. Philosophy articles tend to be organized around major themes, not organized around a collection of opinions of famous commentators/philosophers (see SEP's take on the problem of induction or the Wikipedia article about the new riddle of induction). --Omnipaedista (talk) 19:01, 21 November 2020 (UTC)

@Omnipaedista: One of my previous responses above, commenting on an earlier draft in User:TBR-qed/sandbox, was similar to your most recent comment, and I too pointed to the SEP article. A problem with your comparison to Goodman's new riddle of induction is that the problem of induction is much older, and so an adequate treatment of the subject has to cover more historical material than the article on Goodman's new riddle does; this fact can be seen even in the SEP article. And if you look at the history of the SEP article (some previous versions are mentioned above), it has gone through radical changes over the years, and has had very different approaches to presenting the topic. There are so many different views that it is a challenging subject to summarize, and I don't think one can entirely avoid organizing the subject by different views to some extent, as even the SEP article does in its second half; as the current SEP article says: "A number of philosophers have attempted solutions to the problem, but a significant number have embraced [Hume's] conclusion that it is insoluble. There is also a wide spectrum of opinion on the significance of the problem. ... In this article, we will first examine Hume’s own argument, provide a reconstruction of it, and then survey different responses to the problem which it poses." Having said all that, you are right that Problem of induction § Formulation of the problem is very underdeveloped. To complicate things even further, TBR-qed's approach is largely outside the area of what you called analytic philosophy. TBR-qed relies especially on a chapter by Sloman & Lagnado in The Cambridge Handbook of Thinking and Reasoning that presents psychological research and happens to be titled "The problem of induction" but has little relation to how the problem of induction is treated in analytic philosophy, and as I said above, I have found no evidence that Sloman & Lagnado deserve the prominence in this article that TBR-qed wants to give them. Biogeographist (talk) 19:52, 21 November 2020 (UTC)

Some issues that jump out to me are editorializing in the lede "mysteriously and paradoxically creates illusions of knowledge" and it barely says anything about the early history of the problem of induction. For those interested in ancient philosophy (which, it is important to note, is not dead philosophy. There are modern Stoics, Epicureans, and Pyrrhonists, for example) it is an important subject. Teishin (talk) 22:11, 21 November 2020 (UTC)

I disagree with your judgment that WP should do more than recognise ancient philosophy. The ancients are easily accessible; modern practices with ancient roots are not--a lack suited for WP to fill.TBR-qed (talk) 17:20, 22 November 2020 (UTC)

The sentences introducing the inductivist view and Popper's view

First, I consider the following sentence in the article:

The problem calls into question the traditional inductivist account of all empirical claims made in everyday life or through the scientific method, and, for that reason, C. D. Broad once said that "induction is the glory of science and the scandal of philosophy".

This sentence plays a useful purpose, because indeed there is notable belief, justified or not, that knowledge grows through some kind of reasoning, thus inductive reasoning, because deductive reasoning is not ampliative. The first problem is that I don't feel it's written in a neutral manner. It's not attributed in a clear way. It's not attributed to Broad (except the quote of course), because on the contrary Broad is presented as an innocent witness of the fact. It is not clearly attributed to the traditional inductivist view, because the term "account" again suggests that it is an observed fact. The second problem is that it refers to all empirical claims, but there is no problem with claims of simple observations. It's a nice sentence, but it needs some work. To address the first problem I suggest the following

The traditional inductivist view is that all empirical claims, either made in everyday life or through the scientific method, can be justified through some form of reasoning. The problem is that many philosophers tried to find such a justification but their proposals were not accepted by others. Identifying the inductivist view as the scientific view, C. D. Broad once said that "induction is the glory of science and the scandal of philosophy".

The middle sentence is not attributed, but that's ok, because it is well known. To address the second problem, "empirical claims" must be replaced by something else. I propose "claimed empirical laws". So, it becomes

The traditional inductivist view is that all claimed empirical laws, either in everyday life or through the scientific method, can be justified through some form of reasoning. The problem is that many philosophers tried to find such a justification but their proposals were not accepted by others. Identifying the inductivist view as the scientific view, C. D. Broad once said that "induction is the glory of science and the scandal of philosophy".

Second, let see how this fits with the next sentence

In contrast, Karl Popper's critical rationalism claimed that induction is never used in science and proposed instead that science is based on the procedure of conjecturing hypotheses, deductively calculating consequences, and then empirically attempting to falsify them.

It could be simply:

In contrast, Karl Popper's critical rationalism claimed that inductive justifications are never used in science and proposed instead that science is based on the procedure of conjecturing hypotheses, deductively calculating consequences, and then empirically attempting to falsify them.

I addressed a problem of terminology here by replacing "induction" with "inductive justifications". Of course, Popper never claimed that the inductive relationships usually seen between laws and observations do not exist. If asked why do you accept that the sun raises on the east every day, he would have replied that he sees it every day, just like anyone else. Note that this is no justification at all for the inference of the law. If one is asked how can you infer B from A and one replies because I have A, one has provided no justification at all for the inference. In other words, he believed in the value of corroborations, just like every one else. The only thing he rejected is the existence of a rational justification for the laws.

Dominic Mayers (talk) 17:25, 9 December 2020 (UTC)

This change looks good to me. Biogeographist (talk) 20:18, 9 December 2020 (UTC)

The introductory sentence in the lead

I consider the following sentence:

The problem of induction is the philosophical question of whether inductive reasoning leads to knowledge understood in the classic philosophical sense, highlighting the apparent lack of justification for:

The problem I see with this sentence is that it refers to inductive reasoning which can refer to inductive justifications or to what is now seen as inductive methods in social science and in artificial intelligence or even simply to any relationship between a law and observations that corroborate it. These concepts are totally different and operate at completely different levels. The problem of induction is about missing justifications. For example, this is clearly explained in SEP. The first paragraph in SEP describes the well known relationship between laws and observations: if a kind of bread has nourished thus far, we are not surprised to see the (false) law that it will keep be nourishing. However, when it comes to the problem of induction, the expressions used in SEP are "on what grounds" and "arguments that do not serve". So, clearly, the problem is the missing justifications. Therefore, I suggest that we do not refer to the nowadays vague term "inductive reasoning" and focus on missing justifications. I propose

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 that goes beyond a mere collection of observations, highlighting the apparent lack of justification in particular for:

Dominic Mayers (talk) 18:26, 9 December 2020 (UTC)

I understand your reason for not wanting to use the word "induction". But I would keep the link to Inductive reasoning in there. It could be helpful for people who end up at this page but don't already know what the word "induction" means in "problem of induction". One possible solution: You could link it like this: "mere collection of observations". Biogeographist (talk) 20:18, 9 December 2020 (UTC)

Looks great! I changed the proposal above accordingly. Dominic Mayers (talk) 20:21, 9 December 2020 (UTC)

References

Proposed replacement submitted for review

I have submitted my sandbox proposal. I feel it demonstrates that the existing article is obsolete and that contaminated inductive reasoning exists at pandemic levels. Critique invited.TBR-qed (talk) 17:02, 5 December 2020 (UTC)

It seems fine to me, but my expertise in only in the ancient/early modern history part. It might be nice to add a mention of Taleb's work here, as I suspect if we don't someone else try to add it. Teishin (talk) 18:07, 5 December 2020 (UTC)

I have not read the draft yet and won't have time to read it today, but I would not be as quick as Teishin to declare that it "looks good". Note that it was submitted to AFC and was quickly declined with the explanation: "This submission reads more like an essay than an encyclopedia article." I would say that even TBR-qed's summary of it above is essay-like: "I feel it demonstrates that the existing article is obsolete and that contaminated inductive reasoning exists at pandemic levels." That description seems to indicate that TBR-qed is arguing for a thesis, which is not what Wikipedia articles do. I expect I will have more to say after I read the draft, although it has already been declined at AFC, which I already predicted above when I said that it's unlikely that the other editors here would accept a whole replacement of this article by one written entirely by TBR-qed. The better approach is to consider whether any parts of it are suitable for merging into the existing article. Biogeographist (talk) 23:09, 5 December 2020 (UTC)

I was unaware that it was submitted to Articles for Inclusion. Here it is there: https://en.wikipedia.org/wiki/Draft:Problem_of_induction . Such a submission seems to me to be inappropriate, as we already have such an article. It seems to me that the reviewer should have rejected it on that basis. While I agree that prior versions of the proposed article had many problems, and its author clearly started it based on opinion motivation, this latest version seems much improved. Further, it does not seem to me to be a "whole replacement" but it instead appears to incorporate large parts of the existing article. Teishin (talk) 23:50, 5 December 2020 (UTC)

I haven't read the draft yet, but I saw in its edit history that TBR-qed was the only contributor of content to the draft except for minor technical edits to inline citations, and there is no indication in the edit summaries that he copied material from any other Wikipedia article (indication of the source of within-wiki copied content in edit summaries is required per WP:COPYWITHIN), so it is reasonable to assume at this point in time that TBR-qed is the sole author of it, and a "whole replacement" it is. Biogeographist (talk) 01:25, 6 December 2020 (UTC)

Thank you for continuing to reason inductively with me about induction. If you answer the following question, I will understand how large our agreement might be. Do you agree that my coverage table columns 1&2 is empirical evidence that my classification of existing article as obsolete and biased is warranted, and column 3 is empirical evidence that classifying my proposal as biased and original research is not warranted?TBR-qed (talk) 17:22, 9 December 2020 (UTC)

I don't like the word "obsolete" here: The problem of induction may be obsolete, but we still have an article about it, and it is helpful to describe the history of the problem, even if some of the history is "obsolete". "Incomplete" seems to be a better word: I think we all agree that the article is incomplete. I only think your draft is biased if you are proposing it as a whole replacement of the current article. Biogeographist (talk) 20:18, 9 December 2020 (UTC)

I share this view, but maybe for different reasons. Many many people have written about the problem of induction with different perspectives, different terminologies, etc. often promoting their work as being very different. The article will become a mess if we start to cover every single new perspective on the basis of neutral point of view. Before neutral point of view there is WP:NOTEVERYTHING. What I agree with is that it can be useful to mention the notion of induction as seen in social science and artificial intelligence in the article. This recent view is simply a move ahead in modern research that ignores the problem (not making it obsolete at all) and

• classify the different relationships between laws and observations, which relationships are recognized by Popper and others, or
• consider non ampliative "inductive methods" to generates laws. For example, in inductive logic the problem of induction is not considered at all. Hume is not even cited. One sentence in this article clearly indicates to me that it's not ampliative, but is a set of results that can be proven deductively given hypotheses:

  The [criterion of adequacy] stated here may strike some readers as surprisingly strong. Given a specific logic of evidential support, how might it be shown to satisfy such a condition? Section 4 will show precisely how this condition is satisfied by the logic of evidential support articulated in Sections 1 through 3 of this article.

The challenge to include this kind of material in the article is that we cannot make a link with the problem of induction that is not made in the literature and we should not also accept a link that is made quickly by a researcher that is not an expert in philosophy of science, because it will not be pertinent. In particular, this article in inductive logic is not pertinent, because it makes no reference at all to the problem of induction. Moreover, if there is an article by some experts that makes the link with the problem of induction, it is not sufficient. There must be a few articles so that we can be convinced that it represents the view of at least a significant minority of experts. Dominic Mayers (talk) 21:40, 9 December 2020 (UTC)