Talk:Falsifiability

From Wikipedia, the free encyclopedia
Jump to navigation Jump to search
Former featured article candidateFalsifiability is a former featured article candidate. Please view the links under Article milestones below to see why the nomination failed. For older candidates, please check the archive.
Article milestones
DateProcessResult
April 30, 2007Featured article candidateNot promoted


A basic statement in industrial melanism.[edit]

The purpose of every example in the section is to illustrate the process of demarcation. This process requires a basic statement. The basic statement is given and only then the law that is falsified by the basic statement is given. In the case of industrial melanism, the basis statement is

In this industrial area, the relative fitness of the white-bodied peppered moth is high.

Note that as an observation statement, it must refer to a specific industrial area. A possible improvement would be to use a known industrial area instead of "this industrial area" to make it more clear that it is an observation statement.

If the reader found the basic statement complicated, it is perhaps because he expected an explanation of the law. Or perhaps, the reader completely failed to understand that a basic statement is usually false, because it must contradict a true law. Of course, if one expects some explanation of a law and is given a simple basic statement (which contradicts the law), it can be confusing, as if something was missing or even wrong. But then the fault is not with the simple basic statement.

The fault is likely to be with the lead and how falsifiability is badly introduced in the article. Many people expect the common view on falsifiability when they read the subject title "falsifiability". Unfortunately, how falsifiability is introduced in the article confirm them in their misunderstanding. The common view on falsifiability is simply that the law can be falsified by (true) evidence. It is not a wrong view, because indeed is is believed that all laws, even the law of general relativity is going to be falsified by some future concrete evidence, just as the law of classical mechanics has been falsified. It is not a wrong criterion, but it is not at all this notion of falsifiability that is the subject of this article. Wikipedia is not a dictionary and an article does not have to cover all meanings of a term.

The meaning of falsifiability in this article, when applied to a law, is "the law can be used to make predictions". But falsifiability applies also to other statements, which are not laws. The link between the general definition of falsifiability and its more specific meaning in the case of laws is not obvious. In fact, some authors claimed that "can make predictions" was a better criterion than falsifiability, without realizing that Popper had already explained that it was the same definition when applied to laws. Some people might expect that the examples only illustrate that the laws can make predictions and therefore avoid the use of basic statements. But this would be incomplete and missing an important point, because the notion of basic statement is necessary to give the basic and general concept of falsifiability. There is no way the concept of falsifiability can be understood without it and this entire section does a good job in illustrating the concept of basic statement. It is so sad that some people try to make the article fit with their wrong or limited understanding, which please them, instead of trying to improve the article so that the correct understanding is conveyed. 173.206.25.82 (talk) 17:37, 24 September 2020 (UTC)

173, a simple question - How would you falsify the statement that "this moth is more fit for this environment?" The answer is, you can't, because you don't always know what makes an animal "fit", and it changes from situation to situation. On the other hand "in a soot-polluted areas, you will find more dark colored moths than light colored moths" is a basic falsifiable statement. It doesn't rely on someone's subjective opinion on what animals are fit and what animals are not. You simply measure the pollution and count the moths, and if you don't find dark-colored moths in polluted areas, you know it's false. And, of course, that statement is far easier to understand for everyone who reads are articles, since it does not depend on abstract concepts and doesn't require a background in evolutionary biology. Remember, Wikipedia is not a scientific text, our audience is the non-scientist public. Oiyarbepsy (talk) 18:11, 24 September 2020 (UTC)
You wrote You simply measure the pollution and count the moths, and if you don't find dark-colored moths in polluted areas, you know it's false. Simply report this observation and you have your basic statement. A report could be "In this polluted area, there was many white-colored moths and no dark-colored moth." We have the exact same notion. The only distinction is the emphasis on the basic statement. This emphasis is justified because it is central in the general concept of falsifiability.
To give some context, note that the basic statement itself is falsifiable. It is contradicted by the opposite basic statement "In this (same) polluted area, there was many black-colored moths and no white-colored moth." Yet, none of these basic statements can be used to make (new) predictions. They are not universal laws. The "basic" in basic statement refers to this fact. More precisely, it refers to the fact that they are singular existential statements. The "singular" means that it is about a specific observation in a specific area at a specific time. You also wrote On the other hand 'in a soot-polluted areas, you will find more dark colored moths than light colored moths' is a basic falsifiable statement. This statement is indeed falsifiable just as the basic statements are, but being a law, i.e., a pure universal statement (not specific to any area or time), it is not what we call a basic statement.
The fact that a falsifier must be a basic statement (i.e., a singular existential statement) is essential. Otherwise, the pure existential statement "there were some polluted areas somewhere at some time, in which the relative fitness of the white-bodied peppered moth was high" would be falsifiable, because it is contradicted by the falsifiable (but non-basic) statement "in [every] soot-polluted areas, you will find more dark colored moths than light colored moths". The above pure existential statement, just like any other existential statement, is not falsifiable, because whatever way you try to contradict it, a believer can reply that you have not looked every where or that you looked at the wrong time. It is a fundamental part of falsifiability that non-singular or pure existential statements are not falsifiable. Anyway, the point is that basic statements, because of their role as falsifiers to show falsifiability, cannot be pure universal statements as in your example. They must be singular existential statements. Popper wrote many sections in LScD to explain this. It is an essential part of falsifiability.
I sympathize with those who feel, as you seem to feel also, that it is more natural to explain the criterion directly in terms of predictions: a law is falsifiable if it makes predictions that can be false just as in your example. It's more natural, because it is implicit in the concept of prediction that both the initial condition and the prediction can be measured with some available technologies and also because the prediction is one possibility among many. If the prediction was the only possibility, it would not really be a prediction. There is a lot that is hidden, but yet natural and implicit in the notion of a falsifiable prediction. The concept of basic statement is necessary to make these hidden notions explicit, just like I did with your example with the help of your own argument. It is important that we make these notions explicit, because otherwise we get a confusion with the common notion of falsifiability, which I described above. Popper complained so much about this confusion in the literature. The only way to avoid this so common confusion is the concept of basic statements. Besides, the concept of basic statements is also required to obtain the general definition that does not refer to predictions and is thus not restricted to laws.
173.206.25.82 (talk) 19:01, 24 September 2020 (UTC)

Sources with a synthesis of the common and technical definitions[edit]

Once the language for the basics statememts is agreed upon, the falsifiability criterion is entirely at the logical level: it makes no reference at all to experimental tests.

There has never been a consensus about the definition of falsifiability to use in the lead. Many would prefer the common definition, which refers to actual observations in empirical tests, while adding a qualifier such as "in principle", "eventually", etc. or use expressions such "has the capacity", "has the potential", etc. to account for the fact that these observations might not be possible. This common definition conflicts with the technical definition, which insists that the falsifier is only a basic statement, also called a test statement or an observation statement. The distinction is important, because in the technical definition, there is no direct reference to observations. It is sufficient that a contradicting basic statement exists in the same language that is used to state the law. So, technical falsifiability requires, first, that the language to state the law is agreed upon and, second, that there exists a falsifier of the law in this language. In this way, falsifiability is defined in a purely logical manner, at the level of statements and their logical relationship only, no reference at all to actual observations to evaluate the truth of potential falsifiers.

It is a very serious issue in the literature on falsifiability, because the common definition has been interpreted in a completely different manner than the technical definition by many philosophers, including illustrious philosophers such as Imre Lakatos and John W. N. Watkins. The difference is that they consider that the falsification that exists in principle, i.e., the potential falsifier, must not only be possible in the formal language of the theory, but must also be provably a falsifier in the theory that underlies the observations. For example, to my knowledge, Lakatos never stated in his work the technical definition of falsifiability. He certainly did not in "The methodology of scientific research programmes." In this book, he went instead into an analysis of different levels of falsificationism, namely, dogmatic falsificationism, naive falsificationism and sophisticated falsificationism, raising different issues with the common view of falsifiability at each level. Similarly, Watkins is very explicit that Popper's basic statements as potential falsifiers are "absurdly overstrong", but yet cannot actually falsify a theory. He wrote (emphasis mine):

Popper undertook the impossible task of upholding the falsifiability of just this theoretical core. Apparently forgetting that he had once said 'Duhem is right when he says that we can test only huge and complex theoretical systems', Popper set out to devise potential falsifiers just for Newton's fundamental assumptions. [Popper's examples] are absurdly overstrong considered as [potential falsifiers] of a properly fleshed out version of [Newton Mechanics] but, of course, not strong enough to be [potential falsifiers] of this irrefutable core. They mostly involve planets moving in highly erratic ways. But Newton's laws of motion plus his law of gravitation say nothing about the physical makeup of the planets; in particular they do not rule out the possibility that the planets are enormous rocketlike devices (see O'Hear, 1980, p. 102) that can accelerate themselves in all sorts of ways. Another example has apples that have fallen from a tree rising up and dancing round the tree. But Newton's laws of motion and law of gravitation say nothing against the possibility of such a spectacle being produced with the help of invisible elastic threads.

— Watkins, Science and Scepticism

But Popper totally agrees with the fact that, if we start to consider how the actual observations that are described by basic statements provably or demonstrably falsify the theory, Newton's theory can never be actually falsified in practice by any observation. As pointed out by Thornton, Popper's logic is utterly simple: If an apple does not fall under gravitation, then it cannot be that all objects fall under gravitation. A potential falsifier and its associated contradiction in Popper's falsifiability remain entirely at the logical level. The problems of proving an actual falsification are irrelevant in falsifiability. Thornton's precise statement is

Popper has always drawn a clear distinction between the logic of falsifiability and its applied methodology. The logic of his theory is utterly simple: if a single ferrous metal is unaffected by a magnetic field it cannot be the case that all ferrous metals are affected by magnetic fields.

— Thornton, Karl Popper, The Stanford Encyclopedia of Philosophy (Winter 2019 Edition)

Not only Thornton, but also Chalmers and others understood that falsifiability is indeed on purpose defined in terms of what Watkins qualifies of "absurdly overstrong" potential falsifiers, at a purely logical level. For example, to show the falsifiability of the law "Heavy objects such as a brick when released near the surface of the earth fall straight downwards if not impeded", Chalmers gave the following "absurdly overstrong" potential falsifier:

The brick fell upwards when released.

— Chalmers, What is this thing called science?

Note how this potential falsifier logically contradicts the law in a very simple manner. This cannot be denied. The fact that perhaps an invisible elastic thread was attached to the brick, as Watkins would point out, does not change the fact at the formal level, simply looking at the logical relation between the two sentences, the law is contradicted. So, clearly when Watkins says that these "absurdly strong" statements are "not strong enough to be [potential falsifiers] of this irrefutable core [of Newton's gravitational law]", he refers to problems in an actual falsification. It might be worth to quote Kuhn on this respect (emphasis mine):

To be scien­tific a theory need be falsifiable only by an observation statement not by actual observation. The relation between statements, unlike that between a statement and an observation, could be the conclusive disproof familiar from logic and mathematics.

— Thomas Kuhn, Logic of Discovery or Psychologyof Research?, Criticism and the growth of knowledge, p.14

It is even a part of popular culture. For example, it is well known that to show the falsifiability of the theory of evolution, J.B.S. Haldane gave the "absurdly overstrong" potential falsifier: "fossil rabbits in the Precambrian". Again, there could be plenty of ad hoc explanations for the actual discovery of a fossil rabbit in the precambrian, but it still remains that it is a valid potential falsifier that was clearly accepted to show the falsifiability of the theory of evolution, because the simple logical contradiction with the law is there.

The current article does a very good job at dealing with this confusion. The point of view of Lakatos is fully presented, especially the problems of falsification as seen in the different levels of falsificationism. No controversy is hidden, except one: the controversy about the different definitions, but that is because it is not sufficiently notable. There exist no synthesis that discusses explicitly the two definitions as being opposed. Instead, what we have are excellent sources that very clearly and unambiguously present one of the two definitions, but none present a synthesis that is explicit about the confusion between these two definitions. I am OK with this. Delving with confusion around definitions is not interesting and it is not useful for the readers. Popper, of course, complained a lot about this confusion. For example, he wrote about 10 pages in his 1982 Introduction to Realism and the Aim of Science discussing in details this confusion and insisting that falsifiability is purlely at the logical level. However, this is a primary source.

On the contrary, instead of discussing this confusion between two distinct definitions, some excellent authors have proposed a synthesis of the two formulations as if they correspond to a same definition, while making clear that the unique definition is the popperian definition given in terms of basic statements, no direct reference to observations. For example, Thornton wrote:

Accordingly, [Popper] held that, from a logical perspective, a system of theories is scientific only if it is refutable or falsifiable:

I shall not require of a scientific system that it shall be capable of being singled out, once and for all, in a positive sense; but I shall require that its logical form shall be such that can be singled out, by means of empirical tests, in a negative sense: it must be possible for an empirical scientific system to be refuted by experience.

— Popper, Conjectures and Refutations

This demarcation criterion was most clearly defined by Popper in terms of the relation that holds between a scientific theory and 'basic statements', where the latter are to be understood as singular existential observation-reports of the form 'There is an X at Y'. On this definition, where a theory is scientific, it must exhaustively divide the class of basic statements into two non-empty subclasses.

— Thornton, Popper, Basic Statements and the Quine-Duhem Thesis, Yearbook of the Irish Philosophical Society 9 (2007), p.4

A synthesis that unifies two apparently opposite formulations is even more clear in Nickles' view:

Popper proposed falsifiability (empirical refutability) as the criterion of demarcation. For him a statement is scientific if and only if it is falsifiable in principle, that is, if it can fail an empirical test. This is equivalent to saying that there must be some possible observation statement (true or false) that logically contradicts the claim in question.

— Nickles, Problem of demarcation, The Philosophy of Science: an Encyclopedia, Sarkar & Pfeifer, p. 191

Note that the second formulation is only in terms of observation statements. Note also that despite a mention of "an empirical test" in the first proposed formulation, it is clear in Nickles' view that Popper's falsifiability is disconnected from the "body of belief-worthy claims about the world, let alone demonstrably true claims". He wrote:

Reflecting on the steady weakening of proposed criteria of demarcation, Laudan (1983) concludes that demarcation is no longer an important philosophical problem. Popper's falsifiability criterion, he says, weakens the criterion almost beyond recognition. No longer does the criterion of demarcation mark out a body of belief-worthy claims about the world, let alone demonstrably true claims.

— Nickles, Problem of demarcation, The Philosophy of Science: an Encyclopedia, Sarkar & Pfeifer, p. 194

— Dominic Mayers (talk) 19:29, 29 October 2020 (UTC)

The following statement from WP:OR is important in the situation that is described above: Rewriting source material in your own words, while substantially retaining the meaning of the references, is not considered to be original research. Doing that is necessary here because the technical definition looks weird at first and the common definition is misleading. The important is to retain the meaning of the references. The technical definition briefly stated is "The statement is falsifiable if it is contradicted by an observation statement." The problem is that "an observation statement" seems a weird way of saying "a possible observation". The issue is that the reader has no clue that in the concept of falsifiability, the contradiction must only exist in some language, which abstracts away many experimental issues. When one uses "a possible observation", it can be interpreted and has been interpreted by Watkins, Lakatos and many others to mean that the contradiction must also take into account all possible experimental issues such as invisible elastic threads. The only way to remove the weirdness while retaining the meaning of the references is to make sure that this purely logical context is understood. So, what is needed is one or two natural sentences that define falsifiability by making reference in a very simple manner to the concepts of statements, logical relations between statements, etc., all of this separated from actual observations. I mean that the reader should not have to guess it from "observation statement" only. It should be much more explicit than that. It is likely to be the wrong guess anyway. This is the challenge that we face. — Dominic Mayers (talk) 17:21, 1 November 2020 (UTC)

Search for a formulation in our own words of technical falsifiability in the lead[edit]

The goal is to convey the technical notion of falsifiability in a way that is natural and simple and yet cannot be misinterpreted as the common notion. The logic behind this exercise is that if the weirdness disappears only because the statement can be interpreted to mean the common notion, then the real issue has not been addressed. Once we have a good formulation for the technical notion, a second definition of falsifiability that refers or might refer to the common notion should appear unnecessary, even confusing. The common notion corresponds to falsficationism in the article, as it is the case in Lakatos' book. It is not ignored at all. Only the conflict of definitions must be avoided because it is confusing for the readers and not notable in the literature. The key point is that this statement will have to be in our own words. Please avoid some kind of Witch-hunt against false OR because you see a new formulation. There is no ambiguity that the meaning is verifiable. Here is a first idea, but I am sure that we can do better:

In the philosophy of science, a theory is falsifiable if it respects two conditions. First, it must be written in a language, the empirical language of the theory, which must have a conventional empirical interpretation. Second, it must be contradicted by some statements of that language that can be interpreted as a true or false observation, but not by all them. These statements, which can be interpreted as a true or false observation, are the basic statements and those among them that contradict the theory are the potential falsifiers.

— Dominic Mayers (talk) 18:51, 2 November 2020 (UTC)

Retrieved from "https://en.wikipedia.org/w/index.php?title=Talk:Falsifiability&oldid=994207917"