What do you think about the theory of falsification (Karl Popper)? Do you use the Popperian falsification (e.g. deductive logic) approach in your research?
While Poppers way of postulating scientific statements as a modus tollens formulation is very common in scientific knowledge, just restricting to it is extremely difficult. While I do not think anybody has engaged in trying to translate every scientific some statement to this formulation I would doubt that it is possible. Even restricting scientific thought to just the initial axioms I would think that due to their very nature would not comply with the falsification principle since one could fall into an infinite regress.
This is a very interesting and important question for the current situation in science. I draw attention to only one problem related to the confirmation of published scientific results.
Unfortunately, at present, in most scientific and technical publications there is no detailed information, including both absolute and relative uncertainties of the measured variable, as well as a comparison of the difference between experimental and theoretical data and the achieved overall uncertainty of the studied variable. This data is necessary to confirm the verifiability, reliability and ability to reproduce the claimed results. Therefore, it is difficult for the reader to confirm or reject the formulated models and ideas, and this situation has developed in various fields of science and technology, for example, in psychology, metallography, physics, life sciences, economics, and refrigeration engineering:
1. M. J. T. Milton and A. Possolo, “Trustworthy data underpin reproducible research,” Nature Physics, vol. 16, no. 2, pp. 117–119, 2020. Available: https://sci-hub.tw/10.1038/s41567-019-0780-5.
2. C. A. Chapman, et al, “Games academics play and their consequences: how authorship, h-index and journal impact factors are shaping the future of academia,” Proceedings of the Royal Society B: Biological Sciences, vol. 286, pp. 1–9, 2019. Available: https://sci-hub.tw/10.1098/rspb.2019.2047.
3. M. Buchanan, “The certainty of uncertainty,” Nature Physics, vol. 16, no. 2, pp. 120–120, 2020. Available: https://sci-hub.tw/10.1038/s41567-020-0786-z.
4. M. Baker, “Is there a reproducibility crisis?” Nature, vol. 533, pp. 452–454, 2017. Available: https://www.nature.com/news/polopoly_fs/1.19970!/menu/main/topColumns/topLeftColumn/pdf/533452a.pdf.
5. L. P. Freedman, I. M. Cockburn and T. S. Simcoe, “The economics of reproducibility in preclinical research,” PLoS Biol., vol. 13, no. 6, e1002165, 2015. Available: https://doi.org/10.1371/journal.pbio.1002165.
6. G. Ellis and J. Silk, “Scientific method: Defend the integrity of physics,” Nature, vol. 516, no. 7531, 2014. Available: https://www.nature.com/news/scientific-method-defend-the-integrity-of-physics-1.16535.
7. B. Menin, “Uncertainty assessment of refrigeration equipment using an information approach,” Journal of Applied Mathematics and Physics, vol. 8, no. 1, pp. 23–37, 2020. Available: https://www.scirp.org/journal/Paperabs.aspx?PaperID=97483.
Dear Boris Menin interesting insights. I agree, there is a "problem related to the confirmation of published scientific results" due to the absence and/or incomplete information regarding absolute and relative uncertainties of the measured variable, as well as the assumptions/methodologies/tests, etc. that were applied to generate experimental data. I agree that the problem makes it hard to replicate the results, confirm the verifiability, reliability, etc. However, in economics (except for neuro/experimental economics I guess), it seems less problematic, I relate it more to other sciences mentioned by you.
Personally, I think that the problem can be solved (at least partially) by stricter requirements from the academic publishing journals; they are the ones who make scientific work accessible/visible to society. Of course, some of them already request the information to make replication of the submitted results easier, but it's not the common characteristic for the whole academic publishing industry.
Interesting topic .... following to see more information and comments
Arturo Geigel cannot disagree, restricting science to test theories only from the Popperian falsification theory is not right, as you said (sometimes) "due to their very nature it would not comply with the falsification principle".
In this context, Paul Feyerabend's way of thinking is also interesting. Feyerabend "urged abandoning the search for rules of science altogether; he argued that, since science is a creative and evolving enterprise, there is no specific method it ever did, or should, follow".* For Feyerabend, progress in science is not a result of the rational method. Contrary, "If there has been progress, it is only because working scientists have broken every principle in the rationalist's rule book and have adopted the motto "aything goes" (Broad 1979).**
* https://web.archive.org/web/20191104075438/https://www.iep.utm.edu/sci-ideo/
** Broad, W. (1979). Paul Feyerabend: science and the anarchist. Science, 206(4418), 534–537.
First of all: experiments can be thought experiments. Indeed, that's how real experiments are designed.
Second, trying to prove a theory isn't any different from trying to disprove it. In both cases what's going on is determining what certain consequences are and what would be the possible setups that can display them.
Finally, the statement that one observation that's inconsistent with the predictions of a theory is enough to disprove it is misleading.
The reason is that any theory involves approximations and what's interesting is what the experiments imply about how the approximations break down. And how this breakdown can, indeed, be part of the observation that claims to observe it.
And, of course, what's more interesting is how the approximations get improved, when taking into account the new observations. That's why claims of breakdowns of known theories aren't, by themselves, that interesting.
Stam Nicolis ,
I respectfully disagree with you on your second point. Affirming a theory is not the same as trying to disprove it. The difference can be highlighted in that one argument only tries to disprove but is not required to give alternatives constructions, in the other the construction must follow a logical path of coherence. From a proof perspective both paths can take on radically different strategies depending on the subject.
This was highlighted by Popper where his theory aimed to establish that scientific truths were so until they were tore down. This did not imply that it requires that another alternative immediately replaces it. there might not be one and we have to accept there is no solution.
In order to either ``prove'' or ``disprove'' any theory, it's necessary to produce a calculation within the framework of a theory. And in both cases one is focusing on the answer. If the answer is consistent, the theory is ``proved'', if it's inconsistent, the theory's been ``disproved''.
The difference between the two is that, in the case of an inconsistency, there's no point in even designing experiments, since these can't make sense; while for the case of consistent answers it does make sense, since experiments test not only the theory itself, but allow to discover the phenomena that have NOT been taken into account by it. These are then the discoveries. Precisely because experiments probe effects that are consistent with what's known, but lie beyond what has been discovered.
It's not a question of alternatives; but of possible completions. Scientific truths aren't ``torn down''; they're understood as part of a bigger picture. And the reason, precisely, is that they do possess internal consistency. A framework that isn't consistent isn't wrong; it's meaningless.
Davit Belkania ,
Feyerabend has a point and it can be naively simplified by a simple reasoning chain:
I think this is one point where Fayerabend and Popper do agree. For Popper, science is falsifiable and therefore all its postulates are subject to scrutiny and change.
The point of contention would be, on which are the rules that must/should be followed to achieve consistency in science. There is no clear answer to this.
In my opinion, a point to consider is that what happens with hypothesis also happens with rules with regards to the web of beliefs[1]. Most fundamental rules will be at the center of the web and therefore will be less subject to revision than peripheral ones. This is a kind of bias that we in science hold and must be careful when doing scientific revisions. A counter point is that holding constant revisions to every rule would stop science ad at one point scientific progress must be made. I think, that as with other subjects, the right approach (not truth) is somewhere in the middle.
[1] van Orman Quine, W. (1976). Two dogmas of empiricism. In Can Theories be Refuted? (pp. 41-64). Springer, Dordrecht.
Stam Nicolis ,
You take science as taking place in a vacuum, that is not how science has been done, is done or will be done.
your statement "in the case of an inconsistency, there's no point in even designing experiments, since these can't make sense"
is assuming that the inconsistency is known and all scientific statements that are currently present in our body of knowledge are free of them. Science takes place in context and a framework that isn't consistent is not meaningless, it is one not to be used again and therefore has scientific value (minimises the combinatorial space). One must tread carefully between logical arguments and the reality of science, logic is not the entirety of science (this philosophical stance has been dethroned for a long time).
Science is not a coherent body of knowledge as we have it right now but a fragmented snapshots of reality. stating that 'Scientific truths aren't ``torn down''; they're understood as part of a bigger picture' is a position of an omnipresent being and that is not the reality of science. I think that your argument is better presented within Quine's web of belief and it is just that a belief. This avoids all sorts of problems with your position on "truth" and "bigger picture" which is an ambiguous term that cannot be sustained.
There's science and there's the sociology of science. I'm interested in the former, since it's impersonal. It is possible to ignore the latter, as long as it affects personal feelings.
It is possible to decide, in an impersonal way, what part of any body of knowledge is consistent and what part is not and focus on the former; focusing on the latter might be good, if one is bored, which seems to be the case here.
Stam Nicolis ,
Your last post does not address the shortcomings of the argument you presented and shows a partialized view of science based on your interest. As a final point, the last sentence do not contribute nothing to your position (on the contrary it weakens your stance).
Once more: If you're interested in who's making an argument, that's sociology of science. If you're interested in what the argument is, independently of who's making it, that's science. The former includes all the usual things like credit, awards and fame.
There was a time when doing science was something exceptional-now it's part of normal human activity and an ordinary job. This means that, on the one hand, people can have the freedom to focus on the science and the sociology is a luxury rather than a necessity, whereas previously in order to do science at all the social part was significant. Not so much now. The resources for doing, in particular, theoretical work are available to practically everyone, thanks to on-line repositories, like the arXiv, whereas there was a time when access to journals was not available to people that did not work at universities, for instance. And it's now possible for anyone to ``publish'' their results on a web page that is accessible to everyone. So the content is available to universal scrutiny. So certain parts of scientific work-those that don't involve laboratory apparatus-is accessible to anyone whatever their ordinary job. Getting ``attention'', however, has become harder, simply because the number of people working on any subject is now very large in absolute terms.
What Popper is about is science, not the sociology of science, whether a theory can or not be proved/disproved, not whether people can be convinced about it by the eloquence of whoever is making the argument, or whether they can notice it, but by studying the content of the arguments. Nor what kind of social recognition is conferred on whoever discovers anything. Popper doesn't imply anything about whether someone will become famous, or not. It is, however, possible to develop theories for this, too.
The point is that there's no difference between ``proving'' a theory and ``disproving'' it, since, in both cases, calculating the consequences is necessary and sufficient.
What Popper doesn't mention explicitly is that among the consequences is the consistency of the theory, so checking that a theory is consistent is sufficient for rejecting it, if it turns out to be inconsistent, but not necessary for accepting it, if it is consistent. A theory can be consistent, but incomplete. And the theories of natural phenomena that are now known fall into this category: they're consistent, but incomplete and the interesting part of the scientific enterprise is showing how this incompleteness can be revealed and how the theory can be generalized to take this into account. This is a source, still, of confusion for many people, that continue to believe that scientific discoveries are about inconsistencies, whereas this is meaningless-they're about incompleteness of the current theory, not about its inconsistency, since an inconsistent theory, simply, doesn't make sense. It may be hard to show that any given theory is, in fact, inconsistent, of course. And many people will tend to refuse to believe that what they've been working on is meaningless. That's sociology, however, not science.
``Tearing down'' is sociology, not science, too, since nobody, who's interested in the science part, is interested in who got the wrong or right answer, but in what the answer is. And any such answer involves consistency checks, that show how the approximations are embedded. None of this refers to how science is done in any society, because that's irrelevant to the result. It just doesn't matter-some people will do it, others won't. And the result can be checked regardless of who produced it. How people are actually organized to do science doesn't have anything to do with the content. They're distinct issues. That's why it's pointless to confuse the two.
Stam Nicolis ,
Let me first address the straw man fallacy which you have apparently introduced in my arguments. Please read the statements I put forth carefully.
I will highlight some points worth mentioning that you have not addressed and kept on adding arguments that do not follow as counterpoints.
1) My initial point:
"Affirming a theory is not the same as trying to disprove it."
you did not address it nor the subsequent argument structure for shortcomings in my reasoning.
2) "You take science as taking place in a vacuum, that is not how science has been done, is done or will be done" From this, apparently (you can correct me if I'm wrong and I will correct it), you implied that I am talking about sociology (this does not follow from the statements, therefore it is a straw man fallacy). Talking about the structure of science does not imply sociology, it is a meta argument about science and how it is done. You introduced this fields as a way to weaken my argument, but it does not follow from my statements.
3) My point about disproving is about negative results in science. Negative results as a field of inquiry is as valid and contributes to science as much as positive results
4) The point I presented:
"Your last post does not address the shortcomings of the argument you presented and shows a partialized view of science based on your interest"
is to respond to your introduction into your argument structure of " I'm interested". This introduction in your argument shows partiality
5) A thing about your arguments that I find curious:
Stating arguments as if you are talking to air does not make more formal your argument, you can address me by name and it does not change the validity of your argument as long as you differentiate in your dialogue what is your main argument.
Again, please read the arguments carefully before trying to highlight flaws in them.
There's no difference between ``affirming a theory'' and ``trying to disprove it'', because the same procedure is followed in both cases. Using two different words
(``affirming'' vs ``disproving'') doesn't mean that the procedure's different. It's a question of personal taste, that doesn't affect the result, so it's possible to ignore the personal taste and focus on the result. How a result is obtained doesn't affect its validity.
The expression ``structure of science'' doesn't mean anything, so it has to be completed. It can refer to many things, so, as long as what these are isn't specified, it's not possible to complain that one meaning was used, rather than any other. One meaning is its social organization and I addressed that. If another meaning is provided, I can address that too (in fact I did, when I mentioned the consistency of a theory; I don't see anything else that could be understood under ``structure of science'').
There's no real difference between ``positive'' and ``negative'' results, beyond the use of the two adjectives. That's why treating them as distinct doesn't make sense-both are results and that's what matters. Someone's ``positive'' result is someone else's ``negative'' result, but there's really no impersonal way of stating what the two adjectives really contribute, beyond personal opinion. They don't affect the meaning of the result itself.
I'm always addressing the forum, that's why I prefer not to mention names. Names, for me, make sense outside of discussion fora.
Once more, the arguments only are relevant, addressing the people that are making them doesn't address the arguments.
Stam Nicolis ,
I disagree with your analysis of the last post on the following:
1) "the same procedure is followed in both cases" while I would question this but, I will grant you that point to focus on the main objection. The aims of both are different and it is the point of Popper (that is why he chose modus tollens as a logical construction as opposed to modus ponens). If you are referring to another theory, that is another point which we can take up in another post and will gladly reach consensus if it merits it.
2)" `structure of science' doesn't mean anything"
If you think that is an ambiguous term, you should address it the broadest interpretation and not narrow it down, since, it implies an inference on your part (you can enumerate possible interpretations and state that other could be included).
To emphasise, I am referring to the cumulative web of beliefs (see Quine). This point of view is in regards to the cumulative thoughts and coherence that is formed by scientific belief (this forms a structure and that is what I am referring to.
3) Suit yourself on not mentioning names but, If we acknowledge that this is a philosophy of science thread, then I would say that you need to review how to properly address a philosophical posture of someone else.
I think it is reductionist to think of scientific research today as theoretical. Most of the research being produced today are mainly applied sciences. In many cases, they follow neither purely inductive nor deductive logics. So while I agree that a scientific theory has to be falsifiable (for it to be scientific), it might be harder to apply this to most research being produced today.
Dear Davit Belkania and Readers,
This is an interesting topic. Following.
Counter example is useful for disproving a theory or hypothesis.
But it only works for theory of the form "All of ... is ...", "All ... has ..".
There are also exceptions, such as some famous paradox like Russel paradox.
There's no difference between ``counter''example and ``example'' and ``proving'' vs ``disproving'' a hypothesis; in both cases the same work's been done: to produce the example at all.
This is, by definition, done within a given framework that is the foundation of a thought experiment that, in turn, leads to a real experiment.
A ``counter''example in the framework of a thought experiment, means that the framework is inconsistent and it doesn't make sense to even design a real experiment, since the design can't make sense and any result is meaningless in the framework, since the latter is inconsistent. A prediction is an example that's consistent within a framework but hasn't been the result of a real experiment. This is the basis for the design of a real experiment.
A ``counter''example in the framework of a real experiment means a result that's consistent with the design of the experiment, but, since the real experiment is, also, sensitive to effects that are beyond the approximations that define the framework that allows the experiment to be designed in the first place, it's something new. How this discovery can be incorporated in an extension of the current framework is what matters and there are many ways, typically, each of which is consistent with the previous framework, incorporates the new effect and leads to the prediction of yet other effects, for which new experiments can then be designed. And so on.
It is true that considerable current activity is focused on designing real experiments using a theoretical framework that's well-understood; so the novelty lies more in the results than the theory. There are, nonetheless, many examples in material science and in medicine, to name but two fields, where the theory hasn't been fully worked out to produce the thought experiments that can lead to real experiments.
Beliefs are personal issues, that don't have any bearing on whether a framework is consistent or not or whether an experiment gives rise to a certain outcome or not.
The motivations of any person in doing a calculation or designing an experiment are irrelevant for testing the validity of the calculation or the outcome of the experiment.
Stam Nicolis ,
Your post, apparently is referring to another framework which is not Popper's. You have not addressed how modus tollens (which is one of the focus areas within Popper's argument and it is clear for someone who has read Popper) is equal to modus ponens (besides both being tautologies). This comparison is the simplest and does not take into account the other methods of doing inferences (good luck with that proof anyways). You have not addressed why his thought is lacking and why should somebody should accept your position. Further you have not provided proof of equivalence which is strange for someone who emphasises 'just the argument' and its formal correctness.
As an aside, in philosophy you have to contextualise your position with the works of others (we are in a philosophical thread not physics, computer science or other field). Just throwing your position shows that you are not within the context of this thread. As you mentioned my position was not clear as to what structure meant and I put it within the framework of another thinker from which the idea comes from. Not doing so is not appropriate within philosophy (as opposed to other posts in this thread, you are trying to put forth a philosophical posture without background). This is just like a 'previous section' in a research paper. Both serve to show that one has done her/his homework before listening to a proposal.
Hope to hear the argument in context soon with its proof, so that we can continue this dialogue in its proper context.
Stam Nicolis I guess your comments are not based on pure science nor a mathematical foundation.
You wrote:
"There's no difference between ``counter''example and ``example'' and ``proving'' vs ``disproving'' a hypothesis; in both cases the same work's been done: to produce the example at all. "
Karl Popper was a philosopher of Science, that evolve to epistemologist, which corespponds to the science of Scientific knowledge. The initial idea was to separate science from metaphisics. Initially he rejected the false results, and he call thisrefutation or falsification. This theory applied to all sciences but mathematicsIn my work in mathematics in RG I rejected some articles, the most notable theincompleteness Goedel theorem, the more famous in the last century.
César Rodrigues This theory applied to all sciences but mathematics?
I rejected some articles, the most notable the incompleteness Goedel theorem? Really?
Stam Nicolis RE: « "There's no difference between ``counter''example and ``example'' and ``proving'' vs ``disproving'' a hypothesis. »
Some hypotheses are capable of proof by example but not disproof by counterexample (i.e. in principle verifiable but not falsifiable):
There are 7 consecutive 7s in the decimal expansion of pi.
Dear Readers,
Popper's theory of falsification states that for a theory to be considered scientific it must be able to be tested and proven false. For Popper the scientist should attempt to disprove his/her theory rather than attempt to continually prove it.
My questions:
To put in the perspective, let P denote Popper's theory. If Popper's theory is itself P-scientific (based on his definition), then the theory P must be able to be tested and proven false.
Let H be a theory that P is true. It is unknown if H is falsifiable, too.
One can follow Popper's principle, instead of trying to prove H, let's try to disprove H first (falsifiable?). If H is true, then P is true. If H is false, then P is false.
If H is not falsifiable, then we won't know if P is false or not.
Thus, we see that H is not P-scientific.
One could argue that any attempt to prove/disprove H is not doing scientifically (P-scientific !) :)
As Karl Pfeifer pointed out previously, there are certain examples that are verifiable, but not falsifiable. Interestingly, P is likely to be no falsifiable.
Have fun everyone.
This is an interesting discussion. In my view, it is also important to differentiate between the natural sciences, where you would expect to get identical results in an experiment if the laws of nature hold (and the experiments are conducted correctly), and the social sciences where results are always probabilistic. I am therefore surprised how naively Popper's falsification approach is propagated in disciplines where it does not make much sense.
Here is what Lakatos has to say about it: “There is a third way of interpreting Popper’s falsifiability criterion: namely, by claiming that we should hold as definitely falsified any theory once it has been branded as falsified in specialized journals. […] This idea is too stupid and of course Popper never said anything like it” (Lakatos & Feyerabend, 1999, p. 89).
Lakatos, I. & Feyerabend, P. (1990). For and Against Method. Chicago & London: The University of Chicago Press.
This Link Maybe Useful:
https://explorable.com/falsifiability
Theory falsification is a tricky concept to grapple with. One egregious issue in my view is that simply by virtue of there being counterevidence to one's theory's predictions (i.e. some not-P as observed as opposed to some theory-predicted P) does not entail that there is a straightforward path to knowing what part of the theory is supposed to be amended.
Also, Popper's theory of falsification is not meant to be an empirical theory at all, hence why it cannot be falsified. His theory does not deal with empirical data because it is a consideration of how the logical relations between theory and evidence are supposed to be conceived in order to have theories that are more "truth-approximate". As such, Popperian falsification makes no predictions that can be falsified in the empirical world because his theory renders irrelevant whatever the empirical composition of the world is; all that matters simply is that empirical data exist.
Now, how Popper ends up considering truth-approximation in scientific theory crafting and experimentation is by doing away with induction and introducing his notion of corroboration as a deductive alternative. Unfortunately, history has shown how disastrously contentious his formulations have become.
Refutation/Falsification is the core of epistemology. Mathematics was the methaphisics because all was considered true without beeind subjected to critics.
I myself proved that Goebel theorem is false.
You mean Goedels Theorem? Are you really Sure that you can disprove it? IT IS one of the cornerstones of scientific reasoning. And IT IS very popular, EG. Compare hofstadter's Monographie Gödel Esher Bach...
As an answer read my article: Falsifikationsprinzip von Karl Popper aus Sicht der mathematischen Logik
Have a look at the "Survival of the Print Book! The Falsifiability Approach towards an Inductive Belief"
Article Survival of the Print Book! The Falsifiability Approach towa...
This is an example of what you considered.
The falsifiability approach proposed by Popper, to my mind, does not show that he tended more towards the deductive logic than towards the inductive one. One of his keywords is the use of the word "all" which means "whole" while most people really use it inconsistently such as in the following example: "All people like to be happy." Here the use of "All" does not mean "whole" as in the dictionary, but "all people" can be regarded as the same as "most people" or "nearly all people". His falsifiability is useful to differentiate God's pure revelation from human sciences gained from the limited methods and certain data because of the limitations of human senses and brains. I never find anyone who founded any science with complete data and the method enabling everyone to solute any proble being faced. So following the falsifiability can condition the Popperian to always feel in need of developing sciences in the course of life.
I think that the most important piece of work regarding how scientific method works and how science evolves is due to Thomas Kuhn. For me it is a must-read.
The main application of Popper theory was done by himself: the refutation of some results of Newton Mechanics in favour of Einstein Mechanics. that gave results that corretecd the corresponding Newton results when the speed is approximately the speed of light.
I say that I applied the Falsificaton of Karl Popper with relative success.
César Rodrigues Can you share a copy of your article "C. Rodrigues PA decidable, CCTC, December 2019." ? Is it the article you are mentioning in your attached pdf?
Article PA decidable
César Rodrigues That version is too brief. Can you share a better version with the readers?
For me Popper wants to refute the bad results and separate science from
metaphysics, like Kant.