Karl Popper versus Thomas Kuhn

[Ren]

In this thread we will be discussing the philosophies of Karl Popper versus Thomas Kuhn. Just so that as many people as possible can chip in, I will give a little of context first.

With the relative decline of metaphysics and epistemology in the twentieth century, two disciplines came forward as offering the hope of grounding the possibility of truth: philosophy of language and philosophy of science. Among influential philosophers of language we have Gottlob Frege, Ludwig Wittgenstein, Gilbert Ryle, and many others. On the other hand, I think the three most influential philosophers of science are respectively: Rudolf Carnap (logical positivism), Karl Popper (critical rationalism) and Thomas Kuhn (history of scientific ideas).

What Popper and Kuhn have in common

They both reject logical positivism, according to which a statement is cognitively meaningful only if it reduces to observation-statements which can be empirically verified. This is what philosophers usually refer to as the verification principle or criterion of meaning. Roughly speaking, this criterion says that if a statement can’t somehow be shown to consist in a series of empirical observations, it cannot be considered to be either true or false.

A well-known counter-example would be something like this: ‘There is no extra-terrestrial life.’ According to Carnap and his fellow logical positivists, this statement would have to be meaningless. This seems violently counter-intuitive (and even unintelligent), so philosophers like Popper were unsatisfied with it and came up with another criterion instead.

Where Popper and Kuhn differ

As an alternative to the criterion of meaning, Popper offers the criterion of falsifiability (or testability). Roughly speaking, this criterion provides a demarcation between science and non-science. All it says is that a theory (i.e. a unified collection of statements) must be falsifiable by being submitted to severe tests if it is to qualify as scientific. Note that this criterion is not concerned to locate truth as such. According to Popper, all that scientific theories can hope to achieve is getting closer and closer to the truth, but they are bound to be replaced over time by better and better theories.

Now, this is fundamental to understanding the difference between Popper and Kuhn. Popper accepts the following two premises:

1) Scientific theories are historical products
2) What makes a theory scientific (the criterion of falsifiability) is not a historical product.

While Kuhn would accept 1), he would reject 2). Kuhn thinks of the history of science as dominated by the relationship between ‘normal discourse’ and ‘revolutionary discourse’. Normal discourse is whatever set of rules and norms a scientific community accepts at a certain point of historical time. Revolutionary discourse is an alternative discourse (also born in time) that comes to challenge the normal discourse; if it is successful, it sooner or later solidifies into the new normal discourse. In turn, it might be challenged in the future by a revolutionary discourse; and so on.

A consequence of this is that for Kuhn, the criterion of falsifiability is just a part of our accepted normal discourse. There is no reason why it couldn’t be challenged and superseded in the future by a new revolutionary discourse—in the same way that, arguably, the criterion of meaning was superseded by the criterion of falsifiability. In other words, for Kuhn:

1) Scientific theories are historical products
2) Criteria for what makes a theory scientific are also historical products.

For Popper, the criterion of falsifiability is thought to be timeless: it is assumed that this criterion will forever allow us to separate science from non-science. As I tried to show, Kuhn rejects this. Who do you think is right? Who do you feel more compelled to agree with?

 

[philostam]

Interesting. So theory that wants to be scientific has to be falsifiable? So flat-earth theory is scientific, because it's falsifiable?

What does it even mean to be falsifiable? Some theories cannot be disproved with 100% certainty - is it enough if a hypothesis is rejected with 90, 95 or 99% certainty (like in stats)? Is it enough if it's disproven only by the means of logical argument?

Maybe I am still thinking too much in terms of theories in natural science. I know you made a video about why MBTI is science. I will re-watch that.

But pending further evidence, I will side with Popper for now. I know I haven't even come to the point where they differ - I want to understand the criterion of falsifiability first.

 

[Ren]

Interesting. So theory that wants to be scientific has to be falsifiable? So flat-earth theory is scientific, because it's falsifiable?

Precisely. Popper says something like this: "Many bad theories are scientific, whereas many interesting and fruitful theories are pseudoscientific".

What does it even mean to be falsifiable? Some theories cannot be disproved with 100% certainty - is it enough if a hypothesis is rejected with 90, 95 or 99% certainty (like in stats)? Is it enough if it's disproven only by the means of logical argument?

If the theory is disproved by means of logical argument then it would have to be assumed that the theory is just incoherent. I'm not sure the demarcation between scientific and pseudoscientific theorising is relevant in such a case.

As for falsification, Popper is saying that for a theory to be taken as seriously scientific, it must be capable of being falsified, so if some part of the theory is falsified this doesn't mean that the theory isn't good. I actually don't think the criterion of falsification is meant to help us distinguish good scientific theories from bad ones, in the sense that a theory which is mostly false may still be better than a theory which is mostly true but uninteresting (for example, if it doesn't make interesting new predictions with great explanatory power potential).

The idea is simply that if you submit a prediction to a test and it turns out false, you still have learned something. In a sense, you have gained genuine scientific knowledge from the very falsification of the prediction. You raise an interesting point though, which is roughly: what about predictions which may become testable in the future, but are currently not testable for whatever reason? Maybe @wolly.green or @Aaron Thyne can help us here.

And what is significance in all that - can you direct me to some practical examples or just further reading on the topic?

Well, a first element of significance is that Popper's criterion pretty much destroys many well-known theories' pretensions to being scientific. Among those you can cite astrology, psychoanalysis, and Marxism (dialectical materialism) in its modern form. Because Popper would say that a theory that claims that it can explain everything isn't falsifiable and therefore pseudoscientific.

Another advantage is that once you have your criterion, you can identify which parts of a theory are truth-functional, i.e. made of statements that are capable of being true. And then you can proceed to analyse a theory by e.g. comparing how many true statements it possesses versus how many false statements; and how many true statements may be derived from the true statements of the theory (its predictive power). Presumably a good theory would be a theory with large explanatory content derived from a relatively small number of truth-functional statements.

But pending further evidence, I will side with Popper for now.

Why though?!

 

[philostam]

As for falsification, Popper is saying that for a theory to be taken as seriously scientific, it must be capable of being falsified, so if some part of the theory is falsified this doesn't mean that the theory isn't good. I actually don't think the criterion of falsification is meant to help us distinguish good scientific theories from bad ones, in the sense that a theory which is mostly false may still be better than a theory which is mostly true but uninteresting (for example, if it doesn't make interesting new predictions with great explanatory power potential)

Oh, I see. I like that, yeah.

Well, a first element of significance is that Popper's criterion pretty much destroys many well-known theories' pretensions to being scientific. Among those you can cite astrology, psychoanalysis, and Marxism (dialectical materialism). Because Popper would say that a theory that claims that it can explain everything isn't falsifiable and therefore pseudoscientific.

Hmm. I am not sure how I feel about this. Aren't a lot of our theories just a fiction of someone's imagination and cannot really be disproven? How is anything that Jung came up with different than anything Freud came up with? So MBTI is scientific because it's clearly organized into categories/types and you can "objectively" type a person based on these categories, while psychoanalysis doesn't have that element to it?

I don't know: pseudoscience has a negative connotation. I can agree that psychoanalysis and dialectical materialism are pseudo-scientific in a sense that they obviously weren't reached by any scientific method; but I still think they are excellent theories.

Also: if "many bad theories are scientific", aren't also "many good theories unscientific"? So why does it even matter if psychoanalysis is scientific or not - as long as it's good?

But I understand Popper's ambition here. I think it's a decent criterion, definitely better than what philosophers had before. But I guess I do believe that even this criterion can be improved: so I am team Kuhn!

 

[Lady Jolanda]

But pending further evidence, I will side with Popper for now. I know I haven't even come to the point where they differ - I want to understand the criterion of falsifiability first.

How can you side with anyone if you don't understand the claims being made, and the differences between them?

Note that there's nothing wrong with not understanding and needing further information and explanation, but surely the only rational position to take then is that of being undecided?

If I misunderstood what you meant, and I really hope I did, please tell me so.

 

[philostam]

How can you side with anyone if you don't understand the claims being made, and the differences between them?

Note that there's nothing wrong with not understanding and needing further information and explanation, but surely the only rational position to take then is that of being undecided?

If I misunderstood what you meant, and I really hope I did, please tell me so.

Yeah, you're right. It was not really rational.

I sometimes do like to put myself in a camp just based on feeling/intuition and for fun. Basically ignore this, lol.

 

[Ren]

Hmm. I am not sure how I feel about this. Aren't a lot of our theories just a fiction of someone's imagination and cannot really be disproven? How is anything that Jung came up with different than anything Freud came up with? So MBTI is scientific because it's clearly organized into categories/types and you can "objectively" type a person based on these categories, while psychoanalysis doesn't have that element to it?

No, that wouldn't be the reason why MBTI is scientific. In fact, I think most of the versions of the MBTI theory people operate with is pseudoscientific. For example, CS Joseph's model is, to my mind, almost comically unscientific.

A pseudoscientific theory (again, following Popper) would be a theory that says: "Ooooh you thought you had found counter-evidence for my prediction, but look over there and you'll see that I have an auxilliary hypothesis that explains the counter-evidence in terms of the theory!" Circular reasoning, basically.

So CS Joseph would be able to say: "My theory failed to predict use of x function in this particular instance? Of course not, because right now what was manifested was your superego type, not your real type", etc. So he could theoretically type you as INFP and then just say that when you think you show Ti use, it's just your superego-type speaking. You wouldn't be able to falsify his predictions.

I think it can be easily inferred from the above how he could end up typing Brad Pitt as ESTJ and Paris Hilton as INTJ without thinking twice about how those typings could be challenged. His theory provides him with a trick to answer just every conceivable objection.

We would be very familiar with that kind of tactic in astrology, but it pops up everywhere in the theoretical domain. Popper has a fair amount of respect for early Marxism actually, but he thinks that ever since Marxist predictions have been proven false in relation to the classless society etc., the attempts of Neo-Marxists to salvage those predictions in terms of 'we just haven't gotten there yet' or whatever, are only weakening the power of the original theory. He in fact says explicitly that the original statement of dialectical materialism was scientific.

 

[Deleted member 16771]

We would be very familiar with that kind of tactic in astrology

12 houses, 9 planets, a sun, a moon, lilith, North node, south node... that's like 25 auxiliary hypotheses for a personality description if the first one fails, lol.

 

[Ren]

12 houses, 9 planets, a sun, a moon, lilith, North node, south node... that's like 25 auxiliary hypotheses for a personality description if the first one fails, lol.

Lmao, exactly.

PS. Let's make sure Sandie doesn't see this

 

[Ren]

Just to make it clear, I'm not saying I'm on the side of Popper. I just want to give as charitable a description of his position as possible.

There's no doubt that the criterion of falsification can and should be criticised (in proper Popperian spirit, I suppose).

 

[Deleted member 16771]

@Ren I read your OP.

I'm not sure that Popper and Kuhn are particularly comparable since their work doesn't contradict each other. Kuhn just noticed a pattern in how scientific revolutions proceed, while Popper was attempting to define what science 'is', or rather, establish a condition of validity for scientific theories. They are doing different things.

Are you familiar with Carl Hempel and the DN vs IS (modern) debate?

 

[Ren]

Are you familiar with Carl Hempel and the DN vs IS (modern) debate?

Only broadly and not enough to be able to comfortably talk about it! Do you have a good source on this or maybe you want to give us a summary (especially if it is relevant to the current discussion)?

I agree that Kuhn and Popper have different topics in mind, and do not compete on every level. But some implications of Kuhn's work seem to me to conflict with Popper's—just in the sense that I doubt Popper would be willing to accept that his criterion is just part of a paradigm that could as well be superseded in the future. Maybe he is, but this is not what I gather from his views or from his reaction to Kuhn (see here for example). See also this article:

"Kuhn's view diverged in several important respects from the philosophy of Karl Popper, who held that theories can never be proved but only disproved, or "falsified." Like other critics of Popper, Kuhn argued that falsification is no more possible than verification; each process wrongly implies the existence of absolute standards of evidence, which transcend any individual paradigm."

I think the key term here is absolute standards of evidence. Kuhn seems to believe that these standards are only ever historical and intra-paradigmatic, not absolute. Whereas I think Popper would want to say that his criterion can be applied across all possible paradigms.

That being said, if you think Popper and Kuhn can actually be reconciled or shown not to diverge as seriously as we might think, that would be a cool idea as well!

 

[Krypton]

I agree with Popper that you should vet your theories (although I think we should probably iron out degrees of vetting and such), though I'd say that you can accept something that hasn't been fully vetted as provisionally true when you don't have access to the information required to vet it, provided the impact of screwing up isn't so big. Where the line is, is up for debate of course.

 

[Ren]

I agree with Popper that you should vet your theories (although I think we should probably iron out degrees of vetting and such), though I'd say that you can accept something that hasn't been fully vetted as provisionally true when you don't have access to the information required to vet it, provided the impact of screwing up isn't so big. Where the line is, is up for debate of course.

I'd say most of us would agree that our theories should be vetted. The question is whether the idea of falsification is the best approach to vetting them.

One possible argument against falsification I can think of is the risk that too much focus on testing would prevent brilliant scientists from coming up with wildly 'out there' theories that although not falsifiable, may contribute to revivifying the spirit of scientific research and its scope.

Something like that.

 

[Deleted member 16771]

Only broadly and not enough to be able to comfortably talk about it! Do you have a good source on this or maybe you want to give us a summary (especially if it is relevant to the current discussion)?

I agree that Kuhn and Popper have different topics in mind, and do not compete on every level. But some implications of Kuhn's work seem to me to conflict with Popper's—just in the sense that I doubt Popper would be willing to accept that his criterion is just part of a paradigm that could as well be superseded in the future. Maybe he is, but this is not what I gather from his views or from his reaction to Kuhn (see here for example). See also this article:

"Kuhn's view diverged in several important respects from the philosophy of Karl Popper, who held that theories can never be proved but only disproved, or "falsified." Like other critics of Popper, Kuhn argued that falsification is no more possible than verification; each process wrongly implies the existence of absolute standards of evidence, which transcend any individual paradigm."

I think the key term here is absolute standards of evidence. Kuhn seems to believe that these standards are only ever historical and intra-paradigmatic, not absolute. Whereas I think Popper would want to say that his criterion can be applied across all possible paradigms.

That being said, if you think Popper and Kuhn can actually be reconciled or shown not to diverge as seriously as we might think, that would be a cool idea as well!

Kuhn here is just acting as the voice of one side of the discourse, really, hence why I'm a bit puzzled by your choice of 'Kuhn vs Popper'.

Kuhn's particular contribution to philosophy of science isn't relevant to the disagreement you establish here, which is just another rehash of postmodern challenges to truth, which I don't find that interesting personally (as a topic distinct from your thread, of course - I'm not saying the thread is dull).


The Wikipedia article on the deductive-nomological (DN) model is pretty good at laying out the debate I refer to.

 

[Ren]

Kuhn here is just acting as the voice of one side of the discourse, really, hence why I'm a bit puzzled by your choice of 'Kuhn vs Popper'.

Kuhn's particular contribution to philosophy of science isn't relevant to the disagreement you establish here, which is just another rehash of postmodern challenges to truth, which I don't find that interesting personally (as a topic distinct from your thread, of course - I'm not saying the thread is dull).

Fair enough, I can understand that. But maybe some of us do see a problem and would like to discuss it.

 

[philostam]

No, that wouldn't be the reason why MBTI is scientific. In fact, I think most of the versions of the MBTI theory people operate with is pseudoscientific. For example, CS Joseph's model is, to my mind, almost comically unscientific.

A pseudoscientific theory (again, following Popper) would be a theory that says: "Ooooh you thought you had found counter-evidence for my prediction, but look over there and you'll see that I have an auxilliary hypothesis that explains the counter-evidence in terms of the theory!" Circular reasoning, basically.

So CS Joseph would be able to say: "My theory failed to predict use of x function in this particular instance? Of course not, because right now what was manifested was your superego type, not your real type", etc. So he could theoretically type you as INFP and then just say that when you think you show Ti use, it's just your superego-type speaking. You wouldn't be able to falsify his predictions.

I think it can be easily inferred from the above how he could end up typing Brad Pitt as ESTJ and Paris Hilton as INTJ without thinking twice about how those typings could be challenged. His theory provides him with a trick to answer just every conceivable objection.

We would be very familiar with that kind of tactic in astrology, but it pops up everywhere in the theoretical domain. Popper has a fair amount of respect for early Marxism actually, but he thinks that ever since Marxist predictions have been proven false in relation to the classless society etc., the attempts of Neo-Marxists to salvage those predictions in terms of 'we just haven't gotten there yet' or whatever, are only weakening the power of the original theory. He in fact says explicitly that the original statement of dialectical materialism was scientific.

Oh, wow, I see. This made it much clearer. I already downloaded some introductory podcasts about this Popper guy, because I find this interesting.

In a way he is saying that it's better for "truth" to make a bold claim that can be tested, than to make a cautious theory that cannot be disproved. it's better to be wrong, because it will lead to a "truer truth" down the line. I do really like that.

Quickly reading his Wikipedia page, I see he also has problem with induction. This is also something I have a problem with, so it will be interesting to read his thoughts. I feel Ti is deductive, while Ni is more inductive. I am always baffled when people make predictions.

 

[Ren]

Oh, wow, I see. This made it much clearer. I already downloaded some introductory podcasts about this Popper guy, because I find this interesting.

In a way he is saying that it's better for "truth" to make a bold claim that can be tested, than to make a cautious theory that cannot be disproved. it's better to be wrong, because it will lead to a "truer truth" down the line. I do really like that.

That's great! Yes, to be honest I think there is much of great value to be derived from Popper's thoughts on theories and science.

It also helps that unlike a lot of his peers, he is remarkably clear and easy to read. He never tries to deceive the reader, and is suspicious of any theory that attempts the same. Once his tools become part of your conceptual arsenal they can be put to very productive use in discussions.

Quickly reading his Wikipedia page, I see he also has problem with induction. This is also something I have a problem with, so it will be interesting to read his thoughts. I feel Ti is deductive, while Ni is more inductive. I am always baffled when people make predictions.

Oh yeah, he has a major problem with induction. (The consensus is that he is ENTP, so a Ti user).

I haven't yet read the whole of his Logic of Scientific Discovery, but I'm ready to bet that he provides an account of scientific discovery that doesn't rely on induction. Again, when @wolly.green shows up I'm sure he will be delighted to tell us more about why induction is preposterous

 

[Krypton]

I'd say most of us would agree that our theories should be vetted. The question is whether the idea of falsification is the best approach to vetting them.

One possible argument against falsification I can think of is the risk that too much focus on testing would prevent brilliant scientists from coming up with wildly 'out there' theories that although not falsifiable, may contribute to revivifying the spirit of scientific research and its scope.

Something like that.

I think falsification is a tool to be used along with intuition and everything else. You shouldn't only use one tool, but it's good to keep each tool in mind and understand exactly where and when you use it. I would say in some fields we might need to falsify a little more (those with a replication crisis), but I would want us to have maybe something like an "exploratory paper", where you know that what you're publishing isn't necessarily true, but is a jumping off point for further testing.

I will say I think that some theories can get stronger after attempts to falsify them though. I'm not super well-versed on this, but I recall reading that string theory mutated into various different sub-theories after the initial idea was criticized.

I'm also not really sure what the upper limit of falsification really is, and how you would do it in a way that isn't also kind of attempting to verify something at the same time. Like you could attempt to run a test to see if something happens to contradict a theory, but by doing so, you're also running a test to see if the theory works as expected. And are you supposed to falsify by controlling for other possible factors, by replication, or both? Is it enough to compare to the other possible factors widely known about, or do you have to actively think of them?

 

[Deleted member 16771]

Fair enough, I can understand that. But maybe some of us do see a problem and would like to discuss it.

I read one of those articles you posted - the one by John Horgan - and he's an interesting character himself. He seems to be the Francis Fukuyama of science journalism, heralding the 'end of science' in one of his books (yep, published in the mid-90s).

In his article he kept comparing Kuhn with Popper, I think because he's enamoured with the idea that the ossification of scientific disciplines could result in the 'end' of discovery, and Kuhn is the most famous mouthpiece of that idea, with his 'idea' being its most succinct and articulate version.

It's clear that the Kuhn-Popper comparison is something close to Horgan's heart as a dyed-in-the-wool 'scientific postmodernist'. His selection of 'Popper' as a foil here is telling, I think, since Popper (and falsification) is a highly visible, well-known example used by popular science communicators/optimists/hardliners everywhere.

It seems more like a shot across the bows of his journalistic opponents than anything else. 'Your guy is Popper, mine is Kuhn'.


I'm not dismissing your discussion here, Ren, but maybe my dismissiveness betrays my own position - that I don't find postmodern challenges to truth particularly compelling or revelatory anymore. Whenever I come across something like this, I tend to think that the arguments are tired and trite and so roll my eyes at the attempt to muscle in on 'genuine' discourses. So by that logic, you'd have to put me in the Popper camp, lol, even though ironically Kuhm's idea is part of the general class of ideas that I am most interested in in the. WHOLE WIDE WORLD.