Philosophy of Science 5 - Falsificationism

[@MrAndreaCaso]

Great. As usual. Thank you so much, Kane!

[@rezamahan7109]

THANK YOU THE GREAT KANE! i have ph.sc exam tomorrow and this really helps

[@Ansatz66]

The problem of ad hoc hypotheses seems to be just a misunderstanding of falsificationism. There's no reason at all why we should avoid ad hoc hypotheses; quite the contrary an ad hoc hypothesis is just another way of saying that we modify our theories to better reflect the real world. We can talk about saving a theory from being falsified and make that sound deceitful, but in the end we still get all the benefits that science is supposed to provide. We had one idea of how the world works, then we did a test and the test failed, thereby deductively demonstrating that we are wrong about _something._ We don't know what we're wrong about, but so long as we come up with a new idea about how the world works that succeeds where the old one failed, we're still making progress. Our new idea is that little bit better than our old one, because our old idea is false while our new idea might be true. For example, suppose we do an experiment and the failure forces us to either conclude that Boyle's Law is wrong or our pressure gauge is broken, or any other of the endless ways to explain what we got wrong in our test. It doesn't really matter what we suspect is the source of the failure, because the true point of the test was just to let us know that there is a failure. Before we did the test we probably thought that both Boyle's Law and the pressure gauge were working as expected, and now we know better. Because we know that, we can get a new pressure gauge and try again, and if the test still fails then maybe we'll consider other ways to explain the results and then test those. This is not a problem; this is the exploration that one expects from science; coming up with new ideas and then checking them against reality. Perhaps the reason ad hoc hypotheses strike people as wrong is because they seem akin to lying. When a liar is caught in a lie, the liar will likely come up with a new lie to explain away whatever evidence is causing the problem. How can we be getting closer to the truth when we're just inventing new lies to cover the flaws in our old lies? The point is that science is _not_ getting closer to the truth; this is just what Popper was trying to say. We'll never really know how the world works; all we're doing is telling ourselves stories about how it _might_ work. Science is the storyteller that invites people to check that the story works and updates the story whenever someone finds a problem. Pseudoscience tells a story that is arranged so that it could never need fixing, and therefore it never gives any clear answers. Religion tells a story with the understanding that anyone suggesting there is a problem with the story will be tortured for eternity. There's no reason to think any of the stories are literally true, but science is still the best of the bunch by far. When you look at the competition, it's clear that it doesn't really matter _how_ we choose to fix the story; so long as we keep searching for problems and we keep coming up with fixes, we're still ahead of any of the alternative approaches to storytelling.

[@KaneB]

I suppose if you hold that all statements about the world are "equal", then ad hoc hypotheses would not be a worry. But I think Popper would draw a distinction between general laws such as "f=ma" and more limited claims such as "there is an eighth planet about x distance from the Sun". Furthermore, you will still need to say something about how you rule out the possibility of scientists introducing endless ad hoc hypotheses to save a particular favoured hypothesis from falsification. The Marxist, for example, could keep introducing ad hoc hypotheses to explain away failed predictions, but then make new, specific predictions for the future (and then when those predictions fail, she introduces more ad hoc hypotheses to explain away their failure, etc). There seems little difference in practice between this procedure and the tendency of a theory to make "too many predictions" - i.e. to be compatible with any observation. In either case, we have a theory which is indefinitely shielded from potential falsification.

[@Ansatz66]

Kane B "In either case, we have a theory which is indefinitely shielded from potential falsification." There's more than one way to shield a hypothesis from falsification. Some ways are bad and other ways are good. The example of psychoanalysis is shielded from falsification by making no particular claims; nothing that could happen would contradict psychoanalysis because it doesn't commit itself to anything that can be checked. This is bad because it makes psychoanalysis practically useless, just like all pseudosciences tend to be practically useless. Religions shield themselves from falsification by threatening people who speak of falsifying it with excommunication, eternal torture, or whatever. It's certainly debatable whether this is bad. Science stands apart because it doesn't use either of those shields. Instead science uses the ad hoc hypothesis. Instead of trying to evade problems with our theories, we use problems with our theories to inspire new ideas about the nature of the world, up to and including discovering new planets. If we didn't do this, then we'd abandon Boyle's Law every time a pressure gauge broke, because in science we're not allowed to just ignore a result. Everything needs an explanation and it all has to be consistent, so we end up dividing the world of pressure gauges into those that conform to Boyle's Law and those that are broken. This is practically useful information and it is all thanks to ad hoc hypotheses. Perhaps the most famous ad hoc hypothesis is dark energy, the mysterious something which we've used to explain the accelerating expansion of the universe. No one knows if dark energy is real or if some other explanation would be better, but ultimately all that matters is that when you use dark energy you get an explanation of the universe that fits with everything we see, so you can open a science text book and expect to read things that fit with the real world as closely as we can manage. When there are too many ad hoc hypotheses it can get to be a practical problem because the world seems to become more and more complicated until no one can understand it. Hopefully someone will find a new theory that explains it all more simply, just as heliocentrism allowed us to describe the motions of the planets far more simply than previous theories, but this is just a bookkeeping issue. Science could just as well go on accumulating epicycles forever. "The Marxist, for example, could keep introducing ad hoc hypotheses to explain away failed predictions, but then make new, specific predictions for the future (and then when those predictions fail, she introduces more ad hoc hypotheses to explain away their failure, etc)." Are all the ad hoc hypotheses of Marxism being recorded to accumulate into a picture of the world that explains everything consistently? In other words, are the ad hoc hypotheses being taken seriously, or are they just a means to brush off people asking awkward questions? If Marxists are seriously updating their picture of the world with every failed prediction, creating a more and more complicated model of the world over time, then they are doing science. They are doing all that anyone can do to understand the world; keep our eyes open for problems and seriously try to correct our understanding when we find a problem.

[@KaneB]

"Science stands apart because it doesn't use either of those shields. Instead science uses the ad hoc hypothesis." But the point is that there is a limit to how much of a shield ad hoc hypotheses provide. There comes a point where ad hoc hypotheses no longer cut it. Hacking, in "Representing and Intervening", pg 252, gives the example of Prout's hypothesis that all atomic weights are multiples of that of hydrogen, so that if hydrogen=1, then every other element will have a whole number. This ran into the problem that chlorine had a weight of 35.5. Initially, it was proposed that the chlorine had been imperfectly purified. Various ways of improving purification were developed, but gave the same result. Of course, scientists could have continually insisted that there must be some impurities still not removed - i.e. they could have continually introduced new ad hoc hypotheses to save the theory - but as Hacking says, after many new methods had been tried, this stopped being plausible. Eventually Prout's hypothesis had to go. And for good reason; continually making excuses to retain a particular hypothesis is not especially scientific. It seems to me that these kinds of cases give some credence to Popper's claim that introducing ad hoc hypotheses degrades scientific status. Anyway, I think if you're okay with the idea of fundamental laws being shielded from falsification indefinitely, then you're not really a falsificationist. That completely violates the spirit of Popper's model. For the falsificationist, the basic problem with psychoanalysis and Marxism is that nothing can happen that would lead supporters of those theories to reject their central claims. If the scientific method allows scientists to make whatever ad hoc hypotheses they like - if we don't introduce restrictions on what kind of ad hoc hypotheses can be introduced, and how many, and when, etc - then in practice all other theories are in the same position. "Are all the ad hoc hypotheses of Marxism being recorded to accumulate into a picture of the world that explains everything consistently?" Well, this wasn't a real example. I don't know much about how Marxism actually developed. But we can simply stipulate that yes, they have a picture of the world that explains everything consistently. Even in this case, Popper would probably not be happy if that picture was the result of hundreds of ad hoc hypotheses, all designed to rescue the fundamental Marxist laws (i.e. the foundational claims about the nature of class conflict etc).

[@Ansatz66]

Kane B "Scientists could have continually insisted that there must be some impurities still not removed - i.e. they could have continually introduced new ad hoc hypotheses to save the theory, [but] after many new methods had been tried, this stopped being plausible." There's no point in worrying over plausibility when none of our theories are actually true. All that matters in science is whether our theory fits with our experiments. It would just have been another dark energy. No one knows what it is or why it's there. All we know is that it lives with chlorine and makes it hard to get pure chlorine. There's no real harm done by working under such a theory until we find something that actually contradicts it. Pseudoscience tends to stifle investigation because it makes everything nebulous and impossible to pin down to specifics. Notice how ghosts are virtually impossible to investigate. An impurity for chlorine has the opposite effect, spurring on investigation into chlorine to try to separate the properties of chlorine from the properties of its impurity. "Continually making excuses to retain a particular hypothesis is not especially scientific." We still have to test the hypotheses and take them seriously. They become part of our new theory, published, reviewed, written into text books, and so on. We're not just brushing off an objection and then forgetting the ad hoc hypothesis we used; that _would_ be unscientific. Instead we're just risking a violation of Occam's razor. Obviously if something could be just as well explained by a single hypothesis instead of a hypothesis and a collection of ad hoc hypotheses, then it's better to use the single hypothesis. Since we're just telling ourselves a story about how the world might work instead of chasing after truth, we may as well make it as simple a story as possible. Even so, that doesn't make it unscientific to have complicated theories; it's just inconvenient. For all we know, the actual truth could be far more complicated than anything we've ever imagined, and only people who invent ad hoc hypotheses will ever have a chance of lucking onto the truth. "If you're okay with the idea of fundamental laws being shielded from falsification indefinitely, then you're not really a falsificationist." Even so, a hypothesis is still scientific only as much as it is falsifiable. All real scientific investigation is an attempt to falsify one thing or another. There is no quest for truth; it's a quest to weed out the false. It's exactly falsificationism. The only question is what we should do when we find that our current theories have led us to a false prediction. We need to fix that situation somehow, but do we scrap one hypothesis and replace it, or add a new hypothesis? There's no deductive way to make that decision. Just as our original hypotheses could have come to us in dreams, we may as well let a dream make the decision of how to fix our hypotheses. Or we could give the job to a poet. So long as our hypotheses are fixed somehow and the new hypotheses are still falsifiable, we're still making progress. "The basic problem with psychoanalysis and Marxism is that nothing can happen that would lead supporters of those theories to reject their central claims." What goes on in the heads of the supporters of a theory is not important. What matters for a falsificationist is what claims they make. So long as Marxism makes specific claims that we can test in the real world, then it is falsifiable. If the supporters of Marxism ignore the results of testing their theory, then that's just their mistake. If they come up with a new Marxism (perhaps called Marxism 2) that is like the old Marxism but adjusts its claims to account for the test results, then they are making scientific progress. If we make enough scientific progress then we'll either end up with a version of Marxism that makes no testable claims at all, in which case it becomes obviously irrelevant and the domain of pseudoscience, or else we'll have a version of Marxism that makes claims that withstand all testing, and so it will be the best version of Marxism ever, even if it contains tons of ad hoc hypotheses. "Popper would probably not be happy if that picture was the result of hundreds of ad hoc hypotheses, all designed to rescue the fundamental Marxist laws." Popper probably liked Occam's razor. Occam's razor is fine, but it's not a guide to truth. The discomfort we feel at a hideously complicated theory is only a problem for our attempts to learn the theory, not a sign that the theory is false. Only experiments can show a theory is false.

[@KaneB]

"There's no point in worrying over plausibility when none of our theories are actually true. All that matters in science is whether our theory fits with our experiments." The realism/antirealism debate is not really relevant here. The important point is that Prout's hypothesis became increasingly *unacceptable*; it doesn't really matter whether we explain unacceptability in realist or antirealist terms. And it seems that you agree that Prout's hypothesis did indeed become increasingly unacceptable - see next quote... "Obviously if something could be just as well explained by a single hypothesis instead of a hypothesis and a collection of ad hoc hypotheses, then it's better to use the single hypothesis. Since we're just telling ourselves a story about how the world might work instead of chasing after truth, we may as well make it as simple a story as possible. Even so, that doesn't make it unscientific to have complicated theories; it's just inconvenient." I think you're conceding all that's important here. You agree that scientists should follow the norm of avoiding ad hoc hypotheses, or at least avoiding too many ad hoc hypotheses; you say that this is not a scientific norm, though you grant that it is a norm that scientists should follow and that increasing ad hoc hypotheses make a theory worse. Personally, I'm inclined to say that if some norm X is such that (a) scientists should follow X and (b) failing to follow X makes a theory worse, then we have at least pretty strong reason for treating X as a scientific norm. But not much really hangs on this. To put this another way: Popper would say that if a theory T1 has been saved by introducing various ad hoc hypotheses, whereas theory T2, which is empirically equivalent and applies to the same domain, has very few ad hoc hypotheses, then T1 is less scientific than T2. You say that T1 is no less scientific than T2, but you agree that T1 is worse than T2 and that scientists should adopt T2. I don't see anything interesting in this disagreement. Importantly, it turns out that you agree with Popper that it is a bad thing if fundamental laws are shielded from falsification indefinitely. Popper would say it's bad and unscientific, you just say it's bad. Well, either way, scientists shouldn't do it. "What goes on in the heads of the supporters of a theory is not important. What matters for a falsificationist is what claims they make. So long as Marxism makes specific claims that we can test in the real world, then it is falsifiable." Maybe that's what matters to you - I'm not sure falsificationists in general would agree with that though. I certainly don't think Popper, for example, would approve of a group of scientists exhibiting that kind of attitude to their theories. "Only experiments can show a theory is false." Right, and the trouble is that in the example we're talking about, no experiments could show the fundamental Marxist laws to be false, because the Marxists of our example are such that they will introduce endless ad hoc hypotheses to save their theory.

[@andoreanesnomeo1706]

RIP Karl Popper. His thinking has stood the test of time and continues to bear fruit. Everyone should read him.

[@Lfppfs]

Thank you for sharing these lectures on KZbin, they are great!

[@alexmeyer7986]

Great, as always. Keep them coming! Cheers.

[@thomasmuandersontheneousul4184]

Good stuff Alex - seeing your recent Quine - indeterminacy video however seems to show Popper's route won't work Things that might seem well refuted yesterday do make comebacks (eg recent crises in capitalism suggest Marx had some valid theories, etc.) I think of something like Darwin's theory of evolution - we could now construct AI programs that literally IMPROVE with each new generation - something Darwin could not anticipate - so do we throw out the theory or redefine it? Thanks will look at the rest of the series

[@sosscs]

Did Popper used induction to formulate his falsificationism? He induced marxism, psychoanalysis and theory of relativity didn't he?

[@benjaminprzybocki7391]

The contrast between those theories inspired his position. That doesn't mean he used them as inductive evidence for an empirical hypothesis.

[@vincentdumaij1995]

This video was very helpful! keep up the good work!

[@shihabkhan3218]

Coolest introduction to Kalr Popper.. 😇

[@MatthewMcVeagh]

The difference between astrology and astronomy is not, IMO, to do with preparedness to make bold conjectures (and then abandon the theory when it's falsified). Astrology is not a science and is not meant to be a science; its function is based on a notion of a hidden connection between two phenomena, and this cannot be tested scientifically as there is no empirical evidence of that connection. It also requires interpretation through a human mind, rather than direct physical evidence. I am generally a Popperian but the one area I disagree with him is over the demarcation criterion: I think falsifiability is not the only criterion for science, instead empiricality is too. A scientific theory has to be something science can test with an experiment that gets to the heart of what the theory is about. In the same way Zener card tests can't establish anything about ESP because ESP is not a phenomenon that physically shows up: instead at most an anomalous Zener card test score shows up. I also think the word 'pseudoscience' is only appropriate for things which purport to be science but aren't. That seems to include Marxism and psychoanalysis on many believers' accounts; there are some who say astrology is a science, but they are a minority so I don't think it should be called a pseudoscience. I am an astrologer who is clear that astrology is not a science and is not supposed to be one. If we start using 'pseudoscience' for things which are not meant to be sciences in the first place we will end up using it for arts, history or philosophy, which is absurd. Basically there is more than one form of human enquiry; 'science' as we currently know it has only been around a few centuries and it is ludicrous to consider it taking over and outlawing anything that can't be identified with itself.

[@MatthewMcVeagh]

@Oners82 Haha, well there's an ignorant and prejudicial opinion if there ever was one! Now what is this point I'm supposed to have missed?

[@MatthewMcVeagh]

@Mike Kane I'm not sure I'd agree with that description of astrology. I don't see what 'problem-solving' has to do with it, or in what sense it's 'calculative' except that mathematical calculations are needed to plot locations in space. You're also ignoring the context in which I linked astrology and the arts - and history and philosophy. It was specifically to do with their not being supposed to be sciences, and therefore it not being a problem that they fail to function as sciences. They don't need to have any other connections. The extension of 'pseudoscience' to things that don't purport to be sciences in the first place is ludicrous. It shifts the meaning of 'pseudo-' from "pretending to be" to "not being". I also don't think astrology makes 'predictions' in the same sense as scientific ones. It aims to clarify people's personalities by virtue of hidden, unknown connections. As part of that it can illuminate events in people's lives. It can't give specific, accurate predictions of specific events in the future: that's too complex. And of course "inaccurate and unfounded" is just your opinion; experienced astrologers have experience of massive amounts of accurate connections and correspondences. As I said it takes the subjective interpretation of a mind to determine, and that's not something that can be tested or demonstrated scientifically, so it's a mistake for anyone (pro or anti astrology) to try to treat it as a science that should be judged by the standards of science.