A Theory is Just a Theory (Underdetermination in Scientific Theory)

[@VasselofGod2]

so is one of the ways of saying this essentially that sometimes what we assume is the correct scientific theory because it is supported by evidence X may not always be correct because another scientific theory can also be supported by evidence X?

[@CarneadesOfCyrene]

***** That's contrastive underdetermination. Holistic underdetermination says that when our predictions fail, we cannot show that the hypothesis that we are testing was incorrect, as in fact it might have been one of our other theories.

[@VasselofGod2]

Carneades.org oh, I see

[@highvalence7649]

@@CarneadesOfCyrene shouldn't we in that case say all theories making those failed predictions are incorrect? i don't understand.

[@tnyw872621h8474h9]

Hello, I asked this question to a philosopher and I didn’t get a satisfying answer so I will ask it here. If these problems of underdetermination talked about are the case (which I believe they are) why do they not apply to worldviews as a whole, rather than just scientific theories. And then how is it possible to get to Truth in any sense?

[@CarneadesOfCyrene]

I am a philosophical skeptic, so I don't know if we can find truth at all. You can check out my series on Postmodernism for an explanation of how similar ideas of inability to judge accuracy outside of a paradigm/metanarrative can be found throughout other fields. kzbin.info/aero/PLz0n_SjOttTcLQyeXoDeqR0LGO3JCoLbO

[@gdn5001]

Is ampliative underdetermination plausible? I accept logical underdetermination as a genuine thesis, but I'm not really sure about the ampliative version. Jarret Leplin, a philosopher of science, stated in an article that most modern debates over underdetermination have moved on from logical underdetermination. Here's an example that I think will help me highlight the difference between the two: Theory 1: Lightning is caused by the movement of electrons from an area of low to high charge. Theory 2: Lightning is caused by the movement of electrons from an area of low to high charge, which is directed by the god of lightning Thor. These 2 theories are logically underdetermined but not ampliatively underdetermined. Though both are consistent with the evidence, they are not equally supported by it. Theory 2 arbitrarily postulates an entity that isn't necessary whatsoever, so there's no reason to keep it. If a postulate could be arbitrarily removed from a theory, or replaced with another entity, then it isn't supported by the evidence. For example, the modern day scientific explanation of seasons, which involves the angle at which the earth spins around the sun, can't be easily modified, none of it's components can be easily replaced without demolishing the entire theory. There are no arbitrary postulates used to explain the data. Here's a better way of describing what I mean. There is a difference between a theory that fits the evidence and theory that is consistent with it.

[@CarneadesOfCyrene]

The inherent problem with this solution is that it simply preys on human's propensity to like simple solutions. The only reason that we prefer simple theories over less simple theories is an irrational bias. There is absolutely nothing which says we should keep this ampliative principle over the second theory (beyond another ampliative principle equally subject to scrutiny). I cover this more in the video later in the series on the rationality of science: kzbin.info/www/bejne/bHLZinaKh91qaLM.

[@gdn5001]

I'm not really talking about simplicity so much as I'm talking about fit. Here's a much better example. Electrons are only assigned properties that we have observed. They could in reality be fairies, but there's no evidence for that. We say electrons are particles that make up matter, but don't specify anything beyond that. "Particle" does not mean little ball of energy.

[@emmashalliker6862]

@@gdn5001 “The electron is a theory. But the theory is so good we can almost consider them real.” - Richard P. Feynman

[@PotterSuppositionalist]

I've always thought of scientific theory as most probable explanation available given the body of evidence. It's fair to say that, in all likelihood, chemistry is a superior explanation compared to alchemy. It's more accurate, even if it may be incomplete or fallible.

[@CarneadesOfCyrene]

Potter Suppositionalist The problem as we will see in the series is that we will have trouble distinguishing between scientific theories and non-scientific theories (the problem of demarcation). Or even distinguishing between our scientific theories, and choosing which ones to assent to.

[@PotterSuppositionalist]

Carneades.org I look forward to the series and I agree these are legitimate concerns. I would still contend that the best science is the most probable idea and is always provisional. There is always with a degree of uncertainty and--I'd be the first to admit that--it may not be possible to have an absolutely accurate model of aspects of nature (or its's operations). By my lights, a scientist can be and should be skeptical in general. There is nothing wrong with saying this is idea more likely than that. But science doesn't claim certain knowledge and there is never really a consensus that a theory is true, only that one may be the most accurate and fitting for evidence, currently.

[@CarneadesOfCyrene]

Potter Suppositionalist One of the problems that we'll deal with here is that there is no such thing as "the best science" or the most probable idea. Claiming one theory is correct over another when the evidence is underdetermined, is just irrational.

[@PotterSuppositionalist]

Carneades.org We may never know if a theory is 'correct,' as you say, but I think we can conclude that one theory is more accurate than another, with the lines of evidence available. If you mean there is no theory which could not be rivaled by a skeptical scenario then yes, but that is a trivial claim. It is unanswerable. However, I have a proclivity to choose those ideas which consistently produce results over those which don't. While I don't claim certainty, I learn towards ideas which are elegant, explanatory and beneficial. I may be skeptical, but I wouldn't claim we can't assign probability values or make an educated guess given our situation. I operate my kitchen under the hypothesis that touching a hot stove will burn my hand, so I try not to do that. I'm skeptical, but perhaps also pragmatic.

[@dewinthemorning]

Thank you, now I have learned a new term in philosophy and in science - underdetermination. There would be, I would think, a lot more underdetermination in the science of biology and the social sciences, than in the "exact" sciences (in the physical sciences)?

[@CarneadesOfCyrene]

***** In the social sciences it is a big problem. But it's also something of a problem for new fields in physics, like quantum physics and cosmology. Originally it was posed as a problem only for physics by Pierre Duhem. But since has been expanded to other fields and all of epistemology by Quine and others. In most cases it depends on particular theories and the specific evidence present.

[@CarneadesOfCyrene]

***** Why it is a problem for these other fields will be explained in the coming videos.

[@Elgeneralsimo69]

"Why it is a problem for these other fields will be explained in the coming videos." +Carneades.org If you've done these vids, could you link to them? I'm particularly interested in your philosophical and skeptical analysis of quantum mechanics and cosmology.

[@MonkkSoori]

I feel there's a parallel here between holistic and contrastive underdetermination, and the beta (type II) and alpha (type I) errors in statistics respectively (or the reverse?), or some kind of analogous relationship.

[@CarneadesOfCyrene]

It seems to me that these do not map onto each other perfectly, though they are all errors based on imperfect systems. Type I and Type II errors are talking about incorrect rejection or acceptance of the null hypothesis. Holistic and Contrastive Underdetermination are talking about assuming a particular conclusion over another when there is no evidence to prefer one over the other. With the statistical errors you might have reason to claim that the error is more likely than a correct assumption. with underdetermination, there is inherently no more evidence on one side than the other.

[@Elgeneralsimo69]

4:40 Fourth option; the correlation is trivial. If all kids are playing video games and all kids are anti-social, then the correlation will exist for no other reason than the same population is engaging in two events, playing video games and being anti-social, for reasons that have nothing to do with each other. 5:00 Or you could say that the only conclusion to be drawn from the correlation... _is that there is a correlation!_ Underdeterminism seems to me to only be a factor in what you intend to predict in the *future* or infer from the *past* and not with validity of the correlation in the *present*. Sidenote on skepticism: I get the feeling you are asking for all theories to seek and be based on absolute, all-encompassing, global notions of truth. Thus, because I can't prove with 100% certainty that every time I drop an object it _will_ fall, then I should _always_ be skeptical of my belief that the next time I drop an object it will fall. But I submit to you that while I can't be 100% certain, I can be somewhat certain that there is convincing evidence that _it is more likely than not_ that it will fall; call this 99% certain or 0.99 belief in reference to your Bayesian Epistemology vids. I feel that the fundamental mistake is in believing that certainty (0% and 100%) in any way, shape, or form can apply to this , *or maybe any*, other situation. Is there any reason why the probability range of 0

[@InventiveHarvest]

While correlation alone is not sufficient to determine causality, there is no reason for science to use it alone. Science can see which came first - the video games or the antisocial behaviour. Science can also control for or statistically remove co-causal factors.

[@CarneadesOfCyrene]

The point of that example is not to in any way dispute these options, merely to show that with a correlation alone, you are underdetermined as to what conclusion you should draw (causal or not). Later videos in the series will cover the deeper problem (a component of contrastive underdetermination) that while we can control for factors that we suspect are co-causal, we can never be confident that we have identified all potentially co-causal factors.

[@InventiveHarvest]

@@CarneadesOfCyrene okay, I will watch the other videos. But for now, I maintain that while science does not provide 100% certainty, we can be fairly confident that we have included the relevant factors. This is often done in the pre-paper phase where we present the study to peers and they often point out factors that might have been missed.

[@fernand04ndrade]

I have some points about your video. First of all, the topic is really nice and important but, form me, I thought some explanations were kind of too fast (I am no logicist). I got lost in some parts of the video. I know this must be trivial for some, but a more slow paced video would reach more people (like me, for instance), I think. My second point is, correct me if I am wrong, that in one of these videos you suggest that you would like science to 'work' and to achieve this it would have to be rational (given your definition). In my opinion, science does not have to be completely logical (in your definition), as the focus of science (empirical world) is completly different from the focus of logic (and math, I guess). Logic, at least the classical one, deals with dedutive reasoning... science cannot do this... Just to be clear, I do think science depends on logic (and math), but these are not the only dimensions of science. Science can not be explained by logic and math alone, in my opinion. Another point is that popperian criterai is one of many criteria (by far the most defended, but not the most used in actual science, as Lakatos wrote about). Finally, I have to say that in some points I do agree with you, naive conceptions of falsibility, verification, realism and so on are all around science, and your video helps us to keep the feet on the ground. Thanks for your videos (many other series are also very good) and for you attention in the discussion

[@Chicken_Little_Syndrome]

Duhem is alluding to what we now call confirmation bias. Duhem is pointing to the fact that Academically-motivated Pavlovian conditioning and groupthink influence how we perceive the world. We tend to ignore data that refutes our favorite theories. And we tend to filter all data so we only see what reinforces our favorite theories. This is why Einstein continues to be proven right. And this is why nobody notices the problems with the work of Newton and Maxwell, for example.

[@Gnomefro]

The main point of falsifiability isn't that one conclusively disproves a hypothesis - it's the methodological commitment on part of the researcher that he has stated in advance that he will consider certain observations a falsification and move on if they turn out to be the case. It avoids all sorts of embarrassing appeals to fairies to save bad hypotheses. While it's certainly possible to falsely throw out a hypothesis, I think it's important to remember that the reason one does that is because the hypothesis under test has miserable predictive power during the test and there's a limit to how long one can entertain ideas that consistently fail to produce useful results.

[@CarneadesOfCyrene]

Gnomefro An interesting argument, but actually not the one that I'm going to be looking at. The question here is whether certain observances will actually disprove a hypothesis or not. I won't spoil the argument, but the problem is that even when our hypotheses fail to make correct predictions, we cannot rationally choose to throw them out. Because it may not actually be the hypothesis that is incorrect, but rather the claim that the experiment will actually test the hypothesis.

[@Chicken_Little_Syndrome]

Duhem was right. His work is worth reading. You will never understand why modern cosmology is such a joke or even what I am referring to unless you honestly and logically audit the relevant data. Duhem correctly criticized the work of Newton and Maxwell. What Duhem alludes to is but the tip of the historical iceberg.

[@rezamahan7109]

I dont believe in God, but if there was one, I would thank him/her for creating you to create such informative information.

[@FrozenSpector]

Sweet! Thanks for picking this topic up.

[@CarneadesOfCyrene]

***** Thanks for the link! It was a great topic to follow up the Debunking series.

[@H.J.G]

I’ll have what he’s having 🙋‍♀️

[@Ioganstone]

Though I believe in evolution, this video aged terribly from the snarky youtube atheist days. There's no way given the phrase "Macro evolution is just a theory" to know whether it was said in response to "Macro evolution and abiogenesis is a fact" or as the video said equivocating with the colloquialism to get around "Macro evolution is important for a theory". Underdetermination.

[@grahamers]

This entire video amounts to nothing more than a straw man. Every example/issue discussed is well-known and accounted for when people publish their work and their peers review it. These issues are actively discussed in the papers doing the analysis. Your video amounts to saying "sometimes we can know exactly what is going on." Of course not, and yet we have used the scientific method to develop many models of the universe that make very accurate and novel predictions. No, we don't know everything, but 1) we acknowledge when we cant make conclusion base don data and 2) we know a lot an we use that knowledge it every single second.

[@TheRealisticNihilist]

If you're going to philosophy of science, ESR FTW.

[@CarneadesOfCyrene]

***** Looks cool. I'll have to put it on my list of series to make.

[@MrGreycoat]

28 dollars on plantains and artichokes - can ONLY be 2 artichokes and 6 plantains, because THAT'S the recipe for a plantain and artichoke smoothie. Uhhhhkkk. :)

[@keystothebox]

I don't see underdetermination as a problem. If something can not be determined it shouldn't be believed.

[@substantivalism6787]

keystothebox So if many theories have the same experimental results and successes, should we then just abandon them all? That's just throwing the baby out with the water and trying to avoid the problem by being obscure.