Philosophy of Science 3 - Hume's Problem of Induction

We have noted previously that science appears to be built on induction. David Hume famously argued that induction is not rationally justified: inductive arguments provide no reason whatsoever to believe their conclusions. If science is inductive, then Hume's argument seems to show that the foundations of science are irrational. This video explains Hume's argument and considers a few possible responses.

[maxmax]

Reichenbach's pragmatic justification makes a lot of sense to me. Repeated observations of uniformity (or perhaps even only a single one) are out best candidate for ascertaining causality. And causality, - the relation wherein certain antecedent conditions are understood to bring about consistent necessary consequent states of affairs, - is what induction has always been grasping at descriptively. To make an inductive argument such as "we have seen the sun rise every morning, therefore the sun will rise the next morning, " is precisely to make a causal attribution between elements we observe to be correlated spatio-temporally. This, of course, assumes we even have a concept of causality. But if Kant was at all correct in his response to Hume, we in fact _do_ have a concept of causality, and we can apply this concept to empirical observations in the world.

[rezamahan]

I wish there was a double-like function on YouTube!

[cptblood]

Awesome lecture, fam! Very interesting! Gonna watch all your videos:)

[daswassup]

Nice video, however i genuinely wanted to ask if why an inductive reasoning that rely on a regularity still rationally unjustified like the one exampled on the vid that u observed in the past the fires are hot therefore the next fire is hot, fire being hot has its well established scientific principle

what is think is rationally unjustified is if you use your own experiences to account for the future events. for example i observed that if i ate this certain food ill have an upset stomach
therefore if i will eat this certain food, therefore ill have an upset stomach. because the argument is solely based on individual experiences rather than a broader regularity or scientific principle

what do you think guys?

[Ansatz]

The problem of induction is an excellent illustration of why science would be foolish to depend upon induction. Even if the world were sometimes regular, we have no way to discriminate between actual regularities and mere apparent regularities, so attempting to apply induction in science would be like investigating the world by interrogating a Magic 8 Ball. Science based on induction would be a process of introspection that only reveals how we feel about various apparent regularities rather than revealing anything about the real world.

The pragmatic objection applies to the application of scientific results, where we are faced with forced decisions. For example, do I put my hand in the fire or not? We can't refrain from choosing so we use whatever we can get to help us make a good choice. In that case our success depends on us choosing the correct apparent regularity on which to base our induction. Most people don't have experience putting their hands in fire, so they can't use that as a base for induction, so they have to go by what they have heard from other sources.

Person A might read science textbooks and learn about biology and fire. For A the induction would be that every prediction the textbooks make has been true so far, so the prediction the textbooks make about putting a hand into a fire will also be true.

Person B might talk to telephone psychics and be very impressed by their readings. Suppose when asked about putting a hand in fire the psychic says it's a good idea. For B the induction would be that everything said by the psychic has been true so far, so the psychic is probably right about the fire.

There's no difference between A's induction and B's induction aside from their choice of apparent regularity. Circumstances forced them to choose, but obviously it's not the sort of reasoning we should use when we're not being forced.

One of the reasons why we should seriously consider using scientific results as our apparent regularity is that science does _not_ make use of induction to reach its conclusions. When science is being done well, it doesn't involve trusting that the past will be like the future; such trust is too easily betrayed. Instead, successful science is always suspicious of every apparent regularity, always testing and retesting every conclusion.

First we come up with a hypothesis that seems to be true in our experience. If we were using induction we might stop there since we already have an apparent regularity just from our past experience. Instead we choose to not trust that induction and conduct an experiment to test the hypothesis.

Suppose the hypothesis passes the test. If we were using induction we might stop there since the regularity is even more apparent. Instead, we publish our results for other scientists to examine. They review our work so if we made a mistake in our experiment, that mistake might be revealed. After that, scientists all over the world repeatedly attempt to reproduce our experiment. At no point does good science ever become complacent and trust that we have found the real regularities of nature; it's a never ending process of rigorous testing.

It is because science refuses to trust that science is exactly what we should trust when we're forced to make a choice.

[dmartin]

excellent video thanks. would it be true to say that the approach of bayesianism is really just a particular form of probabilistic inference?

[amrelnashar]

At 4:00 you give an argument about why it is not possible to justify induction using deduction, that's because deduction can only give us 100% certain conclusions while the conclusions of induction are only likely but not certain.

That's where I disagree, the deductive method can also give us probabilistic conclusions, for example the deductive method in a branch of mathematics called probability theory can give us probabilistic conclusions like : "It is 25% likely that when you toss three coins they all give the same face".

In other words, perhaps induction could be justified using probability theory . What do you think ?

[adamwisniewski]

no good solutions in my opinion. but i dont know if there are any.

[jimmyfaulkner]

Does this reveal that philosophy of science is important for science as science cannot solve the problem of induction and this problem can only be tackled using philosophical inquiry and reasoning?

[MatthewMcVeagh]

Surely the issue with induction is that it doesn't necessarily produce truth? This was certainly Popper's point, and it was implicit in the discussion of how a hitherto uniform universe might not be uniform tomorrow - it doesn't matter how many times induction has 'worked' as in predicted a correct result, it does not do so necessarily and therefore cannot be relied on to produce truth. A scientific theory 'proved' by induction could be applied in technology that then fails in some situation because the theory is not overall correct, there had just only so far been observations that corroborated thee theory.

[Lembo]

I am okay that science asks that any claim that wants to be touted as a law or truth is expected to do so to 6+ sigma levels of confidence. It concedes to Hume that maybe all the observations are indeed a 1 in a million fluke that makes the false "law" seem true, but we'll go with it until the incredible streak of luck ends and we have to change our conclusion.

[darrenparis]

Why give equal footing to assuming that something will change and to its remaining the same? If /everything/ was always changing, there would be no use in attempting to understand anything at all because to understand something is to pin it in our minds and our records as having static properties.

And if you accept classical physics, it is a violation of the laws of physics that something should not either remain the same or in a same cyclical state without interaction with a foreign body. Here, to assert that change is possible is to assert the existence of an external object causing the change, which is violating Occam's razor.

[tylerhulsey]

ngl thinking about the problem of induction makes my head hurt 🤕🙃

[humeanrgmnt]

Scientists just don't want to hear it; like birds they prattle about the trees trying to figure out how the world works using inductive reasoning.