Thread: Problem of Induction

Originally Posted by Omnis Dei

Absolute truth is unnecessary. Furthermore, it's an obstacle in discovery.

I absolutely agree with this.

Originally Posted by Omnis Dei

And by the way empirical knowledge is an oxymoron

Did you want to elaborate on this? I have come to this conclusion often, I am just wondering your reasoning on the matter.

Can you give an example of empirical non-knowledge and non-empirical knowledge?

I think all knowledge insofar as we can have knowledge at all is ultimately empirical.

I'd agree with that. The point I was making is that empiricism, in it's most simplistic description, is an attitude that if you cannot argue it, it is not true. This does not make it false, it just means that the concepts we argue are only as good as the foundation they rest upon and nothing is true in it's own sake.

However, because ultimately nothing is objectively provable, there is also no such thing as empirical knowledge, if we equate knowledge with truth.

In reality, knowledge is everywhere because learning is constantly occurring. But it's important to remember that the foundation of this knowledge has no absolute foundation to rest upon. Much like a building, we may be able to calculate the foundation of the building but we do not understand the consistency of the soil we build it upon.

Originally Posted by Xei

I was wondering what philosophy of science people here hold, and why.

In particular, what do you think about Hume's problem of induction: that there is no basis for thinking that if certain things are repeatedly seen to act in certain ways, then they will continue to act in certain ways? I think it's clear that science and induction can never be proof, but what is our justification for even suspecting that, for instance, all swans are white, or if you drop an object it will fall to Earth? Does science even imply that we should suspect these things? If not, what use is science?

It's worth noting that 'it has tended to work in the past' is itself an inductive argument.

If something has been proven to be true every time in the past, multiple times, then we can assume it will always be true in the future.
If it is not true for some reason later on, the knowledge of that thing will just change to say "This is usually true, but very, very occasionally it is not true".

So if you are making a decision, you can take the minor exception to the rule in to account. But realise that the possibility of it not being true again is very very small.

That's what makes science useful. Probability finding.

Of course, almost nothing is true 100% of the time. Eventually it will be wrong in some instance. And that anomaly just gets assimilated in to the current scientific knowledge base.

I don't see how there is a problem tbh.

Originally Posted by tommo

If something has been proven to be true every time in the past, multiple times, then we can assume it will always be true in the future.

Why can we assume this? Can't we only assume it's been true every time we've tested it? This being in a very small fraction of time (even if considering the whole lifespan of humanity) could we really ignore that the "laws" we've made for the physical processes of nature could really be trying to describe things that are dynamical, even if so slowly that in a life time they are very accurate and useful?

Originally Posted by tommo

If it is not true for some reason later on, the knowledge of that thing will just change to say "This is usually true, but very, very occasionally it is not true".

That "very occasionally" may very well be an indication that our whole idea of this process we observe is wrong and that a better idea should be sought out. If your talking about just the observation itself being usually true, what if it depends on a dynamical process that up till now has been slow and unnoticeable but can start to accelerate or change more abruptly? You couldn't be sure that the deviation from that usual observation will always stay rare.

Originally Posted by Wayfaerer

Why can we assume this? Can't we only assume it's been true every time we've tested it? This being in a very small fraction of time (even if considering the whole lifespan of humanity) could we really ignore that the "laws" we've made for the physical processes of nature could really be trying to describe things that are dynamical, even if so slowly that in a life time they are very accurate and useful?

That "very occasionally" may very well be an indication that our whole idea of this process we observe is wrong and that a better idea should be sought out.

Hm ok. I think this is coming from the abstract nature of my argument.

Take, for example swans. We know they are almost always white, but black when born.
There may be an occasion where a swan is born brown.
But it doesn't mean that the whole idea of the process is wrong. Because we would know WHY the swans are black, and become white.
Just because occasionally there is a deviation in the data, does not mean it will continue to become more apparent.
And there's no reason it would, because the underlying mechanism is the same; the swans have some sort of pigment the causes the colours or whatever it is).

And it doesn't mean that a new idea or theory should be sought out. It just means that the current one needs to be expanded upon. i.e Sometimes, and very rarely, the pigment mutates and causes them to be born brown.

Originally Posted by Wayfaerer

If your talking about just the observation itself being usually true, what if it depends on a dynamical process that up till now has been slow and unnoticeable but can start to accelerate or change more abruptly? You couldn't be sure that the deviation from that usual observation will always stay rare.

No, of course not. But for the foreseeable future, it will be the same. So it is still incredibly useful for making decisions.

Originally Posted by tommo

Hm ok. I think this is coming from the abstract nature of my argument.

Take, for example swans. We know they are almost always white, but black when born.
There may be an occasion where a swan is born brown.
But it doesn't mean that the whole idea of the process is wrong. Because we would know WHY the swans are black, and become white.
Just because occasionally there is a deviation in the data, does not mean it will continue to become more apparent.
And there's no reason it would, because the underlying mechanism is the same; the swans have some sort of pigment the causes the colours or whatever it is).

And it doesn't mean that a new idea or theory should be sought out. It just means that the current one needs to be expanded upon. i.e Sometimes, and very rarely, the pigment mutates and causes them to be born brown.

This observation rests on a dynamical process that changes dramatically with time, natural selection. If some environmental factor starts killing off white swans while brown swans stay camouflaged or something, swans will normally be brown while some mutations may be blue and purple, or eventually turn into a completely different species that climbs trees or something lol

Originally Posted by tommo

No, of course not. But for the foreseeable future, it will be the same. So it is still incredibly useful for making decisions.

It's our best guess, the best we can possibly do, and that's incredibly important and awesome, but as far as how well it will work in the future I don't think we can say. The process will still go on and we will keep making our best guesses based on anything that comes our way, even if it is completely unexpected.

Originally Posted by Wayfaerer

This observation rests on a dynamical process that changes dramatically with time, natural selection. If some environmental factor starts killing off white swans while brown swans stay camouflaged or something, swans will normally be brown while some mutations may be blue and purple, or eventually turn into a completely different species that climbs trees or something lol

Yes, evolution takes longer than our lifetime though. And if it started changing, we'd observe that, figure out how fast it's changing and determine at what point all swans would be brown. And probably keep a few white ones in a zoo

Originally Posted by Wayfaerer

Why can we assume this? Can't we only assume it's been true every time we've tested it? This being in a very small fraction of time (even if considering the whole lifespan of humanity) could we really ignore that the "laws" we've made for the physical processes of nature could really be trying to describe things that are dynamical, even if so slowly that in a life time they are very accurate and useful?


That "very occasionally" may very well be an indication that our whole idea of this process we observe is wrong and that a better idea should be sought out. If your talking about just the observation itself being usually true, what if it depends on a dynamical process that up till now has been slow and unnoticeable but can start to accelerate or change more abruptly? You couldn't be sure that the deviation from that usual observation will always stay rare.

I don't really think I addressed the problem of induction specifically yet, more so just the goal of science.

I think it is only logical that if something if observed to happen an exact way 10 million times, like the force of gravity on an object, It would make sense to form laws and assume they will behave in a certain way to progress in mathematics, science, and many other outlets of understanding the core foundation of physics. Just as it would make sense that if a cure for a disease has worked 10 million times, it would be smart to administer it again.

Now to the problem. Can we assume that this is objectively sound and is not subject to change, that we have arrived at an immutable and perfect law. What would be our basis, if there is one?
Our basis is a large set trials that have occurred with a substantial amount of variances that would suggest that this law hold true. it happened to a bowling ball, to a airplane, gravity would seem to be a justified law.

It seems as though the answer to this problem of induction is that we just need to progress with observable laws and hope they hold consistently to a variety of tests that we will put them through. When variability, if any, happens then we simply adjust and move on. Therefore, sure, maybe there isn't a 100% or even a 1% chance that gravity is immutable and perfectly understood knowledge because the probability of all this being chance or influenced by some lurking variable still persists, but what is certain is that if we can't assume consistency in phenomena, we wouldn't have progressed to where we are now.

Originally Posted by Dreams4free

but what is certain is that if we can't assume consistency in phenomena, we wouldn't have progressed to where we are now.

I guess I just think "assume" is a strong word. We work off of what we can, and it has worked, but how can we "assume" things will forever stay constant?

Many people responding to this thread don't seem to understand what the problem of induction is.

The problem is that, there is no reasonable, non-circular answer that can be given to the following question:
Why can we be reasonably sure that, given that something has happened many times in the past, it will happen in the future?

Seriously, try to answer the question. I guarantee that cannot come up with an answer that doesn't assume that "given that something has happened many times in the past, we can be reasonably sure that it will happen in the future."

You can say that you don't care, because it's such an abstract philosophical question and we're incapable of taking its implications seriously. And that's fine. But it's still true, that we have no valid reason to believe that induction should work. At least recognize that fact, before expressing that you don't care because it's irrelevant to your life.

It makes no sense Dianeva. Why WOULDN'T it happen if it has happened in the past consistently?

Sure it may not ALWAYS happen forever, but most of the time it will. Especially when you get down to things which aren't organism and don't change as much.
If there is no reason for something to change, you can be reasonably sure that it will be the same in the future. Like photons having no mass for example.

Originally Posted by tommo

It makes no sense Dianeva. Why WOULDN'T it happen if it has happened in the past consistently?

Every day for the 10 years you've lived next to him, your neighbor starts his car at 6:35 AM, lets it warm up for 5 minutes, then drives away. You know because his car is LOUD.

Will he do it tomorrow?

Originally Posted by tommo

It makes no sense Dianeva. Why WOULDN'T it happen if it has happened in the past consistently?

Because things change?

Sure it may not ALWAYS happen forever, but most of the time it will. Especially when you get down to things which aren't organism and don't change as much.
If there is no reason for something to change, you can be reasonably sure that it will be the same in the future. Like photons having no mass for example.

But how can you be sure things "that don't change very much" won't start changing very much for reasons we don't know, or could be theoretically possible? Photons having no mass is an assumption based on special relativity, which works for us, but could be subject to change. Maybe the whole nature of light will turn out to be something we've never imagined before. We aren't even experimentally sure that the photon has no mass.

Originally Posted by Wayfaerer

But how can you be sure things "that don't change very much" won't start changing very much for reasons we don't know, or could be theoretically possible?

You can never be sure of anything, unless you have "faith" of course; anything can only be on a scale probability.

We can be fairly sure something would continue to do the same thing, because there is no reason to expect that it wouldn't stay the same. There is no data suggesting it would change at any time. No forces we can figure out which would cause it to change.

Originally Posted by Wayfaerer

Photons having no mass is an assumption based on special relativity, which works for us, but is subject to change. We aren't even experimentally sure that the photon has no mass.

It was an example. You don't argue with examples.

Originally Posted by tommo

We can be fairly sure something would continue to do the same thing, because there is no reason to expect that it wouldn't stay the same. There is no data suggesting it would change at any time. No forces we can figure out which would cause it to change.

If something keeps happening, it must have some cause, and we have no reason to think that that cause will change. But, we also have no reason to think that that cause should stay the same, without assuming that induction is valid.

Originally Posted by Dianeva

If something keeps happening, it must have some cause, and we have no reason to think that that cause will change. But, we also have no reason to think that that cause should stay the same, without assuming that induction is valid.

Oh.... I get what you're saying now. The problem is that you're using a method of reasoning to describe an actual process.

Originally Posted by Dianeva

If something keeps happening, it must have some cause, and we have no reason to think that that cause will change. But, we also have no reason to think that that cause should stay the same, without assuming that induction is valid because something has always happened, it will always keep happening.

I still don't see how this matters practically though.

Originally Posted by Wayfaerer

Sorry if you think I was arguing, I was just addressing one of your points which happened to be an example lol

haha, no, no problem.

Originally Posted by tommo

You can never be sure of anything, unless you have "faith" of course; anything can only be on a scale probability.

We can be fairly sure something would continue to do the same thing, because there is no reason to expect that it wouldn't stay the same. There is no data suggesting it would change at any time. No forces we can figure out which would cause it to change.

I'm not sure, I think there is some theoretical work based on the possibility that the constants of nature are dynamical. If that's possible, it doesn't seem to far of a stretch that the change could accelerate at any time, since we really can't be sure about the nature of the universe, only give our best guess. I think we can only hope things will stay the same for convenience.

Originally Posted by tommo

It was an example. You don't argue with examples.

Sorry if you think I was arguing, I was just addressing one of your points which happened to be an example lol

Originally Posted by tommo

It makes no sense Dianeva. Why WOULDN'T it happen if it has happened in the past consistently?

Answering 'why' with 'well why not?' won't do, lol.

I don't think it matters practically.