"Natural Laws" and their domain of validity

madness

What is the philosophical justification for extending observed phenomena to "laws of nature". For example, Galileo dropped massive objects and saw that they fall at the same rate, but to then say that all massive objects fall at the same rate requires a leap of faith. Similarly, we used to believe that the laws of nature here on Earth had nothing to say about the workings of the heavens, but now we think differently.
Clearly we need to be careful not to overly generalise our observations outwith their range of validity. On the other hand, if we reject inductive reasoning altogether then we can't even accept the most limited laws of nature. It seems some kind of middle ground is necessary, presumably there is no clear demarcation here.

zomgwtf

Uh, no... it wouldn't require a leap of faith to say that objects fall at the same rate. A leap of faith means specifically 'believing or accepting as true something that is intangible or unprovable'. AKA Without empirical evidence.

Objects falling at the same rate can be concluded in a very simple experiment. (Dropping things with different mass from the same height... go right ahead try it!)

Indeed we might have believed that laws of nature were different from the laws of the sky or whatever 'heavens' implies. THAT is a leap of faith, because it was unprovable at the time. However as time went on and on a paradigm shift occurs. Science has come to terms with this and most scientists do NOT give opinions or 'theories' of things which are 'unprovable' (requiring a leap of faith). A main component of scientific theories is falsifiability...

So in conclusion: Scientist don't need to 'seek a middle ground' they need to go all out scientifically only on those things which they can apply scientific method to. (Which, in my opinion, they do.)

madness

No. Unless you test every possible mass comprised of every possible type of material at every possible location etc. you can't say for sure that it will generalise. Even then, you can't prove that a past experiment (no matter how many times you repeated it) will behave the same way in future unless you use inductive reasoning. Please don't respond if you don't at least have some basic understanding of the philosophy of science.

zomgwtf

The point of science is not to 'prove' anything for exactly the reason you stated. Science does not deal in the realm of proofs.

The fact that science uses inductive reason does not mean it invokes a 'leap of faith'. The change of scientific views of the world is known as a paradigm shift (funny how you respond with the snark remark of me not understanding basic philosophy of science )

What a leap of faith means is this:

When there is something that is 'unbelievable' yet a person 'wants to believe' in it they make a decision in their mind to put FAITH in its validity. This means it goes from non-belief ---> belief through FAITH that it is true.

In my previous example I should have stated that the person who knows we shouldn't make opinions on the nature of the 'heavens' will have to make a leap of faith. Because they 'don't believe we should comment on the nature of heaven' but they would claim as truth 'the natural laws of Earth do not apply in heaven'. This requires a leap of faith.

There is no need to limit inductive reasoning in the scope of science, that's rediculous. (If my reading of what you wrote is correct) Instead we should limit the scope of science to deal with only things that are able to fall within the scope of science! That's a given though.
So again, inductive reasoning =/= leap of faith. For two reasons:
A. It's not assigning an objective truth value to anything
and
B. It is based on emperical evidence AND logic, both of which are not present for a leap of faith.

madness

Why does inductive reasoning not fall into this category? And my comment was referring to the fact that you seem to have no knowledge of Hume's problem of induction - which is basically what this thread is about.

Inductive reasoning is not based on empirical (spelling) evidence or logic and assigning a truth value is exactly what it does. I already gave the need to limit inductive reasoning, i.e. that we don't want to extend a law outside its range of validity.

zomgwtf

The problem of induction does not talk about any such leaps of faith. It talks about KNOWLEDGE (objective truth value) of inductive claims.

Can knowledge be attained through an inductive statement? No, not if you use an objective truth value standard. That's a given and is accepted in science.

Someone like Popper would argue however that it does lead to knowledge however I don't believe that to be true. Induction, in my opinion, is not a problem since no scientist, or person with understanding of science, in their right mind would argue that science has an objective truth value. The term knowledge however can be debated to determine if scientific reasoning can lead to knowledge.

There is no doubt in my mind that science is an extremely usefull tool for humans without which we would still be living in caves hunting with our bare hands.

Please go and look up inductive reasoning, this is the very reason it is a different word from 'deductive reasoning'.

madness

You are aware that inductive reasoning is ubiquitous in science, right? The problem of induction states that knowledge cannot be obtained by inductive claims - this is why it's a problem. Therefore the problem of induction does talk about leaps of faith. And Popper didn't believe induction leads to knowledge that's why he tried to get rid of it in his philosophy of science. The point is we want to use science as a means to obtaining knowledge.

I don't doubt that it is. The problem is defining science and being precise about what types of reasoning are acceptable within the scientific method.



"Inductive reasoning, also known as induction or inductive logic, is a kind of reasoning that allows for the possibility that the conclusion is false even where all of the premises are true.[1] The premises of an inductive logical argument indicate some degree of support (inductive probability) for the conclusion but do not entail it; i.e. they do not ensure its truth. Induction is employed, for example, in the following argument:

All of the ice we have examined so far is cold.
Therefore, all ice is cold." from wiki.

What's your point? This is just further evidence that we need a middle ground for inductive reasoning in science.

apeiron

Zomgwtf has it right. Science is about modelling - formal statements that make predictions that can be measured. And those formal statements are developed first by abduction (a smart guess), then refined via the dual and mutual process of induction~deduction.

Induction is the move from particulars to generals - from observations of specific or local instances to the generalisation of universal "truths". And deduction is about going the other way, from the assumption of some global or universal truth to a prediction about what we are likely to find at a specific location.

This is how minds themselves work. Our perceptions of the world are formed by the interaction of ideas and impressions. Generals and particulars. If I see a cat, I am both aware of its general cat-ness (and would be surprised if it barked or squeaked) and also its specific cat features, whether it is a black cat, a skinny cat, etc.

As children, many individual experiences of cats (and other animals) builds up (induces) a crisp idea of cat-ness. Then this context, this generalisation, is the framework for perceptual deduction - I am probably seeing a cat because of this arrangement of light and shade (whoops, it was just a quirky shadow on a twilight wall).

The middle ground for induction~deduction is the zone of productive interaction. It is where the top-down and the bottom-up actions are mixing in a fruitful, adaptive, fashion. The more cats (and non-cats) I see, the more I fine-tune by knowledge (internal model) of cats.

Science too depends on the exact same fruitful balance. The more detailed instances we can observe, the more extensively we can generalise.

Of course, you are talking about how far can we safely go when we are trying to get beyond the easily measurable.

This would be like saying, well, we have a pretty good mental model of cats. But what about "cats" on other planets? What can we say about alien lifeforms?

Here we cannot (yet) make any measurements. We cannot employ the inductive path. We instead have to rely on deduction or generalisation. And actually, when it comes to attempts to imagine alien alternatives, we do a pretty poor job (although perhaps that is just because carbon, water and something like DNA are the only realistic ways to go).

Anyway, science is not "all induction" and so suffers a problem about its demarcation. Science (like all modelling, including neurological) is about the interaction between bottom-up impressions and top-down ideas. Induction and deduction are synergistic partners.

Progress is achieved by moving the two scales of action ever further apart.

In the beginning, when our notions are vague, as in the infant brain, we have to make abductive guesses about the world. The inductions and the deductions, the ideas and the impressions, are all still lumped together at the same scale. Babies will point at a cat and say confidently "doggie". A sort of correct beginner's generalisation.

But once our models become well developed, the generalisations become as broad as our measurements of the world are fine. The bigger our ideas, the smaller the details, the fine discriminations, we must be noticing.

Which is why science depends on refinements in experimental technique. The zone of knowlege is built out of the interaction of ideas and impressions. To increase the size of this zone, we must push out in both directions equally - towards more sweeping theories based on more precise measurements. The two ends must be in active touch with each other to be able to continue to expand.

madness

Sounds a bit vague. Too much induction and you will make erroneous assertions, not enough and you can't make any progress. How do you define the zone of productive interaction? In any case, using induction based on it's productivity requires inductive reasoning- you use induction because it "works". You can't justify induction using induction.

I don't see a clear demarcation between the induction we use say to extend Galileo's experiments to all masses and the stronger inductions like aliens. I see a continuum. The original point in the thread was to ask how much induction should be used in science and how its use is justified.

apeiron

More "induction" would be not measuring more of the same but more of the different. Confirmation does not change the state of your knowledge much. We would model this fact concretely as the law of diminishing returns - asymptotic approach to certainty. If I've measure the same thing five times and got the same result, then I have increased my knowledge arguably. But if I measure it 10 times, I have not doubled my confidence.

Where real change happens is when something novel or unexpected (unmodelled) is observed. Then we repeat observations to try to get more data points.

We then know we have hit the zone of productive interaction when models and measurements are in good statistical agreement. When novelty in measurements can be safely explained away as just measurement error. Of course, we cannot know all outlier measurements to be just flukes, but we can have a pragmatic confidence.

We can measure all masses within our reach and the relation seems firm. But yes, things could be different over the observational event horizon and we would never "know".

So our deductions cannot outpace our inductions if we are being "scientific" - insisting on the importance of a working interaction between the two aspects of modelling.

The story on mass and on aliens are a little different. For mass, our natural expectation is that mass would be the same elsewhere (though are there theories of g varying with scale, or time). With aliens, our expectation is in fact the opposite - that it would not be like life on earth. And the surprise would be if it was.

Both would be the consequence of deductions based on inductions. A rule about mass has been built up locally by induction and is extended to places beyond our reach by deduction (from the rule, we predict the measurements that would be observed).

The same with the evolution of life - except because life is presumed to be accidental in the paths it takes, the general rule would we should expect aliens to follow some different path. However this is still a rule developed by local observation then extended to other out of reach places via deduction.

JoeDawg

This is incorrect.

Induction moves from the known to the unknown, it does NOT require that one move from particulars to generals.

And, madness, is also incorrect:

The problem of induction is about justification, not about whether we can derive knowledge through induction. David Hume who defined the problem of induction was an empiricist, that is, he believed that ALL knowledge is derived from observation. The problem is not whether you can derive knowledge from induction, you can, the problem is that we have no justification for using induction, which is very different.

apeiron

Time to buy a dictionary Dr Dawg.

madness

I would consider this to be less induction. By "more induction" I meant making stronger assertions and generalisations based on the limited empiricial knowledge we have. This is what I am getting at with this thread - we need to make some unjustified generalisations, but shouldn't make too many.

This relies heavily on induction.

So basically we try all different amounts of induction until one gives a good fit? Of course there is no reason to think this amount of induction will work in a different setting. Again we need induction to justify the use (and amount) of induction here.

Mkorr

One could view "natural laws" as descriptions of reality that becomes more reliable as more scientific evidence accumulates.

JoeDawg

Hahah. While, I'm sure everyone here is as impressed as I am, with your ability to do a google search, and come up with a 'free dictionary' definition, this is not quite the level of understanding I was going for.

Dictionaries are designed to give a general understanding of a word, to a general audience.
As such, they are NOT often referenced in high level philosophical discussions, but more often when the discussion has degraded to the level of a 'definition war'.
While this happens quite frequently... on the internet... it's never that productive, and usually implies stubborn ignorance on the part of the person providing the definition.

Inductive reasoning can be as simple as:
My apple is red, therefore your apple is red.

Or... more complex....
These ten apples are red, therefore all apples are red.

Or....
I have 20 apples in a bag, I have removed 19 red apples, therefore the next apple I remove is red.

It actually can involve going from:

particular -> general
particular -> general -> particular
general -> particular
particular -> particular

The important part therefore, is NOT general/particular. It is making an inference from a known sample to an unknown sample. This is one of the problems with constantly trying to fit everything into overly simple dichotomies, you end up losing more complex ideas.

In case you were wondering, some people get confused when it comes to inductive inferences from general to particular, as this seems more like deduction.

But there is an important difference. Deduction starts with defined quantities, induction starts with observed quantities.

apeiron

Impress us by providing an inductive argument that actually proceeds from the general to the particular then.

Your apples example relies on tacit generalisations and hence deductions by way of abductive reasoning. ie: If we both have the same thing (an apple) then our best guess is that the apples share the same properties (like red).

Turn it round and ask the question: you have a red apple in your hand, someone else also has something red in their hands. How secure are you in generalising to argue that it too is going to be an apple?

You can see how your example invokes a tacit knowledge of the world, a knowledge of apples and their properties, and is not a step from the known to the unknown but instead a pincer movement making use of the available information - some of it local (what you hold in your hand), some of it global (apples are restricted in colour, colours are not so restricted in the hand-held objects they may represent).

This pincer movement is what abduction is all about. And the reason why the proper definition of induction and deduction (as held to be basic in any dictionary much to your discomfort) is about the dichotomy of particular~general.

I can understand what you mean in saying that deduction starts from defined quantities, I just don't accept it as a psychological reality. It is a hangover statement from the rationalist tradition and so a self-delusion about how knowledge is actually derived about the world.

I stick with the psychological realism of Peirce. He got it right. But maybe that is the subject for a "higher" level philosophical discussion once you've had a chance to read up on Peirce and abduction, vagueness, etc.

JoeDawg

Can I use google?
Wrong. Abduction is just inverted Deduction. They both rely on premises and necessity, they just move in different directions. They are definitional, not observational.

With induction, I don't have to know what an apple is. With induction all I'm doing is making an assumption that the unobserved apple will resemble the observed one. Its not a guess, its an assumption. We then base probability on this assumption.

One can use any criteria one chooses to make a best guess.
Abduction can be used together with both Deduction and Induction, but the fact remains Induction works differently.
If you are using Induction, there is no 'security', the problem of induction makes it clear, there is no justification for induction. With Abduction, you can use induction to help justify your guess.
This, invariably is where you run into problems.
Oh, I googled him, but I'm not sure what that James Bond guy has to do with philosophy... maybe your talent for google searches is just superior to mine.