View Full Version : From Godel & Tarski to the Scientific Method


Cybermorphic
03-17-03, 02:56 PM
Although I probably do not understand the math behind Godel any better than someone else who has read GEB, I feel that looking at the theorem even from a level more basic than TNT has led me to consider new implications from the Incompleteness Theorem on our basic system of reasoning; the Scientific Method. I will explore my considerations of both Godel's first Incompleteness Theorem, and Tarski's Undefinability of Truth Theorem using only English to express their important meanings.
When either Godel's theorem or Tarski's are expressed in English one is tempted to ignore them saying things like, "This is meaningless". If it is possible one should keep in mind that these theorems can both be expressed in mathematics so it is not a question of meaning but a question of what the implications of these theorems are. Tarski's theorem is not as well known as Godel's theorem and it is definitely not as understood. Tarski's theorem is constructed in the same fundamental way as Godel's theorem but with more startling consequences.
I will briefly go over Godel's First Incompleteness Theorem. Godel proved inside of the system Principa Mathematical which was designed not to allow self-reference which could lead to paradox that a self-referential statement could prove that Principa Mathmatica was incomplete, and also that all other formal systems complex enough to do simple arithmetic were incomplete. He did this by creating something that when translated from math looks like, "This is unprovable inside of Principa Mathmatica.". If you could prove that you would of course also prove that you can't prove it, and if you can't prove it then it is true but unprovable. Because of this we can say Principa Mathmatica is incomplete.
Tarski's Undefinability of Truth Theorem looks just the same as Godel's sentence but we replace 'unprovable' with 'not true'. It looks like; "This is not a true sentence of number theory". If that sentence were true it would be false, and if it were false we would see that it was true. To avoid this contradiction Tarski tells us we can not define what truth is inside of number theory.
Tarski's Theorem has implications which need to be considered. If we can not define what truth is in math I have to ask myself if we can define what truth is anywhere. Certainly we can not define truth in English because I have just explained his proof with English. Tarski spent the rest of his life after creating this theorem trying to find a way to define truth with other formal systems of his own imagination. What I consider concluding from studying this work is that any system of reasoning capable of referring to itself and defining truth can not use a definition of truth without contradictions. The scientific method is a system of reasoning that I am considering as meeting these conditions. The method, which is to first make a hypothesis and then test the hypothesis and finally make a conclusion based on the results of the test seems to fall appart when you test a Tarski sentence. Try this hypothesis, "This is not a true sentence when tested with the scientific method". To create the scientific test we only need to adopt a logical process of making an assumption about the hypothesis and following it through to see if we reach a contradiction. Sense we are using the scientific method the sentence contradictory when we have a definition of what truth is, or when we allow for testing the scientific method itself. So we might conclude that either the scientific method can make no conclusions or that the scientific method can not be used to test the scientific method without contradicting itself. Because we assume the scientific method does make conclusions, then we are left to consider that it may not be able to study itself without contradictions. I am considering that sense the scientific method may not be able to test itself without contradictions there is another method that may be able to test the scientific method and prove just as use full. A Cybermorphic Method.
I await your responses.

HallsofIvy
03-18-03, 08:07 AM
Nothing in your post says anything about the scientific method. I do not know what you mean by "test the scientific method" - the scientific method is tested by its results.

In any case it wouldn't make sense to use the scientific method to "test the scientific method". Whatever you mean by that it is not a scientific question.

Cybermorphic
03-18-03, 10:39 PM
That is exactly my point. The scientific method can not be used to study the scientific method itself. You agree with me, but for your own reasons which you do not explain. You also say that the scientific method is tested by its results. This sounds a lot like your saying that the scientific method can study itself, you just left out that a hypothesis is made first about the scientific method. If your openion is that it is not a scientific question then I suggest that you stick to your guns and try to understand my post before you make any final conclusions.
Here is the outline:
(1) We make a hypothesis that the scientific method can not be used to study the scientific method.
(2) We create a test of the scientific method. We could try a whole number of tests about the scientific method. We can try "The scientific method can not prove this." instead of my last suggestion.
(3) We conclude that if the scientific method can prove that then it is contradictory and if it can't prove that then it is incomplete. So in this example which perfectly illistrates godel the final conclusion is that this test does not prove that the scientific method can not study itself but is incomplete to fully study itself.

It isn't too hard to understand my point but I hope this makes things clear. If you don't feel this is scientific then please try and explain your reasoning so that I my decide if I agree or disagree with you.

HallsofIvy
03-24-03, 09:41 AM
Since all scientists agree that "The scientific method cannot prove the scientific method" and that, indeed, the scientific method cannot "prove" anything (it is designed to disprove, not to prove), I don't see what you are arguing about.

Canute
03-24-03, 06:12 PM
I think you have expressed an important truth, although I see it a little differently.

You say that "that the scientific method can not study itself but is incomplete to fully study itself".

As I understand Godel this is not quite completely true. It is perfectly capable of completely studying itself, verifying its truths, as long as it not complete. It is only when it is complete that Godel finally insists that there must be a piece of illogic in it somewhere. In other we cannot have a complete and completely true scientific explanation of everything since it would have to be in some respect not completely true. The implications of this rattle our paradigm, since the same illogic must apply to the thing we are explaining.

HallsofIvy
03-24-03, 10:40 PM
You say that "that the scientific method can not study itself but is incomplete to fully study itself".

Please do attribute things to me that I did not say. That doesn't even make sense. If you are going to quote, please copy and paste so that you get it right.

Goedel did not say anything about the scientific method. His theorems were entirely about axiomatic systems and science is not an axiomatic system.

Canute
03-25-03, 11:35 AM
Originally posted by HallsofIvy
Please do attribute things to me that I did not say. That doesn't even make sense. If you are going to quote, please copy and paste so that you get it right.

Goedel did not say anything about the scientific method. His theorems were entirely about axiomatic systems and science is not an axiomatic system.
Pardon me. I forget to make clear that I was talking to cybermorphic. It was him/her that said it.

I would say that all scientific explanations are made within, and in the terms of, axiomatic logical/conceptual/mathematical systems. In fact I can't see how it could be otherwise. I agree though that it would be woolly to say that 'science' itself is such a system.

HallsofIvy
03-25-03, 12:54 PM
Ah, Okay.

I disagree that "all scientific explanations are made within, and in the terms of, axiomatic logical/conceptual/mathematical systems"

Not all "explanations" are or have to be in terms of "axiomatic logical/conceptual/mathematical systems".

Certainly many are but those making such explanations have to be aware that they are only approximations. Axiomatic systems are based on "undefined terms" and using such a system in science involves choosing one so that assigning specific meaning to the undefined terms results in the axioms being verifiably true. As soon as that involves measurement (which is always approximate), we can, at best, say that the given axiomatic systems is approximately correct for that use.

Canute
03-25-03, 08:05 PM
Originally posted by HallsofIvy
Certainly many are but those making such explanations have to be aware that they are only approximations. Axiomatic systems are based on "undefined terms" and using such a system in science involves choosing one so that assigning specific meaning to the undefined terms results in the axioms being verifiably true. As soon as that involves measurement (which is always approximate), we can, at best, say that the given axiomatic systems is approximately correct for that use.
Theoretical physics seems to be a very good example a system of undefined terms trying to put specific meaning to them. Isn't language itself another example? Isn't any system that relates terms to each other in a logical way an example?