The scientific mechanism for accepting one theory over another.

[wasteofo2]

Are there any sort of essays or generally accepted criteria for accepting one theory over another? I ask, becuase for most any theory, it seems that you can find evidence for it. If you did your research, you could come up with tons of evidence for a totally wrong theory (like all the evidence that existed for the geocentric model of the solar system).

It seems to me, that the only scientific mechanism for accepting a theory as valid would be to compare and contrast it with as many alternate theories as possible. Test all the theories against each other, and see which is the most consistant.

I mean, that makes sense to me, but that's just what I think. Is there any sort of literature backing my assumptions?

 

[quantumdude]

There isn't just one scientific method. But there are two broad catagories of methods by which theories are ruled out.

1.) Falsification by experiment.
2.) Internal contradiction.



Method 1 is based on the logically valid syllogism known as modus tollens which says, "If p is true, then q is true. q is not true. Therefore, p is not true."

More formally:

$p \longrightarrow q$
$\neg q$

$\therefore \neg p$

This model would be called naive falsificationism, and has been criticized by a number of philosophers of science. But in some instances falsifying evidence is so overwhelming that such objections can be safely ignored.

The rise of quantum mechanics was brought about largely by Method 1. In one relatively short period of time it was found experimentally that there were several phenomena that could not in any way be accounted for by classical physics. For instance look up the photoelectric effect, the stability of the atom, and the ultraviolet catastrophe.

Method 2 is used when it is found that a theory is logically inconsistent. If it is possible to start from the axioms of a theory and, for some statement $p$, derive both $p$ and $\neg p$, then the theory is rejected. In logic this is known as a violation of the law of the excluded middle.

This method is in large part responsible for the birth of special relativity. As the 1800s were coming to a close classical mechanics and classical electrodynamics were the prevailing theories of the day. If in addition to these theories one accepts the idea that the laws of physics should be the same for all inertial observers then one encounters an apparent contrdiction. For the laws of classical mechanics are left invariant under one set of coordinate transformations (known as the Galilean transformation), and classical electrodynamics is left invariant under a completely different set of coodinate transformations (known as the Lorentz transformation, of Special Relativity). Relativity resolved the contradiction by proposing a modified version of mechanics, of which classical mechanics is a limiting case.

Are there any sort of essays or generally accepted criteria for accepting one theory over another?

The entire discipline known as "Philosophy of Science" is devoted to this. Philosophers of science will readily admit that nothing in their field is set in stone.

I ask, becuase for most any theory, it seems that you can find evidence for it. If you did your research, you could come up with tons of evidence for a totally wrong theory (like all the evidence that existed for the geocentric model of the solar system).

You are touching on a basic issue of the philosophy of science here: the inadequacy of confirmation theory. We gain confidence in theories not so much for withstanding many tests to confirm it, but rather for withstanding many tests to falsify it.

And when scientists do accept a new theory over an old one, it is expected that the new theory will be confirmed in a larger experimental domain than the old one, not a smaller domain. For example it is now recognized that relativistic mechanics is superior to nonrelativistic mechanics, and one reason is that the former is valid over a larger range of speeds than the latter. Using this criterion it is impossible for the reverse process to happen, even in principle. In other words, if relativistic mechanics had been discovered first, it could never have been supplanted by nonrelativistic mechanics, if we were to adhere to the principle enunciated above.

We don't look for theories that account for some observational evidence, we look for theories that account for all of it.

 

[HallsofIvy]

It seems to me, that the only scientific mechanism for accepting a theory as valid would be to compare and contrast it with as many alternate theories as possible. Test all the theories against each other, and see which is the most consistant.

No, that's not science, that's mathematics!

"Science", by definition, relies upon the "scientific method" and Tom Mattson described that very nicely: falsification by experiment. The scientific method is
1) Observe and experiment to see what is actually happening
2) Construct as many theories of how the things you observe are connected as possible
3) Find deductions from your different theories that would give different results to experiments
4) Do those experiments to see which theories you can discard.

Notice that experiments will disprove a theory- they can never prove a theory. Also note that the standard for acceptance of a theory is not "consistency" but how the theory corresponds, according to experiment, with "reality". The philosophy underlying any science is "realism".

With mathematics, it is exactly the opposite. We do not do experiments to see which mathematical theories correspond to experiment, we construct logical proofs from axioms to see which theorems are consistent with the the axioms- the standard for acceptance is consistency. The underlying philosophy for any mathematics is necessarily "idealism".

 

[selfAdjoint]

Notice that experiments will disprove a theory- they can never prove a theory. Also note that the standard for acceptance of a theory is not "consistency" but how the theory corresp

This does not agree with actual scientific practice. A blatantly inconsistent theory with numerous adjustible parameters ("epicycles") that can be tuned to match experiment will not really be accepted. It is usually regarded as a fatal flaw in a candidate theory that it can be shown to be logically inconsistent within itself.

This dictum is of course somewhat modified by the doctrine of effective theories, whereby a theory is accepted if its inconsistancy is not manifest at the energy scales at which it is effective, and the range at which the inconsistancy appears can be partitioned off cleanly and ascribed to some as yet undiscovered theory, effective at higher energies/