3

I'm trying to prove that all physical THEORIES are just experiments and tests that conduce to assumptions about correlations between causes and results. BUT how physicists conclude that a relationship is a CAUSATION and not only CORRELATION ?

When can something related to something else be classified as a cause and not only as an accompanying phenomenon ?

Positive examples : mass and gravitation , electric charge and electric field , night and sun

Negative examples : rain and rainbow , wings and flying .. (they are related but are not a reason or result for each other)

Sofia
  • 6,826
  • 3
  • 20
  • 38
Mostafa 36a2
  • 133
  • You have to start from somewhere. First you assume correlation does imply causation. Then you test it. A LOT. If you initial assumption breaks down under some circumstances, then probably in that case your correlation does not imply causation. E.g mass and gravitation: We have no evidence of mass/energy not causing gravitational attraction. – PhotonBoom Jan 21 '15 at 14:01
  • What if they are both caused by a third thing ? Don't you think that this makes the whole humanity knowledge based on HOW FAR CAN WE GO following the reasons ? and nothing can be absolutely trusted nor true , what if we then will base on it to get further ! – Mostafa 36a2 Jan 21 '15 at 14:16
  • Yes of course. Thats why I said we need to test something a lot. Something similar actually happened with relativity. We assumed that Galilean invariance was true and we kept getting wrong results with EM. Up until then we had no reason to assume that G.I was wrong until we extended it further and got wrong results. And then we invented Special Relativity – PhotonBoom Jan 21 '15 at 14:19
  • So : nothing is 100% true , what ever it is ! based on the way we are getting these results , we can extend the argument when we talk about mathematical theories that have been proved by testing billions of cases using computers, this is not enough to prove it whatever LOTS of testing we have done ! – Mostafa 36a2 Jan 21 '15 at 14:24
  • 1
    In science nothing is 100% certain, or to be fair, very little is 100% certain. In physics we are pretty much always using effective theories (i.e approximations) and these theories change all the time in light of new evidence. – PhotonBoom Jan 21 '15 at 14:27

3 Answers3

2

Experimentation is the key. You find things correlated in data. You hypothesize a law that quantifies the correlation. Then, you take the two variables, in a controlled environment, and you vary one of them. If the dependent variable changes in the predicted way, then you've shown that your independent variable(s) have a causal relationship with the dependent variable.

The key point is that you don't really know until you've done the experiment. There is more sophisticated stuff you can do, but this is really the essence of this. For instance, Newton's law of gravitation was not really shown to be a truly causal theory until Cavendish did his experiments with the lead balls.

Zo the Relativist
  • 41,373
  • 2
  • 74
  • 143
  • 1
    And "proved" is really a dirty word for this sort of thing. In physics, we typically use "proved" in the mathematical sense -- showing that if we make a set of assumptions X, then it is logically and universally true that a set of conclusions Y are true. In this sense, it is impossible to "prove" a scientific law true, because we can't do universal experiments. – Zo the Relativist Jan 21 '15 at 15:13
0

This probably belongs more in philosophy but usually if things are strongly correlated but one isn't the cause of the other, they are both caused by a third thing. So for example, global temperature is strongly correlated with the number of pirates, both which are probably "caused by" an increasing industrializing human population. One can rule out pirates as a cause of global warming by manipulating the number of pirates independently. On the other hand, if we directly manipulate that variable, it's a strong indicator that a third, independent thing isn't at work.

From a physics point of view, I'm not sure rain and rainbows are a good example. Rain can be both a cause of rainbows and correlated with it. In other words, something can have multiple causes.

lionelbrits
  • 9,355
0

Let me take an example. The human population increases every year by a huge number, on average. Similarly, galaxies and clusters millions of light years away from us move away by millions and billions of kilometers every year. There is an increase in both quantities. (Population and Distance.) Does this imply that the increase in human population cause galaxies to move away from us at a farther rate? No chance. This implies that correlation does not necessarily imply causation. (people could even have a complete physical model about human population and spacial expansion but that still wouldn't imply that humans cause galaxies to move away.)

Hritik Narayan
  • 7,318
  • 3
  • 33
  • 46
  • Why ? how can you be sure about this ? and why not apply the same argument with gravitation and Mass for example ? correlation does not necessarily imply causation , even if the gravity increases with respect to mass , that is not enough to conclude the causation !! – Mostafa 36a2 Jan 21 '15 at 14:23
  • Why is it not enough to imply causation? – Hritik Narayan Jan 21 '15 at 14:25