Cause and Effect

[booker]

What are cause and effect? Is there really anything to distinguish the two other than saying that one comes before the other?

 

Edit: In the context of relativity you might say that there are two time-like interdependent events. According to any clock, effect would proceed cause. But is there more?

 

Perhaps I could have posted this in the quantum mechanics folder where observation of the wavefunction could be an issue.

 

Re: Distinguishing cause and effect using a clock. Feynman would call a clock a thermodynamic device, where the direction of advance would be determined by conditions of lower entropy in the past light cone than the future light cone. That is, the clock is advanced when in a spacetime region of lower entropy.

Edited July 29, 2008 by booker

[Mr Skeptic]

On a small scale, though, I think causes and effects must occur simultaneously.

[Pete]

On a small scale, though, I think causes and effects must occur simultaneously.

If that were true then what is the distinction between them?

 

Pete

[fredrik]

What are cause and effect? Is there really anything to distinguish the two other than saying that one comes before the other?

 

This could be made a deep philosophical question and maybe I missed your point but...

 

A cause is like a condition, that together with an "implication law" imples an effect.

 

We may have a set of conditions A.

We may have a set of conditions B.

 

And a law that says that if A holds is true then we can conclude that B holds.

 

A causes B (if A is a sufficient cause)

 

But the interesting philosophical question is from the epistemological perspecitve is how we come to conclude that the law itself holds?

 

If the law is given, then the causation is a plain deduction. If we associate the progression of events (time) with the causation, then we have a sequence of deductions, where the "past, together with the laws of physics, causes the future".

 

This is a bit like a deterministic philosophy, still leaving the big epistemological question on howto infere the deterministic structure without using induction (which should have no place in a deterministic world)

 

But IMHO from the point of a sound scientific method laws are discovered or inferred from experience/experiment (how else does it happen?). So the interesting part here isn't the deductive case of causation, but the inductive case where the question arises what causes the law of causation in the first place?

 

We all know the drill that correlation does not imply causation. But from the point of inductive reasoning based on experience, what is a sound logic of discovering the causal laws?

 

There might not be a sharp line between inductive logic, in the sense that a correlation implies causation to "a certain degree of certainty", and the common sense metod of the scientific discovery.

 

This inductive reasoning is much more in line with an indeterministic worldview.

 

I think these questions are interesting but goes beyond relativity itself. Relativity does not explain these things. The just supposedly constrains the effect of any cause to the future light cone, that constrains the causal structure. But there are question on "what is causality" that is not satisfactory answered by relativity.

 

So what Relavitiy does is to somehow suggest a structure on the boundaries of causation. Still leaving the deeper meaning of causation in the context of science unresolved.

 

/Fredrik

 

An inductive version of causation instead of an implication, A implies B, one could say that

 

A imples B, with a certain "probability", also conditional on the very reasoning beeing done.

 

This suggest that A can be likely to cause B, and yet later we find that A might not cause B, but then this gets consequences for our degree of confidence in the very reasoning.

 

The motivation for the fuzzy reasoning could for example be that, it's the best guess we can come up with. It's somehow the subjectively judged best guess we can produce given our massive ignorance and incompetence.

 

/Fredrik

[ajb]

In relativity you say something like "effects must be in the future light-cone of the cause". This is the statement of causality. Penrose diagrams are very useful in describing the causal structure of a space-time. I am not very familiar with them, well nothing like what I should be.

 

I think the sort answer to your question is that cause happens before effect. As you have stated. But then we do have the question of Closed Time-like Curves (CTCs) and time travel in general relativity...

 

In quantum field theory you have the notion of micro-causality. This is the statement that the (anti)commutator (equivalently the propagator) of two space-like separated fields vanishes. Basically, this means they "cannot know about each other".

[Mr Skeptic]

On a small scale, though, I think causes and effects must occur simultaneously.

If that were true then what is the distinction between them?

 

Pete

 

None, really. I'm thinking here of things like particle interactions. Any changes to one particle will occur simultaneously with the changes to the other particle. If a cause and effect are separated by spacetime and have an indirect interaction, then the travel time is always positive so that the "cause" must come before the "effect". But for an indirect interaction, there can be an interruption, such that the "cause" does not necessarily result in the "effect".

 

For example, if you aim a gun at me and shoot, that does not necessarily mean that I will get a bullet wound. I might move out of the way, or an object may block the bullet's trajectory. So it would be somewhat improper to say that firing the gun causes bullet wounds in people the gun is aimed at. But firing a gun accelerates the bullet, and being hit by a fast-moving bullet causes bullet wounds.

 

So the only reason that we say that cause comes before effect, is because we like to refer to things separated in spacetime as having a cause and effect relationship, rather than saying that the "cause" accelerates particles or generates photons, and these may later interact with and a second object to cause the "effect".

[booker]

This could be made a deep philosophical question and maybe I missed your point but...

 

Thank you for the reply, Fredrik. You'v obviously given this a great deal of thought!

 

Maybe I do have a point--or a small objective. I ask, can any given law of cause and effect be states as a set of convervation laws equipped with labels that tell us which elements were cause, and which were effect. Then it's a natural question to ask if the law is invariant under relabeling--or if a new and tenable law is obtained.

 

A cause is like a condition, that together with an "implication law" imples an effect.

 

We may have a set of conditions A.

We may have a set of conditions B.

 

And a law that says that if A holds is true then we can conclude that B holds.

 

A causes B (if A is a sufficient cause)

 

But the interesting philosophical question is from the epistemological perspecitve is how we come to conclude that the law itself holds?

 

If the law is given, then the causation is a plain deduction. If we associate the progression of events (time) with the causation, then we have a sequence of deductions, where the "past, together with the laws of physics, causes the future".

 

This is a bit like a deterministic philosophy, still leaving the big epistemological question on howto infere the deterministic structure without using induction (which should have no place in a deterministic world)

 

But IMHO from the point of a sound scientific method laws are discovered or inferred from experience/experiment (how else does it happen?). So the interesting part here isn't the deductive case of causation, but the inductive case where the question arises what causes the law of causation in the first place?

 

We all know the drill that correlation does not imply causation. But from the point of inductive reasoning based on experience, what is a sound logic of discovering the causal laws?

 

I don't know the drill. Could you tell me about it?

 

There might not be a sharp line between inductive logic, in the sense that a correlation implies causation to "a certain degree of certainty", and the common sense metod of the scientific discovery.

 

This inductive reasoning is much more in line with an indeterministic worldview.

 

I think these questions are interesting but goes beyond relativity itself. Relativity does not explain these things. The just supposedly constrains the effect of any cause to the future light cone, that constrains the causal structure. But there are question on "what is causality" that is not satisfactory answered by relativity.

 

There's was method to my madness in posting here. From a couple recent posts I'd read in the relativity folder, it seemed the present company was capable of thoughtful consideration. (As a benefit, I've enjoyed reading the content relating cause and effect to relativity.)

 

So what Relavitiy does is to somehow suggest a structure on the boundaries of causation. Still leaving the deeper meaning of causation in the context of science unresolved.

 

/Fredrik

 

An inductive version of causation instead of an implication, A implies B, one could say that

 

A imples B, with a certain "probability", also conditional on the very reasoning beeing done.

 

This suggest that A can be likely to cause B, and yet later we find that A might not cause B, but then this gets consequences for our degree of confidence in the very reasoning.

 

The motivation for the fuzzy reasoning could for example be that, it's the best guess we can come up with. It's somehow the subjectively judged best guess we can produce given our massive ignorance and incompetence.

 

/Fredrik

 

-booker

[fredrik]

I don't know the drill. Could you tell me about it?

 

It's a common fallacy in interpreting statistics.

 

Se for example http://en.wikipedia.org/wiki/Correlation_does_not_imply_causation

 

For example, someone collects statistics on a particular type of cancer patients and finds say that smokers are over-represented in that group.

 

Ie. we have a correlation between smokers and cancer. this can be thought to be a fact.

 

But can you conclude hat smoking causes cancer? (Now that may be true for othre reasons, but the above logic is not a deduction.)

 

The same can be applied to other cases where the absurdity becomes more clear. Support that in the group of cancer patients it is found that they watch far more TV than others, or knit more often.

 

One can by the same logic then suggest that watching TV or knitting causes cancer.

 

Sometimes funny things appear in the media which are extracting causal connections from statistics. And by choosing the survey, you can produce statistical data that with the same fallacy support just about anything.

 

The correlation is a statistical fact (although still subject to statistical uncertainty due to sampling issues), but the cause can not be deduced from the data.

 

At best, one can induce to various relative degrees of belief of "probable causal connection". If he inductive nature of the reasoning is admitted, it all comes out as more honest, and the reasoner is fully aware of that he is not making deductiosn - he is guessing. But that's is often

 

The emphasis in my post was not questions like, "given a causal law then ...(insert question)". My emphasis that I personally find interesting, is how knowledge of a causal law is acquired, in particular in the light of the difficulty of making deductive conclusions from observations and statistics.

 

This is also connected to old philosophical questions of

- howto go from observation to law

- Howto count/rate evidence.

- howto combine evidence

- howto rate the confidence in "induced/acquired opinon"

 

Originally, in the early days of science, or pre-science, alot was a matter of opinon, and probable opinon was those of authorities. Eventually it was found that this is pretty ambigous, and unsatisfactory. So people started to question these things. It gave birth to many philosophical questions, and also probability theory. Very long time ago, one could not for example unambigously computer odds of games. Alot of it was probably based instead of experience "playing with dices", rather than analytical reasoning of the mathematics of gaming.

 

Out of this questions regarding the philosophy of the scientific method also comes. Because science is somehow the idea of inferreing from experience and experiment, laws of nature. How is this done without running into fallacies like the above?

 

/Fredrik

 

I suspect this is going more philosophical, but not less interesting.

I'm not sure if someone would want to move this to philosophy.

 

Maybe I do have a point--or a small objective. I ask, can any given law of cause and effect be states as a set of convervation laws equipped with labels that tell us which elements were cause, and which were effect.

 

I can't say for sure I understand you exactly, but to reflect anyway.

 

My view of the causation concept (the view I feel makse sense to reality) is not of deductive type, but of the inductive type. And deduction could be seen as a limit case of induction. About realism inspired deterministic induction I can't comment. It makes no sense to me, but prehaps others can comment.

 

I'm tempted to say that a fundmental causation is time evolution. This is basically the problem of trying to induce the "future" from the present and possible part of history, which is somehow the ultimate goal of science.

 

In this view, the assymmetry between cause and effect is related to the arrow of time. The degree of certainty to which we can distinguish cause and effect is the degree of certainty which we can distinguish the arrow of time.

 

From a the view of statistical induction of causal law, one can imagine that in the case of very little evidence the uncertatiny in the inductions are very uncertain, and in the extreme case almost random in which cause it's effectively impossible to induce the causallaw.

 

Then it's a natural question to ask if the law is invariant under relabeling--or if a new and tenable law is obtained.

 

Interesting question. I wonder how you define the relabeling. Somehow the use of the word label implicity suggest that it lacks physical significance. They are merely symbols.

 

So is the "probabilistically" induced causal law, invariant under relabelling of events or evidence? Is that something what you ask?

 

A very interesting question not so easy to answer and I think it depends on wether we consider a non-physical relabelling (like change or coordinate system) or if you consider a physical relabelling where the labelling is considered to originate from a physical process. In the former case I think it's trivially invariant, in the latter case I think the question is still unclear. I think we need so supplement the physical meaning of the labels in the first place. If they have no physical meaning, then perhaps we should reformualte the question in terms of what has physical meaning.

 

In my very personal thinking, the induction of causal law are not uniqe. It's observer dependent. But, OTOH I see the slight disagreement in this law, the very basis for interactions. This recoveres the consistenty. The subjectivty of "induced causal law", predicts interactions, that is due to this subjectivity.

 

So maybe one possible suggestive answer to your question is that - except for the trivial meaning of labels - is that the eventual "invariance of causal law" during another indexing/labeling of evidence, is to be identified with interactions between the two index systems.

 

Edit: One might however try to construct a limited special theory where invariance is guaranteed by construction, and so to speak consider only the group of indexing systems (observers) who does leave the causal law invariant. But does this set exhaust are actual observers? To extend the group of obsevers, is then as per my suggested reasoning, equivalent by introducing further interactions.

 

/Fredrik

Edited July 30, 2008 by fredrik
multiple post merged

[booker]

Fredrick-

 

The physics is probably not as sexy as all that you've written about. The simplest example is a free Newtonian body in motion.

 

The cause in this case, is say, a particle traveling with momentum in the x direction with coordinates (0,0,0) at time t0. The effect is the particle traveling in the x direction with coordinates (10,0,0) at time t1. If you want you can have collisions with other particles at times t0 and t1 so it has more the look and feel of cause and effect, that you might expect.

 

One conserved quantity is momentum. The convervation law is Newton's first law (A body remains in motion unless acted upon by an exteral force). The labels of cause and effect are derived by comparing the values t0 and t1. The lesser value is cause; the greater value effect. Relabeling, like you say, by some transformation that inverts the time, can feel a little more physical than simple relabeling.

 

There's an evolution operator for the object's position that describes how some things are not conserved--like spatial coordinates--they do not remain the same from cause to effect-- x(t) = vt + x(t0).

 

In quantum physics the same element appear as long as one assumes wave a function corresonds to some element of physical reality--otherwise we have to speak of a potentia of some sort--or knowables--or the evolution of a calculational device.

Edited August 1, 2008 by booker