How do we prove causation?

[Thurlor]

How exactly can causation of anything be proven (beyond any doubt)?

Most of the time when I hear about x causing y it is just a correlation between x and y. Cigarettes and lung cancer, poverty and crime, etc.

It seems rather unethical to claim knowledge of a cause when at best you have knowledge of a correlation.

I have heard people claim that a causal relationship is obvious in most cases because of common sense but that is just another form of bias.

On a philosophical level it almost seems impossible to ever prove causation.

 

[Grayman]

In a controlled environment. If a test is repeated and the result is 100% all the same then you have direct causation.

In the normal environment there are many 'contributing factors' and each factor might have a different degree of impact.

When talking about 'the cause', generally the factor with the greatest impact that you can control is considered the trigger or cause. So if we are talking lung cancer, you can control daily smoking but you cannot control genetics, the many viral/bacterial infections, or accidental inhalation of toxic substances.

 

[Grayman]

Oversimplified Example:

So if a person put a bomb in a building and blew it up, what would be the direct cause of the building blowing up?

Multiple factors are 1) the bomb, 2) the oxygen required for ignition, 3) the person positioning and triggering the bomb

All of these factors have equal impact. If any one of these factors was not there, the building would not have blown up. Because the impact of each factor is the same we look at what can we control.

1. The bomb: We can make bombs illegal and limit access so control is possible but is limited.
2. oxygen: We cannot control oxygen being in the atmosphere so oxygen is not considered a cause of the explosion.
3. The person: We have limited control of human behaviors through laws and social constructs.

So the cause was the person and the bomb. The proper answer is to say both were the cause in part and both need to be regulated to prevent the explosion.

Instead what people do is they focus on the control that is most convenient for them. Banning all access to bombs and regulating the purchase of materials that might be used to make them OR increasing the punishment for such crimes even considering the death penalty.

So using this structure we figured out that the person bombing the building was the cause but then what caused the person to bomb the building?

In order to figure out the cause of the person choosing to bomb the building we have a lot, lot, more factors to consider. This is because people are so much more complex compared to physical examples like the above.

Factors can include:
Mental Disorder - several possibilities
Bad upbringing, abuse, or foster care systems
Fear, or hate of a people or group
Etc...

So let's say all of these were factors but to different degrees. Should we waste time arguing that racial hatred is the cause and just make laws about race crimes OR do we address them all... mental issues by creating programs so that people can get help, revamp the foster care system and address child abuse, and then create more social programs and events were strangers can intermingle and humanize each other instead of stewing over our differences from afar?

 

[Thurlor]

@Grayman

So a repeatable test in a controlled environment would be required? What if the environment can't be controlled? For example what if the studied environment was the solar system, how could it be controlled?

 

[Grayman]

@Grayman

So a repeatable test in a controlled environment would be required? What if the environment can't be controlled? For example what if the studied environment was the solar system, how could it be controlled?

You find environments where the 'controlled variables' fit the conditions you need to make an observation and record the results.

Einstein theory of relativity wasn't proven until there was an eclipse and light was found to bend around the moon. How light moved through gravity was essential to the theory but there was no way to test it in a controlled environment and so the conditions required to prove it had to be found.

 

[Hadoblado]

While I think you're correct that 100% proof is impossible, I don't think it's unethical to claim knowledge without 100% proof, and I don't think it's necessarily bias either.

Do you know that it's not proven?
Do you know that only correlation is established?
Do you know that it's bias?

I think it's great to be able to think like this, but I also think like this tends to be tedious and often applied selectively. It's just frustrating and boring to constantly acknowledge the limits of our understanding when functionally they're a technicality of no consequence. So people don't, and we all trust each other to know the limits of our understanding without explicitly stating it all the time.

Also, where it counts (in science journals), the possibility of error is constantly acknowledged (albeit in some fields their standards aren't high enough) - but this is rarely communicated to non-scientists because it's generally not of interest to them.

 

[Rebis]

Truth is few and far between. Correlation does not equal causation, but in a lot of cases we're making a knowledge claim. We know x has something to do with y given it produces y in a controlled environment: We're not making claims that z (a hidden factor) also influences y, all we know is x implies y.

Control comes with a reduction in the scope of the knowledge claim as there are less variables the further you reduce the proposition. When something becomes univaried as stated above, we know x definitely implies y.

 

[Happy]

Correlation does not equal correlation

 

[Ex-User (14663)]

I suppose you can never prove causation (due to the problem of induction). Something might look causal with 100% consistency yet one counterexample will be enough to falsify the causality.

Statistical analysis of causality is another thing. If you have a time series and one variable is correlated with another one across time, it's a form of causality (e.g. crop yields of wheat can be correlated with precipitation in a certain time frame before harvesting, in which case one can say precipitation causes certain changes in crop yields). Formally referred to as predictive causality, or "Granger causality". (Which btw makes the expression "correlation doesn't imply causation" a pretty imprecise one – in the end nothing implies causation, but certain temporally dependent correlations imply Granger causation)

 

[Rebis]

The mistake has been corrected

 

[Inexorable Username]

I love this question. More people should be asking questions like this these days, as we take far too much for granted. Here's some points to consider:

(These are just my thoughts. I could easily be wrong. Feedback is welcome)

A dog's dental health, physical size, and emotional state may have all contributed to the dog biting someone. However, all of these other aspects are part of the dog, itself, therefore, the dog caused the bite. Someone's behavior may have aggravated the dog, but the most direct, immediate "cause" of the bite, itself, was the dog (and its components. ie: teeth.)

When it comes to assessing a cause, we follow the action.
Then we choose one of the following - the most immediate contributor, or the thing which caused the action which was the most necessary prerequisite to the consequence.

So if there is traffic, and one car rear-ends another, causing a 5 car pile-up, the cause would have been the first car to rear-end another car. The first car is the most immediate contributor to the pileup (closest to the instigating action), and it was also a necessary prerequisite.

Look at these examples as a sequence :
Person shouts > DOG BITES > Because of distress > Because it was tall enough to jump the fence > Because it had all of its teeth, and therefore was able to bite.

What caused the human to be bit? The dog, because DOG BITES is the most immediate reason for the human sustaining a bite, and in order for the human to have sustained said bite, the dog was a necessary element.

There is traffic > FIRST CAR CRASHES > More cars crashed because that car crashed.
What caused the final car in the 5 car pile-up to be rear-ended?
The first car. Despite there being traffic, the other cars could have proceeded normally. The first car set the action sequence in motion, and therefore it was the "key element" in causing the pile up.

 

[Inexorable Username]

I think the next thing to consider would be "cause beyond a reasonable doubt".

If one person drinks coffee, and feels energetic, that's a correlation.
It's still a correlation if 2, 3, or 10 people drink coffee and feel energetic.

Now, lets consider an extreme example. Let's say that 200,000 people all drank coffee, and felt energetic. At this point, you could say that it is unreasonable to believe that coffee does not cause people to have energy.

Now, you might say everyone is different. So maybe it a component in a person's biology that causes the energy in response to the coffee...
But consider the case of immaculate conception. Assuming the Virgin Mary was, in fact, a virgin when she gave birth - she would be a pretty outstanding exception to the whole "copulate and create kids" theory. Since everyone else still has to copulate to create kids, Mary probably isn't enough to say that copulation does not cause pregnancy.

So to some degree, I would say that causation could be proved by demonstrating that the probability of the cause NOT leading to the effect is statistically insignificant.

----------

I think the reason that causation is confusing is because people because people abuse it. Some examples:

In society: "He did it because he was drunk. If he wasn't drunk, that never would have happened." Well...yes, but he was still the most immediate cause of the guy getting punched in the face. So - he caused the punch. (Whether or not he should be punished is a different matter)

In science : "Our need to use tools caused us to evolve to have opposable thumbs"
There's very few things that could actually be considered causation when it comes to evolution. This happens a lot, in science, unfortunately. In this instance, there isn't enough information to assume a causation.

One of the biggest offenders when it comes to inferring causation, in my opinion, is science. Take nutrition for instance. If a number of people drink milk, which contains calcium, and their bone density increases - many scientists would say that ingesting calcium causes increased bone density. This isn't necessarily the case, though. You can say that milk increases bone density, but you cannot assume that a component within the milk increases bone density without more testing.

That's why I love that you asked this question.
The decrease in the quality of our scientific research has really been bothering me these days. The issue is that many, many people will use scientific studies to argue a stance. The idea is that the authoritative source should be accurate, and in the case of a scientist - unbiased. I think, traditionally, that has been true. Lately though...science is falling a bit short of that expectation, and we've become so dogmatic about scientific studies, that people become emotional if you contradict the findings.

Arrogance has been an issue in science for a very long time. However, we live in an era where laymen have access to information, and everybody and their neighbor can become a scientist. We need to shelve our arrogance and make it socially acceptable to question the validity of incomplete, or theoretically-based scientific arguments.

 

[Cognisant]

In science : "Our need to use tools caused us to evolve to have opposable thumbs"
There's very few things that could actually be considered causation when it comes to evolution. This happens a lot, in science, unfortunately. In this instance, there isn't enough information to assume a causation.

That's not science, that's journalism, the peer review system makes scientists extremely pedantic because like philosophers on a forum anything they say/write that isn't carefully worded could be misconstrued as a mistake that another could take advantage of to further their own reputation.

Take nutrition for instance. If a number of people drink milk, which contains calcium, and their bone density increases - many scientists would say that ingesting calcium causes increased bone density. This isn't necessarily the case, though. You can say that milk increases bone density, but you cannot assume that a component within the milk increases bone density without more testing.

Again that's not science, that's marketing, there's vested interests in various industries that want to "prove" the benefits/superiority of their product to try and convince people that for example they should have at least a glass of milk every day so as not to suffer from osteoporosis in old age. They do this by stretching the truth, making claims that aren't entirely false (calcium is needed for strong bones and a lack of calcium can cause osteoporosis) to mislead people to a false conclusion (that they need to drink milk every day) or funding multiple studies with small test groups and cherry picking the ones that get the results they want so they can say "studies show that <buy our product>".

It's not science that's the problem, it's capitalism.

 

[Daddy]

How exactly can causation of anything be proven (beyond any doubt)?

Unless you can control every variable in the system, you simply can't. You need to be God.

Though if you 'define' the system and its associated variables, you can squeeze induced causation, even if it's still just a castle in the sky when it comes to actual reality. That's kind of mathematics. But it promises so much conclusive potential, maybe no one cares.

 

[Inexorable Username]

I absolutely love thy is que

In science : "Our need to use tools caused us to evolve to have opposable thumbs"
There's very few things that could actually be considered causation when it comes to evolution. This happens a lot, in science, unfortunately. In this instance, there isn't enough information to assume a causation.

That's not science, that's journalism, the peer review system makes scientists extremely pedantic because like philosophers on a forum anything they say/write that isn't carefully worded could be misconstrued as a mistake that another could take advantage of to further their own reputation.

Take nutrition for instance. If a number of people drink milk, which contains calcium, and their bone density increases - many scientists would say that ingesting calcium causes increased bone density. This isn't necessarily the case, though. You can say that milk increases bone density, but you cannot assume that a component within the milk increases bone density without more testing.

Again that's not science, that's marketing, there's vested interests in various industries that want to "prove" the benefits/superiority of their product to try and convince people that for example they should have at least a glass of milk every day so as not to suffer from osteoporosis in old age. They do this by stretching the truth, making claims that aren't entirely false (calcium is needed for strong bones and a lack of calcium can cause osteoporosis) to mislead people to a false conclusion (that they need to drink milk every day) or funding multiple studies with small test groups and cherry picking the ones that get the results they want so they can say "studies show that <buy our product>".

It's not science that's the problem, it's capitalism.

Well, I know it's not science. That's the issue! Still, glad to hear someone agree with me, because if you would believe it - I actually get a lot of people disagreeing when I make this point. The general consensus is that if we don't take studies for granted and trust in the authority of degrees and universities, we'll either (A) Fail to make substantial progress, or (B) Get too wound up in little details like wording things properly.

Anyways, I have a mind to start collecting "scientific articles" that demonstrate a poor scientific perspective. It shouldn't be difficult. If I think of this post, I will send you examples at some point of the poor quality, arrogant scientists I've been observing these days.

 

[Cognisant]

Scientists don't write articles, they write white papers and scientific journals.

Journalists write articles.

 

[Inexorable Username]

Scientists don't write articles, they write white papers and scientific journals.

Journalists write articles.

I know that. Poor choice of words. I'm referring to studies. Properly formatted scientific studies with abstracts.

I don't read the news, and the only time I read articles is when I'm trying to find information that is non-scientific in nature, or evergreen. Evergreen, meaning content that either never changes, or is expected to not change for a very long time. Ie: the way in which L-Tryptophan is converted to Seratonin.

One of the sources that is low on my list of credibility is the NCBI (ncbi.nlm.gov)
What are your thoughts on using that as a source?

I've not had the best of luck when it comes to finding quality sources there, so I've been shying away from it, but unfortunately, it can be pretty hard to find multiple sources on specific subjects without being affiliated with a university.

Your tips are, of course, appreciated. Always looking to improve my research skills.

 

[Grayman]

I love this question. More people should be asking questions like this these days, as we take far too much for granted. Here's some points to consider:

(These are just my thoughts. I could easily be wrong. Feedback is welcome)

A dog's dental health, physical size, and emotional state may have all contributed to the dog biting someone. However, all of these other aspects are part of the dog, itself, therefore, the dog caused the bite. Someone's behavior may have aggravated the dog, but the most direct, immediate "cause" of the bite, itself, was the dog (and its components. ie: teeth.)

When it comes to assessing a cause, we follow the action.
Then we choose one of the following - the most immediate contributor, or the thing which caused the action which was the most necessary prerequisite to the consequence.

So if there is traffic, and one car rear-ends another, causing a 5 car pile-up, the cause would have been the first car to rear-end another car. The first car is the most immediate contributor to the pileup (closest to the instigating action), and it was also a necessary prerequisite.

Look at these examples as a sequence :
Person shouts > DOG BITES > Because of distress > Because it was tall enough to jump the fence > Because it had all of its teeth, and therefore was able to bite.

What caused the human to be bit? The dog, because DOG BITES is the most immediate reason for the human sustaining a bite, and in order for the human to have sustained said bite, the dog was a necessary element.

There is traffic > FIRST CAR CRASHES > More cars crashed because that car crashed.
What caused the final car in the 5 car pile-up to be rear-ended?
The first car. Despite there being traffic, the other cars could have proceeded normally. The first car set the action sequence in motion, and therefore it was the "key element" in causing the pile up.

Why do people often simplify things into a single cause rather than distribute cause?

In other words, all cars involved in the pile up were the cause of the pile up to varying degrees. Some were not paying attention. Some were following too closely. Etc...

 

[Inexorable Username]

I love this question. More people should be asking questions like this these days, as we take far too much for granted. Here's some points to consider:

(These are just my thoughts. I could easily be wrong. Feedback is welcome)

A dog's dental health, physical size, and emotional state may have all contributed to the dog biting someone. However, all of these other aspects are part of the dog, itself, therefore, the dog caused the bite. Someone's behavior may have aggravated the dog, but the most direct, immediate "cause" of the bite, itself, was the dog (and its components. ie: teeth.)

When it comes to assessing a cause, we follow the action.
Then we choose one of the following - the most immediate contributor, or the thing which caused the action which was the most necessary prerequisite to the consequence.

So if there is traffic, and one car rear-ends another, causing a 5 car pile-up, the cause would have been the first car to rear-end another car. The first car is the most immediate contributor to the pileup (closest to the instigating action), and it was also a necessary prerequisite.

Look at these examples as a sequence :
Person shouts > DOG BITES > Because of distress > Because it was tall enough to jump the fence > Because it had all of its teeth, and therefore was able to bite.

What caused the human to be bit? The dog, because DOG BITES is the most immediate reason for the human sustaining a bite, and in order for the human to have sustained said bite, the dog was a necessary element.

There is traffic > FIRST CAR CRASHES > More cars crashed because that car crashed.
What caused the final car in the 5 car pile-up to be rear-ended?
The first car. Despite there being traffic, the other cars could have proceeded normally. The first car set the action sequence in motion, and therefore it was the "key element" in causing the pile up.

Why do people often simplify things into a single cause rather than distribute cause?

In other words, all cars involved in the pile up were the cause of the pile up to varying degrees. Some were not paying attention. Some were following too closely. Etc...

Well, if it’s a pile up, it’s almost always because all of the cars were following too closely to some degree, and usually whether or not you’re paying attention doesn’t matter at that point. Pile ups don’t generally happen when people are going 20 miles an hour. They usually happen on highways.

^ this information comes from my experience of living in Silicon Valley before and during the legalization of pot. They laid off most of their cops, and it was like the apocalypse. Crazy times. Pile ups probably about once every three days. Only time in my life I ever listened to the radio, because I needed to know which roads were closed.

As far as your point about multiple causes...well- my interpretation of the question was essentially “how do you pinpoint the cause”

I feel like what you’re making a point about is blame, really, which is different.

Someone’s punch to the face might have triggered you to punch them back. You punching them was caused by them punching you, BUT you are entirely to blame for the punch you threw.

If we’re all running out of a burning theatre and someone gets trampled alive - the fire was the cause of the trampling. However, the people who ran over the person are to blame.

I guess you could at that cause is fairly objective, where as blame is subjective.

You’re right though, this is oversimplified! I simplified my argument because it’s really the most concise...and I think most coherent way for me, personally, to express a complex topic.

Before you said something about it, I didn’t really think about blame...but you raise a really good point. There does seem to be an incredibly valuable distinction between cause and blame that I’d never thought of. I learned something today!

Comments like this are what I live for. I’m such a hermit...I barely talk to anyone. Since coming to this forum, I’ve been challenged by so many people and it’s forced me to think in ways I’ve never had to think before. I feel so excited. It’s weird.

Thanks for responding to my post
I hope you respond to more!

 

[ZenRaiden]

It kind of matters what means proof or what evidence there is and what results one seeks. For all practical purposes and intents its most about what is more likely.

I mean best way we can look at human behaviour and just look for repsonses we get and what sort of situations we get. Obviously people can tell you sometimes why they behave the way they do and that it self can be enough of evidence.