Causality Vs Correlation

4 minutes



Let us do one thought experiment.

You see, at a fair distance, line of tall trees. After a while, you see a bird comes flying and sits on a branch of a tall tree. It does not matter which type of bird or tree they are. Just moments after bird sits on tree branch, tree starts falling. Bird flies away and tree falls down completely.

Let me ask you a question. Why did that tree fall? Do you think it has anything to do with bird sitting on a tree branch? You have to consider that bird, most probably, is not too heavy as it was able to fly. And tall trees do fall once in a while. Or may be, tree was too old. If you think like me, you would say “I do not know, there is not enough information, but it is more probable that bird has nothing to do with fall of tree.”.

Remember your answer, whatever it is.

Let us continue with our thought experiment. Now, you look at that line of trees again. Of course, there is one less tree standing. Then, you see another bird flying down onto a branch of another tree. To your astonishment, that tree starts falling. Before that tree falls completely, bird flies away.

I am going to ask you again. Why did this tree fall? Anything to do with bird sitting on it?

May be it is coincidence. Why would a big tree fall because of a bird? Trees can even withstand strong wind!

But you see even third tree falls in same way. A bird comes, sits on a branch, tree falls. What now?

Let us assume you have time and you watch that line of trees for whole day – and you count that 189 trees have fallen after a bird sits on a branch. What do you think now? Do think a bird causes fall of a tree when it sits on a branch? If you are still not convinced, let me ask you another question. How many trees will need to fall after a bird sits on a branch, for you to be convinced that causal relation is really there between bird sitting on a branch and tree falling?

If you have put thinking effort, you should see that it is not that straight forward to differentiate between correlated/coincidental events and causal relations between events.

I hope you would agree that, before establishing causal relations between events, you have to observe, many many times, those events happening in temporal sequence in consistent manner. Of course, still that does not guarantee causal relation. Like you, there should be many more observers should observe those events in same way. Many of those observers should get creatively critical to have controlled environment & then observe those events. Some would grow farm of trees of various kinds and let birds of various kinds sit on them. Or see what happens when you hang weight, equal to a bird’s, on branch of tree. Some should do ‘postmortem’ analysis of fallen tree to find clues of how it fell. And so on. This should go on for years! These activities should produce lot of knowledge and if such knowledge if affirmative towards causal relations then may be, may be you gain confidence that it is causal relation and not just coincidence. Even then, it is just not guaranteed as you do not know ‘real’ reason why tree falls when bird sits on it. You ‘just’ know that tree has fallen every time bird sits on it!

Keep thinking…