Wednesday, December 31, 2014

Blogger's Desk#4- Probability in Science: Why


At the end of 3 years of blogging with 130+ posts and more than 131K views in total, I want to thank those people who have constantly questioned me and sometimes have asked me questions that left me running for answers. I have been able to answer a few of them, with a large chunk of reply attributed to the rescue statement "I'm not aware of any research or text in response to your question". But the path has been rewarding though. This blog page has been a product of my appreciation of research and researcher's alike.

However, there seems to be a question that hit me which is exquisitely simple, but to convincingly tell you an answer, seems to be so difficult. But lately, I think I have an answer and let me make an attempt. Let me re frame the question like this. "Why do scientists always talk about a finding in terms of 'probability' and not certainty? This question has also come up when I was data mining regarding understanding of science (Link) and its public outreach.

Let us consider a scenario. you are mixing a chemical X with Y to yield X2Y4 under specified conditions. If I say there is a statistical chance that you will not get the same reaction 15 times out of 100 attempts despite everything being same. What is your reaction? Absurd right. But that's the way Biology works, since it has variability. That means, at least simply by fluke you end up getting a deviated result, simply by chance. It is possible to quantify this degree of uncertainty and the possibility of chance to a certain extent by using standard bio statistical models. Here's is the first point why biologists talk in terms of probability, regarding any claims.

Second point, studying biology is a complicated process. There are many a times too many variables to consider, with each variable studied one at a time. This is called a reductionist approach. The puzzle is divided into multiple parts, with each individual part solved one at a time, and then the whole thing is stitched together. However, there is a catch here. There is nothing to tell you that the puzzle is complete and hence you cannot be sure. I want to quote from Richard Dawkins (I think it was in his book, "The greatest show on earth"). You could never prove anything with absolute certainty in science. Inability to disprove is a proof of correctness. This stands in line with the standard that any theory should be testable and falsifiable (Read more here).

Let me give you an example. Shamelessly, I'm replicating this from David Eagleman from his book "Incognito". Let us consider that a person who has never known anything about radio waves gets a radio which produces music. Having never seen such a thing he decides to scientifically deduce how it works. After years of research he figures out that by arranging circuit in certain way, he is able to generate the music. As per the science hypothesis would be, he is correct. But the actual way is there is a component which he has never studied and unknown to him. The telecasting radio waves. You see, despite everything being right there was a completely hidden component in the puzzle, but there is no way of knowing there is such an extra component. Now you get the point.

This provides an additional explanation of why with better technologies we come to know better and sometimes may even overturn a standard hypothesis based on fresh new understanding. In a summary, the extreme variance in biology and the inability to predict that a problem has been solved is the reason why a responsible sane scientist always talks about his results in terms of probability.

Wish You all a Happy New Year.

No comments:

Post a Comment