Thursday, June 25, 2009

Falsifiability and having it both ways

Long time readers might have figured out that I have it in for Karl Popper's philosophy of falsifiability. I don't believe Popper gave an accurate description of how science works, nor how science should work. I think Popper's philosophy is at best a clunkier version of induction, and at worst, simply wrong. It is also prone to abuse and/or misunderstanding.

However, I will say that there is more than just a grain of truth to it. At the heart of the philosophy of falsifiability is the recognition a particular logical fallacy. I am not sure if this fallacy has a common name, but I will call it "having it both ways".

Take the following illustrative real-life example (via Phil Plait). Patricia Putt, a claimed psychic applied for the James Randi's Million Dollar Paranormal Challenge. In this challenge, Putt and the testers agreed on an experiment which would be able to test her ability to "read" people. In the experiment, Putt read ten women and wrote down descriptions of each of them. Each of the women was asked to choose the description out of the ten which fit her best. If five of the women picked out the correct description, Putt would move on to the final testing.

As it happened, none of the women picked out the correct description. At a later time, Putt decided that this meant she had actually gotten 10 out of 10, since each of the women had picked out one of her descriptions. Nevermind that each of the women were asked to pick out a description, and that none of them picked out the correct one.

There are only two possible results to this experiment:
(1) Most of the women pick the correct description.
(2) Few or none of the women pick the correct description.

If (1) had occurred, then that would be evidence in favor of Putt's psychic abilities. Not absolute evidence, mind you, since it could also happen by chance. But it would be evidence enough that we would want to investigate it further. However, what actually occurred was (2); none of the women chose correctly. Putt claims that this is nonetheless evidence in favor of her psychic abilities. But she can't have it both ways! We might say that one possible result is evidence for psychic ability, or that another possible result is evidence for psychic ability, but we cannot say that all possible results would be evidence for psychic ability.

If all possible results of an experiment have the same conclusion, then there would hardly be any need to perform the experiment in the first place. Even before the experiment occurs, we know that either (1) or (2) will occur. According to Putt, either way, this means the experiment comes out in her favor. But then, what is the point of the experiment? Can we really call it experimental evidence if we think it is unnecessary to ever actually perform the experiment?

The bottom line is that if you want to entertain the possibility of evidence for something, you must also entertain the possibility of evidence against it. You can't have one without the other.

I propose that this fallacy doesn't just make for bad science, it makes for bad reasoning, period.

2 comments:

Larry Hamelin said...

You're wrong about Popper. I'll explain later on the other post.

Larry Hamelin said...

I've commented on the original post.

Keep in mind that Popper was a philosopher, not a scientist. Much of what he wrote, even about science, was bullshit, but philosophers are held to a lower standard than ordinary people. That he had one good, original philosophical idea is enough to classify him as a genius, but that doesn't mean we have to take the rest of his writing as gospel.