To: Francis M.
From: Geoffrey Klempner
Subject: Essay on Popper: must good science be falsifiable?
Date: 30 November 2005 10:17

Dear Frank,

Thank you for your email of 21 November, with your University of London essay in response to the question, 'Must good science be falsifiable? If so, why?'

Although you've said nothing wrong, you have only mentioned Popper and O'Hear's Introduction to the Philosophy of Science. O'Hear's criticisms are confined mainly to questions which help to elucidate Popper's theory.

If I was answering this question, I would show that I was at least aware of views which question Popper's 'rationalistic' approach to science, such as Lakatos (Criticism and the Growth of Knowledge) and Kuhn (Structure of Scientific Revolutions), or, at a greater extreme, Feyerabend (Against Method). You don't have to read these books, but it will help a lot if you are aware of them, and the questions which they raise.

I am not going to tell you what you should read. Start by looking at books and articles which O'Hear refers to. If you read just one other book or article whose author you can refer to, it will make a difference.

Your introduction gives a somewhat incorrect impression (although you correct this later) that falsifiable theories are a reliable guide to action. This is obviously absurd. I have a theory (which is easily falsifiable) that my guardian angel will never allow me to be run over by a truck, even if I walk across a six lane motorway with my eyes closed.

Later, you refer to 'corroboration'. The more corroborated a theory is, the you can rely on it as a guide to action. But is that necessarily true? Suppose you meet someone who holds a theory that he is immune to AIDS, a theory which has been corroborated over many years of indiscriminate sex.

I would say to such a person, 'What on earth makes you think you are immune? On what basis have you formed that hypothesis?' Surely it is irrational to form hypotheses purely on a hunch, or on no basis at all. But that is exactly what Popper appears to be saying when he claims that any 'conjecture' is good science provided it is falsifiable.

Or maybe Popper is right. The conclusion we should come to is that not all 'good science' by Popper's definition is a reliable guide to action. And that brings us to the nub of the essay.

The question is about 'good science'. What is good about 'good science'? What is it good for?

Good science advances human knowledge. The fruits of good science are shown in many ways. What is so radical about Popper's approach is that it allows scientists to 'play', to come out with outlandish ideas and theories because these are all grist to the mill. This is the very opposite of the picture you get from traditional empiricist approaches to knowledge, where every step up the ladder of knowledge is based on evidence and observation.

But Popper has been criticized as excessively rationalistic, and I would tend to agree. At one point in your essay, you refer to the possibility of 'supplementary additions' to a theory to save it from being falsified. This issue is more commonly referred to as the problem of 'auxiliary hypotheses'.

How far should we go to save a theory in the face of an apparent refutation? Below, I have copied a paragraph from the Stanford Encyclopedia article on Popper. The gist is that the more important a theory is, the more researchers strive to insulate it from falsification. You don't throw out Newton's laws of motion just because you come across an anomaly in the observed motions of the planets which the theory is unable to explain. Yet Newton's theory was finally overturned. How did that come about?

Going back to your contrast between Popperian 'science' and psychoanalysis, astrology and marxism, I have a slightly personal axe to grind in that my wife and one of my sisters are both art psychotherapists. A lot has been written about the scientific status of various schools of psychotherapy and psychoanalysis, for example, see Farrell 'The Standing of Psychoanalysis' OUP. Is it in fact the case that psychoanalytic hypotheses are not falsifiable? In clinical practice, the therapist will form hypothesis and 'test' them, rejecting those which 'fail'. Isn't that good enough? If you reply that practitioners are not prepared to accept the possibility that their very practice and the assumptions which underlie it might be falsified, the same holds to a considerable extent of any 'very important' theory in science.

So my main criticisms: You didn't spend enough time looking at what it means to say that science is 'good science', and the lack of references to other contributors to the debate.

All the best,

Geoffrey