Science And Pseudoscience

Last weekend's edition of The Philosophers Zone looked at Imre Lakatos, at his name changes, his persuading a woman to commit suicide, his escape from Hungary, his swinging between Marxism and right wing politics, and his relationship with Karl Popper. In between, the philosophy of science is discussed and, specifically, the difference between science and pseudoscience:

What Kuhn's saying is not at all surprising. Nobody should ever say that if you've got a theory that involves 15 assumptions in order to derive an observational consequence and that observational consequence turns out to be false, that you've always got a hit, and that's what sometimes it looked like Popper was saying, that anything else would be defensive, dogmatic. But nobody should really say that it should always hit what we might refer to as the central theory, Newton's theory or Maxwell's theory, Einstein's theory. It's perfectly reasonable to say No, it's one of these other assumptions that might be wrong, it's not really denying the Popperian method which at its simplest is completely undeniable, that if you've got a logical deduction of a consequence from a set of assumptions and that consequence is false, then you'd better regard at least one of the assumptions that you derived it from as false. But there's nothing that says it's got to be the central assumption.


But then you've got to say, well, that's also what people do in terrible cases we intuitively put on the wrong side of the line. I don't know if Uri Geller is caught cheating, he says that he had to cheat because although he does have genuine psycho-kinetic powers, he had to cheat because there were sceptics around and the sceptics affect his powers of spoon-bending. Well that looks exactly like the sort of ad hoc defence of a central theory that we all intuitively regard as pseudo-scientific. So there has to be a difference between holding on to a central theory by rejecting some auxiliary assumption in a scientific way, and doing so in a non-scientific way. And all that Imre really suggested, and I think this is to a good degree accurate, was that the scientific way is when you don't just absorb the initial difficulties, so you've got an initial theoretic framework that yields a false observation, you move to a new theoretical framework which gives you the correct version of that false observation. That's easy to do, that's true also in the unscientific cases. What distinguishes the good shifts from the bad shifts is extra independent predictive power. So you work back, Adams and Everier worked back from the details of Uranus's orbit, assuming that Newton's theory is correct, and say what do we need to assume about this other planet in order to give us the correct account of Uranus' orbit. Well naturally then, they get the correct account of Uranus's orbit from the revised system. But it also makes the extra prediction that there is another planet, that there is something that you've misidentified as a fixed star that is in fact moving, and if you look you'll see it, and that's exactly what happened with the discovery of Neptune. So taking away all the arcane terminology, progressive problem shifts are just shifts in which you get some independent testability and the new theory survives that test, whereas degenerating problem shifts are cases where you only absorb previously negative observations in an ad hoc way without any further means of testing.