Sunday, June 2, 2013

Science and inference to the best explanation

To: Chris M.
From: Geoffrey Klempner
Subject: Science and inference to the best explanation
Date: 4th March 12:33

Dear Christian,

Thank you for your email of 22 February, with your essay for the University of London BA Logic module, in response to the question, 'Does inference to the best explanation constitute a distinctive way of assessing scientific theories?'

As you point out, there are different ways in which a method of assessing scientific method may be described as 'distinctive'. However, I think that the point of the question turns on what we are distinguishing it from. If IBE is just a fancy way of describing a collection of methods that are already well known and documented, then it is not 'distinctive'. On the other hand, if there is a valid or useful contrast to make between IBE and other methods, then (right or wrong) it is 'distinctive'.

Then again, as you also point out, what are we looking for here: a more accurate description of scientific practice, or a normative standard which we can employ in order to choose one theory in favour of another? Clearly, a lot depends on what one understands by the notion of 'assessing'.

I would have liked to have seen a bit more exposition of the basic idea behind IBE. How does it differ from, or compare with Peirce's notion of 'abduction' (if at all)? How does it differ from Popper's account of 'conjecture and refutation' (if at all)? In my own usage, I tend to employ the terms 'inference to the best explanation' and 'hypothetico-deductive explanation' as if the two notions were synonymous. Am I wrong in doing this? What is so new about Lipton's idea that we assess theories on the basis of their 'loveliness'? Hasn't it always been known that aesthetics plays a vital role?

Lipton clearly wants us to take notice of something, which is sufficiently 'distinctive' to be worth writing a book about. I think what the question is really asking -- between the lines -- is whether he has contributed something new to the debate or merely rehashed ideas that were already current.

Producing scientific theories is a three stage process: We generate possible theories or explanations; we select theories or explanations to test; and then we test them. Lots of theories are not worth testing (because they are so 'unlovely'). Some are more interesting but untestable. When we have done all our testing, there may be two or more theories still in the running. Even if only one theory survives, we do not necessarily 'suppose it to be true' in van Fraassen's sense. However lovely you find your dream partner, it is sensible to spend some time together before you propose marriage.

The crucial idea here concerns consequences: there are consequences in selecting a theory for testing, and there are also consequences in accepting it 'as true', or 'as the best theory'. A great deal of money goes into funding scientific research projects so it is no small matter to regard a particular theory as 'worth' taking an interest in, or to reject another theory as 'not worth' taking to the next stage. However, the philosophy of science isn't really concerned with economics. As philosophers, we are concerned with truth. We want to know how to decide which explanation is true, or more likely to be true, or (at least) less likely to be false.

Lots of theories were involved in sending a man to the Moon. Given the enormous care taken to avoid disasters, you wouldn't rely on a theory just because you 'liked' it. If when the stakes are the highest they can be you are prepared to act on a statement, then you have demonstrated in the clearest way that you regard that statement 'as true'.

But many theories are not like that. Their consequences are more diffuse, less easy to pin down. Relative to the practical needs of technology, arguments over the relative merits of different cosmological theories are no different from armchair discussions over different interpretations of the Presocratic philosophers. The only 'consequence' is satisfaction of the intellect.

It seems to me that the term 'assessment' ultimately has to be understood in a normative sense, and has to concern how we decide which theories to accept (as true) and act on, where appropriate. Popper has an important point here, that we test theories in order to reject them. But you wouldn't assume the truth of a theory just because it has so far passed all the tests. You wouldn't build bridges or rockets on the assumption that the theory is true, until you've done a lot of testing. Hence the point about marriage.

I don't see from what you have said that inference to the best explanation is a 'distinctive way of assessing scientific theories'. What it is, is a useful way of collecting together all the considerations that are involved in generating, selecting and testing scientific theories. As a contribution to the philosophy of science, it is neither descriptive or normative but aims to be both, in a weak sense. Not all scientific practice (as you note) conforms to the IBE model, nor, as I have argued, can you derive norms from what is basically an aesthetic notion of that which pleases the intellect.

I think an examiner would conclude from your essay that you had a good grasp of this topic, and would note that you have made some points of your own. I liked the way you attempted to analyse different notions of being 'distinctive' and also liked your point that different considerations arise when we are concerned with 'big theories about the fundamental forces of the world', or when we are doing 'normal' or everyday (bridges and rockets) science.

All the best,

Geoffrey