Philosophy of Science Work

It was a cold and almost frosty morning. The heating comes on slightly later at the weekend and the house was still cold as I started work - sorting out the study plan for today and tomorrow. Mostly reading and taking notes on ideas on confirmation and more reading of Kuhn on Copernicus.

But first a trip outside to take some photos of the spiders' webs that are highly visible this morning. Not sure how good the pictures will look on this blog, but they look great on the PC.


And so to work - a day spent virtually entirely at my desk in our home office shown below. What a lot of books I seem to have (and you can't see the wall of books behind where I took the photo from!)


Most of the morning is spent on Howson's Hume's Problem: Induction and the Justification of Belief. This is a very interesting book in many ways. One the one hand, it believes that Hume's argument is correct (which I also think). But then seeks to argue that inductivism (as formulated within Bayesianism) is actually part of logic, following a suggestion from Ramsey that the laws of probability are laws of consistency. Hence inductivism occurs as part of remaining consistent about the degree of belief in hypotheses subject to changing evidence. Not sure I accept that.

I think large problems remain - in particular, I don't accept the view that science shows the success of inductive reasoning. Unless induction means something like the realist assumption of the regularity of nature, then I think this is simply untrue. This links back to the work I've been doing on Lakatos's meta-methodology for characterising methodologies of science. The sort of science we would expect to see if inductivism was the main methodology of science is the steady accumulation of generalisations covering wider and wider areas. What we would not see are radical revolutions that overthrow previous theories. For this would suggest that past theories were incorrect inductions and it is not clear how these could occur if the method of science generally is inductive.

My own view is based on the "no-miracles" argument for realism combined with a naturalist explanation via evolution and the acceptance of much of the Taleb thesis in The Black Swan. So the real world is a regular world, which is why some of our generalisations hold. But inductivism implies a false justificationism which will often prove incorrect in practice.

A break just after lunchtime to watch some football and try to avoid getting distracted by the new John Peel book that arrived this morning - a selection of his writings from the last 30 years of his life. I still really miss John Peel. Listening to some Peel show tapes recently has only emphasised this.

More work through the late afternoon - still focused on Howson's book. Now focused on probability approaches to inductivism.

It occurred to me that one way to re-cast some of this discussion would be to consider (again) the question of what picture of science would be generated from the model, were it to be correct. Howson has an annoying habit of simple asserting that inductive arguments are there in science all the time (a bit like Salmon in this respect and perhaps going back to the Broad quote about induction being the pride of science . . .). At a meta-level, we might suggest the "test" of inductivism by the assessment of whether, over time, and relative to the history of science, it does tend to produce the result that false hypotheses are probably false and that true hypotheses are probably true.

In this way, the problem of revolutions in science becomes more apparent. For any revolution is a case where inductivism is shown to have considered false hypotheses to be probably true - this is also closely linked to Taleb's turkeys!