Philosophy of Science, Who Needs It?

This post at Siris, and the post it links to at Pharyngula, reminded me of my first methods course, lo these many years ago. The first week or two of the course were spent on the "philosophy of science" and the "scientific method," including a lot of definitions of important concepts like "fact," "theory," "law," "hypothesis," and "model." After those first two weeks, the only concept we actually used in discussing research methods was "hypothesis," and we used that concept in both a very loose and very specific sense: what you start with when designing an experiment. A hypothesis, which in the beginning of the course had a very exact meaning in relation to those other concepts of the scientific method, suddenly became just a position statement of one sort or another, of some level of exactness, that you then tested with whatever study you were conducting.

Later, when I took a philosophy of science course, we spent a lot of time talking about falsification. This was a word I had heard often in the methods courses I had taken (by this time, I'd taken 2 or 3). First, I learned all about Popper's concept of falsification, and then I learned that falsification had some serious theoretical problems, as the Quine-Duhem thesis had shown. Apparently individual hypotheses couldn't be falsified. This would seemingly pose a problem for all those scientists using the methods I'd learned in my methods course, because those methods were all about falsifying individual hypotheses. That's what those methods do. This was disconcerting to say the least.

Finally, I started working in a lab as an undergraduate research assistant, and talking to the researchers, grad students and faculty, about all of this confusing stuff I had been learning in my methods and philosophy of science courses. From them I learned that actual practicing scientists, by and large, don't know much about philosophy of science (several had never heard of the Quine-Duhem thesis, though they all knew who Popper was). It was at this point that I had an epiphany: scientists just care about doing science, and since it turns out that doing science doesn't require knowing much, if anything, about the philosophy of science, most scientists don't really care about philosophy of science. "So you say we can't falsify individual hypotheses?" a scientist might ask, and when the philosopher of science replies affirmatively, the scientist will say, "Well, acting as though we're trying to falsify individual hypotheses has gotten us pretty far, scientifically, so phooey on you!" The scientist will then conduct his or her next experiment designed to falsify an individual hypothesis.

Which brings me to the question, who needs philosophy of science? Now, I will admit having read some very interesting papers on the philosophy and history of science. They've often helped me to place certain scientific ideas and findings in their historical and theoretical context, particularly when those ideas and findings are in sciences far removed from my own knowledge base (e.g., physics or microbiology). What's more, I've read some interesting papers by philosophers attempting to apply approaches from one science to another. For example, a couple years back I read a paper about the role of causation in physics, and the implications for the role of causation in the social sciences. It was really interesting stuff. I've yet to hear of anyone applying it, in any way whatsoever, to their work in the social sciences, but it was still a fascinating read. But finding something interesting, or learning something about another science (that I probably could have learned by reading a book about that science, but I'm lazy), isn't quite the same as actually being relevant to science or scientists and their work. So I wonder, aside from a few exceptions (Popper's falsification being the obvious one), to what extent does the vast majority of philosophy of science matter to the doing of science? I doubt it matters very much. And if it doesn't really matter to scientists, to whom does it matter?