To what extent are the various methods of science rationalist rather than empirical?

Question

I know something about the vigilance in which science is considered empirical, especially in analytical philosophy which is deeply rooted in logical positivism and British empiricism. Yet, I can't help but wonder how a lot of the sciences seem to actually make fruitful use of thought experiments and scientific models to come to conclusions: We discuss objects like black holes, and electron orbitals, which are hard even to imagine, let alone measure or observe.

Is it fair to say that modern science, while grounded in empiricism, observation and measurement, also meets many of its claims using the methods that seem closer to the rationalists in the history of epistemology? Or can the definition of "empiricism" be extended to account for the evidence we have of scientific models like space-time, light as waves or light as particles, and entropy? If so, how would you define empiricism to account for this?

Answer 1

You're not doing empirical science if the objects postulated by your models are beyond hard to measure or observe, even indirectly. Given all the layers of indirection between e.g. a chair and your awareness of it (hi photons, photoreceptor cell voltage, action potentials in retinal ganglion cells, depolarization of simple cells in V1, ...), we shouldn't be the least shy of indirect observations.

The reason why special and general relativity and quantum mechanics are so successful is precisely because they make oodles of predictions, some of them very weird, and measurements have confirmed those predictions over and over again.

And inasmuch as there is a crisis in string theory (not really, but neither is it entirely comfortable right now), it is because it isn't really science at this point; people have had incredible difficulty narrowing down the elegant mathematics to testable theories. So it is certainly mathematics, and inspired by physics, but a lot of scientists (e.g. me) reject that it's science yet. Everyone is simply trying to do science when coming up with hypotheses and models, but you've got to see it through to finding telling agreement with observations before it actually meets that bar. It's just really hard to close that loop in some fields; when you fail, you've also failed to advance scientific knowledge.