To finish up: Weinberg sets out the problem of explanation as a question of fundamentality, generality and derivation. Are Newton’s laws more fundamental than Kepler’s? Maybe not, if the laws are derived from Kepler’s orbits. And Kepler’s orbits, he points out, apply to electrons where Newton’s don’t, so we can’t easily determine which are more general.
I’m suggesting that these are not the right questions at all. It’s a question of, well, questioning. When the purely descriptive inductive inference is falsified, there is no question to ask. You can’t say, why did the key fail? You’ve simply got a statistical degrading. But if you’ve got a theory about the machinery, then you can ask “what went wrong with it?” Then you can fix it. That’s what “why” is all about. How to fix.
When you have a background hypothesis that is more than just a statistical correlation among phenomena, that’s when you can ask, “why?” Conversely, if you ask “why?,” you’re appealing to some hypothesis beyond the statistical description. The two approaches, empirical and generative/law-governed, give very different answers to “how does it work?” One leads to a successful fix, the other to an inferential failure.
To be fair, Weinberg recognizes up front that explanation is part of what science does, and he criticizes those who try to eliminate explanation by reducing it to mere description general or description fundamental or description derivational. As he says, it’s not for scientists to redefine our common language words. If a scientific explanation happens also to be a description, then it is both a description and an explanation: a description of a broad range of phenomena that the description classes together; and, sometimes just by virtue of classing them together, an explanation of those phenomena. No reason for description and explanation to be mutually exclusive.
But I think that those explanations that do no more than categorize — this flower inclines towards the sun because all flowers incline towards the sun — may be explanatory, but it’s little more than a broader inductive generalization. It gives a sneaking feeling of question-begging. Why does this flower incline? Because all flowers do. But why do all flowers do? Somehow, the inductive generalization hasn’t got beyond just describing. Why does this flower incline? Because it’s a flower. Why is it a flower? The answer to that will likely be genuinely circular unless there is a deeper hypothesis, like photosynthesis, than “flowers incline to the sun.”
I don’t see why it would matter whether photosynthesis is more general than inclining, or whether scientists discovered photosynthesis by deriving it from inclining, or whether inclining is more fundamental than photosynthesis. What’s important is that the explanation appeals to something other than flower behavior. Photosynthesis is a phenomenon independent from the flowers. It’s not about the mere inclinations. The science of photosynthesis regards a separate theory, and that theory added to the knowledge of flowers predicts the behavior the empiricist observes among the flowers. But the empiricist doesn’t observe photosynthesis directly. Photosynthesis does not logically require inclinations at all. It’s about sunlight and chemical transformations.
A true non-question begging explanation has to appeal to a theory that is separate from the mere observation.
Now, the curious may ask, so why is there photosynthesis? And there should be a further answer reaching back into genetics, and if the curious ask about that, there should be a further answer reaching even further into basic chemistry and physics, and, if we keep asking, we may get to a final answer about the fundamental physical “laws of nature.” And it’s likely that someday we won’t be able to ask “why” any further. But it won’t be because we don’t know how to look beyond. It’ll be because the universe is just so and there isn’t anything beyond. Maybe.
Will that end in circular answwers and question-begging? I guess, but at least it won’t be our fault.