Kyle Stanford from UC-Irvine was here for a visit this past weekend, and talked about his novel argument against scientific realism. I thought there was a problem with the argument, but I now think maybe not.
Realism, here, is the epistemic thesis that we have good reason to think that the our present theories in the more mature sciences are true (or, at least, that the terms for unobservables in such theories succeed in referring). Kyle’s concern about this thesis derives from the fact that scientists are not that good at conceiving of all the alternatives to the theories that they propose.
On one level, this point is obvious: Newton didn’t, and in some interesting way, couldn’t conceive of quantum mechanics. The realist has a rejoinder that Kyle was willing to grant: exceptionalism. The exceptionalism response says that we now have a mature science and that the explanatory successes are sufficient to undermine the inductive argument from past failures to conceive of alternatives. In short, our situation is different from the situation of past great scientists, and so the induction fails.
Kyle’s response is to consider what he calls “close call” situations–cases where the alternative theories are readily available and well-confirmed by the evidence. So what he’s looking for are historical cases where a famous theorist seems incapable of formulating an alternative to their present approach to a problem.
One example Kyle used involves theories of inheritance, and relationship between Darwin’s theory of pangenesis and Galton’s alternative stirp theory. What’s interesting is that Darwin’s theory is a causal chain theory, and Galton’s a common cause theory, but Darwin simply couldn’t grasp Galton’s theory, even when he read Galton’s paper and corresponded with him about it. He keeps saying he doesn’t get it, doesn’t understand, and he never got the idea of a common cause alternative to his own theory. (This reminds me of some of my old professors–ossification of the mind perhaps, and an inability to comprehend new approaches. The sad part is that when you get to this point, you can’t detect it because once you get there you’ll think that it’s the result of the younger crowd just not being as good…and each side’s ad hominem cancels out the other…)
The correspondence is fairly compelling here–Darwin just didn’t get it. So Kyle has one example of the sort he wants–a “close call” situation where alternatives remain unconceived.
What I wanted to know about is how to test the unconceived alternative hypothesis. I understand what a confirming instance will look like–that’s pretty clear in the Darwin/Galton case. What I want to know is what a disconfirming instance will look like.
When we don’t find the textual evidence for unconceived alternatives, what conclusion will we draw? I expect we’ll just move on to the next case and hope for better results there. But we really need a better methodology than that. We need to know whether it is in general true in close call situations that the alternatives remain unconceived or whether they sometimes remain unconceived but in many cases are thought of but rejected for whatever reasons.
So my question concerns how to deal with cases where the record doesn’t substantiate the case as one involving unconceived alternatives. There are two possibilities that we need to be able to differentiate: one where there’s just no evidence about whether there are unconceived alternatives and the other where there is evidence that the alternatives are conceived. My worry is that every case will be categorized as the first sort if we don’t find evidence of unconceived alternatives. After all, what would a case have to look like for it to be a case where all “close” alternatives are conceived? We can describe what that would be in the abstract but with the vicissitudes of historical record, I doubt we’ll ever find such cases, but that’s no reason to think that there aren’t lots of them.
But perhaps Kyle doesn’t need a preponderance of cases in which unconceived alternatives remain. Perhaps all he needs is enough cases of that sort to undermine our confidence that there is a reliable connection between being a great scientist and having conceived of “close” alternatives. If that is all that is needed, Kyle’s job is easier, since it doesn’t take that many case studies to undermine our confidence in the reliability of this connection. There’s still a hint of a problem, but not as strong, since it is much easier to convince us that there aren’t enough gifted scientists than to convince us that most scientists are not gifted in this way. That is, we don’t need evidence for ungiftedness, we just need a defeater for giftedness.
So the moral of the story is this. Instead of describing the argument for denying realism by saying, “Scientists, even the very best, are not very good at conceiving of all the alternatives to their preferred theory, even when we restrict the alternatives to “close” ones,” the argument is more sympathetically interpreted as follows: for scientific realism to be sustained, we need a reliable connection between being a great scientist and conceiving of all the “close” alternatives, and there exist enough case studies of great scientists who simply lack this ability to cast legitimate doubt on the claim that we are justified in believing that our mature theories are true. So, we shouldn’t be scientific realists. The argument doesn’t need to strong thesis that scientists are not very good at conceiving of alternatives to their preferred theory, just that they are not good enough at it to sustain the epistemic thesis that we are justified in thinking that our present mature theories are true.