In the scientific realism vs instrumentalism debate, realism is the position that the elements of a scientific theory represent reality. So when general relativity talks about space warping, space really is warping. Instrumentalism, or anti-realism, is the stance that scientific theories are just prediction mechanisms, with no guarantee that they represent reality. Under instrumentalism, general relativity accurately predicts our observations as though space were warping, but whether it actually does or not can’t be determined.
Scientists, by and large, tend to be realists. It’s hard to find motivation to do the often boring and sometimes dangerous work involved in gathering scientific data, to dedicate years of your life to it, unless you see yourself in pursuit of truth. But as I noted in our last discussion on this, scientists tend to be realist about some theories and instrumentalist about others (although which is which depends on the scientist).
The argument in favor of instrumentalism is theory change. Many historical theories have been successful at making predictions, but eventually end up being replaced by a better theory, often with a radically different view of reality. The example I usually cite is Ptolemy’s model of the universe. For centuries it more or less accurately predicted naked eye astronomical observations, but we now know its model of a stationary Earth, with everything else in the universe revolving around it, is wrong. It eventually gave way to a Newtonian view of the universe, which in turn later had to give way to an Einsteinian view.
Realists often respond with the no-miracles argument. If scientific theories do not reflect reality to at least some approximation, then the argument is their ability to reliably make accurate predictions amounts to a miracle. Under this view, theories may change, but each one is getting us to a closer and closer approximation of the truth. This argument seems bolstered when we consider that often multiple theories are consistent and can be reconciled with each other.
As I’ve noted before, I tend to think this debate is misguided, because what we call “real” or “truth” are themselves just another predictive model. Still, while emotionally I’m a realist, I’ve long had intellectual sympathy with instrumentalism, although I prefer scientists to aim for realism and only settle for explicitly anti-real theories when there’s no choice. All in all, I’d prefer to be a realist, but wonder if there’s an intellectually honest way to get there.
Maybe there is. I recently came across the concept of structuralism or structural realism. The idea starts with the fact that what tends to survive theory change are the mathematical structures of the old theory. More specifically, these structures still tend to be useful in whatever domain the old theory was reliably accurate in. The prime example here is Newtonian physics, which gave way to general relativity and quantum physics in the early 20th century. But Newton’s mathematics remain so reliable for most situations that they’re what NASA uses for most space missions.
Under structural realism, it isn’t necessarily the concepts discussed in the theory descriptions that are real, but the structures and relations revealed in the mathematics. There may be multiple possible underlying realities from which these structures emerge, but the structures and relations appear to be what endures. If so, a philosophy of realism that focuses on the structures seems much more defendable than traditional realism.
There are multiple types of structural realism discussed in the articles linked to above, but they seem to fall into two broad groups: ESR (epistemic structural realism) and OSR (ontic structural realism). ESR says that all we can know about entities in the world, “the furniture of the world”, are their structural relations. It doesn’t rule out that there might be intrinsic non-structural and non-relational aspects of these entities, just that we can know anything about them.
OSR on the other hand, asserts that all there is to the entities are those structures and relationships. OSR apparently gets support in the fundamental physics community, particularly from researchers working in quantum field theory.
When considering these two views, ESR seems like the more defendable position to me. But I can see the argument from the OSR advocates. If there are intrinsic non-structural non-relational properties or aspects of entities in the world, how would we ever know about them? If they have no structural relations with anything other than the entity itself, how could we ever interact with them to observe them? Can they exist for us in any meaningful manner?
It’s worth noting that many panpsychists make an appeal to these intrinsic properties, arguing that they could be what provides consciousness to matter. In fact, it was Philip Goff’s description of causal structuralism, and his arguments against it, that first alerted me to the concept of structuralism. Ultimately the appeal to intrinsic properties seems to imply an epiphenomenal type of consciousness that seems problematic.
The motivations of the OSR advocates seem to be to close off these kinds of discussions. But the distinction between ESR and OSR strikes me as utterly metaphysical and unknowable, which I think defaults me into the ESR camp.
There are a number of criticisms of structural realism discussed in the SEP article. Two I find notable. One is that it implies a sort of Tegmarkian (mathematical universe hypothesis) view of reality. But this doesn’t seem right to me. Structural realism doesn’t seem to be arguing that all structures modeled by mathematics are real, only ones that have some reliably causal relationship with empirical observations.
The second one appears to have more bite. It’s that, strictly speaking, structures are not preserved across theory change. In reality they’re approximately preserved. But I think when we consider that the typical realist position isn’t that scientific theories are absolutely perfectly true, only that they approach ever closer approximations of truth, this criticism loses most of its bite. But I don’t know that it loses all of it.
Consider the examples I cited above. Arguably the structures of Newton’s theories are preserved well enough that we can say they mostly reflect reality, again to the extent they remain useful for NASA. But can we say the same thing about Ptolemy’s model of the cosmos? We could argue that his structures are preserved, but only from the standpoint of an observer on Earth, a pretty limited scope.
No one at NASA with their head screwed on straight would use any of Ptolemy’s structures in planning a mission. A structural realist could argue that the Ptolemaic model was a pre-scientific one, that well tested scientific theories are unlikely to be as far off. But that assumes we may never be as far beyond our current theories as they are beyond the Ptolemaic system. It seems like an example that should make us cautious.
On the other hand, I don’t know that we should expect complete perfection in these outlooks. A philosophy of science view, to be useful, arguably only needs to provide useful heuristics, assumptions we should still always be prepared to revise. In that sense, structural realism may be reliable enough for most purposes.
What do you think? Does structural realism make scientific realism a more plausible position? Or does it sacrifice too much to retain the “realist” label? Is the remote possibility of major structural change a fatal flaw? Or should we just stop letting the problems of fundamental physics constrain our philosophy toward the rest of science, where traditional realism is a much easier proposition?