The trick to understanding the title of this book is to see the space between “every” and “thing.” It is not advocating throwing out all of philosophy, just our usual notion of a “thing.”
I actually own a physical hard copy of this book. Someone asked me at Christmas what book I wanted for a present, and this was the only one I could think of off the top of my head. I have been gnawing my teeth at it, ever since
The book is known for introducing what it calls “ontic structural realism,” and my interest in that stems from the word “structure.” I am always talking about how energy makes arrangements of itself, and so I want to compare “arrangements” (as in thermodynamics) with “structures”(as in philosophy).
There are, however, multiple usages of the word “structure,” depending on the discipline using it. In the philosophies of “Structuralism” and “Post-Structuralism,” the subject is mostly society and how it is organized (into structures). It was a rage in the 1960’s as evidenced by Kuhn’s famous book The Structure of Scientific Revolutions. I am not going to go here into French philosophy except to point out that in these philosophies a structure is an abstraction. We cannot reach out and touch it, even with instruments.
And another usage of the word “structure” occurs in mathematics where, again, a structure is arguably abstract (as many would claim all of math is).
What Ladyman’s book is about is applying similar arguments to physics. And so immediately we can see the issue before us. Is physics about abstractions (rather than about physical objects)?
Ladyman argues that structures are more real than physical objects (which is why we need to reconsider, he argues, our concept of “things”). And why would he want to argue that? Well, that gets into the part about “realism.” The argument comes down to recognizing how science often changes its mind and how that can be embarrassing to realists who had been claiming that the old theory being superseded had been ultimately real. So the argument in structural realism is that, even though science might change its mind, the underlying abstract structure of what it is talking about does not change. It survives untouched by the revisions.
Thus the abstraction (the structure) is said to be more real than the physical objects themselves.
And if we stop to think about it, that is a familiar contention in philosophy. Much of Medieval philosophy was an argument over which is more real, the Universal (the abstract generality about some particulars) or the physical particulars. Also, Plato had his otherworldly Forms which were said to be more real than the physical world.
So now Ladyman is updating that with abstract structures which are ultimately more real than physical objects.
two kinds of structural realism
Structural realism is a relatively new idea invented by Worrsall in 1989. It appeals to Ladyman, but he has found several problems with its original version, so he has updated it. The old version he calls “epistemic structural realism,” and his new version he calls “ontic structural realism.”
Both versions hold that structures are abstract and more real than physical objects (which must be the case for the structures to survive the changes in how science describes physical objects).
But the difference is that the epistemic version utilizes science as a methodology, whereas the ontic version holds that, in order to arrive at abstractions, we have to do so with metaphysics, not with science and its study of physical things. Writes Ladyman, “There are no things. Structure is all there is” (p. 130). He adds, “There are objects in our metaphysics but they have been purged of their intrinsic nature, identity, and individuality, and they are not metaphysically fundamental” (p. 131) because, instead, abstract structures are fundamental.
If that sounds extreme, it is no more extreme than saying that Universals are more real than particulars.
I will recap Ladyman’s arguments against epistemic structural realism later in this review.
Personally, I am not averse to talk of structures (because of their similarity with arrangements). But I hope to explore the difference. And Ladyman’s claims that only abstractions are real invites a lot of questions, such as the following:
How does causality arise, if only abstractions are real?
Do we ever directly experience a structure? If not, is it only a concept? How do we know?
How do we get from abstractions being real to verification via physical testing?
And are abstract structures just a means of reintroducing essentialism into philosophy? (That argument historically resulted in Post-Structuralism).
But the really big question is this: What is going together to make the structure? What is the structure made of, if it isn’t physical, and it isn’t just a concept, and yet it is ultimately real?
Another way of looking at a structure is that the structure is telling us how things are related. They are related as per being together in the structure. But then what are these “things” being related? What are the relationships among?
That is another way of asking what is going together to make an abstract structure. We can see how that might work with numbers. Each number is itself an abstraction and yet it can still have abstract relationships with other numbers. But how does that work with physical objects? If the structures are abstract, as Ladyman claims, then it is not physical objects that are having the relationships.
Ladyman can answer that the relationships are among objects which themselves consist of an organization of relationships. Then what might seem to us to be physical objects are really just abstract structures made up of other structures. And so it is all the way down. At any level of scale, a structure is made of still smaller structures each made of woven abstract relationships.
Yet that would seem to be an infinite regress. Do we ever get to the point where we learn what is being fundamentally related?
Ladyman has an answer.
At this juncture, however, I have to interject that to my mind Ladyman argues by obfuscation. He states very clearly what he wants to prove, and he provides very detailed explicit definitions. But then he wanders off and starts talking about what seems to be irrelevant. And on these other subjects he might be perfectly right. But he never connects the dots and tells us how these other subjects prove his main contention. He can be very frustrating to read.
What the reader has to do is guess what he might mean and then see if that seems to check out as the reader keeps slogging through. It leaves a person wondering how much of his argument he has really thought through.
So the following is my own reconstruction of what I think he might be saying. I consider it to be a fair reconstruction—perhaps it even makes more sense than his original—but I wouldn’t be surprised to find out later that he ends up telling us he means something completely different.
So here goes:
I think Ladyman is suggesting that there is not an infinite regress of abstract structures made out of more basic abstract structures because eventually we come to quantum nonlocality. And at that point—when what an object even is depends on what is out of the picture—then it no longer makes sense to ask the question, “What are the relationships among?”
At that level, it’s all just a blur.
Then, once we have hit bottom, so to speak, we can start rebuilding complex abstractions out of these fundamental quantum obscurities. Or more exactly, we can build up the structures which give rise to what is studied in the other sciences. (Ladyman calls any science that is not quantum mechanics—even the rest of physics—the “special sciences”). It works, he claims, to build one structure out of more basic ones because of the so-called “unity of science,” which means that everything else is made out of quantum physics. So when we see a “thing,” it is actually a vast organization of abstract relationships made out of other abstract relationships, starting most fundamentally with the blur of quantum physics that is too nebulous to study with science (so we have to study it with metaphysics and his “ontic” reasoning).
To his credit, Ladyman does not so much make this argument in terms of nonlocality (which can be controversial) but in terms of the “quantum indistinguishability of individuals,” which I agree is right there in the math. It is an often-overlooked aspect of quantum mechanics, and Ladyman is right to bring it up.
It has to do with using eigenvalues and how that means that it becomes impossible to tell one individual from another individual from just the math. That is so even to the extent that two individuals might be identical but still different in the sense of being in different locations, and yet even that cannot be used to distinguish them, just from the math.
So Ladyman’s point is that, in our infinite regress, where one abstract structure is made out of other smaller and smaller abstract structures, we eventually get to quantum mechanics where we cannot tell one individual from another individual. And at that point, it makes no sense to ask what in an abstract structure the relationships are among. There are no individuals to have the relationships.
Science cannot study what is indistinguishable. So indistinguishability also rules out epistemic approaches to describing structure. Further, even the higher order structures that we call “things” are basically made of what is indistinguishable. So we are left with metaphysics telling us that everything is abstract structures.
Or at least that is what I think he is saying.
My own critique of this argument is as follows.
I think he is wrong to argue that the indistinguishability of individuals occurs only in the obscure portions of quantum mechanics. It occur all over the place in the special sciences, and yet that is no handicap to using measurement. For instance, it happens anytime that we use averages.
Consider a classroom test where we find that the average test score is 80. From just knowing the average, we still do not know what each student scored. There is no way to distinguish one individual from another (at least by just using the math). But this example also illustrates how individuals can exist physically as separate things—really—even if they cannot be perfectly described by some overriding math. Being a physical individual is not math-dependent.
So the indistinguishability of individuals in quantum mechanics is neither unique in science nor is it a way of “hitting bottom” in our infinite regress. It does not constitute a way of finally no longer needing to answer the question of what is being related in an abstract structure.
As for the “unity of science,” that is where supposedly the special sciences can be completely derived from physics. Or conversely, it is where the special sciences can be “reduced” to—or stated over again—completely in terms of physics, so that biology reduces to chemistry, and chemistry reduces to physics. The usual answer to that contention today is to dispense with the arguments and simply say, “Okay, try it.” Take any number of physics textbooks you like, and using only what you find in there, derive, say, Darwinism or recessive genetic traits. It cannot be done (and it has been tried by those very eager to succeed).
The reason it doesn’t work is that the world is not organized just vertically into categories, as in hierarchies, but also it is organized horizontally so that what is present on the same level of scale is a factor in what happens. We are not justified in skipping the contribution of these horizontal relationships, but that is what happens when we only acknowledge things in vertical hierarchies. My favorite example is to think of Johnny who knows everything about car mechanics. But that doesn’t mean that he therefore knows the route to Grandmother’s house. No, to know the route depends on other things existing on the same level of scale as the route, such as knowing about roads, bridges, intersections, and traffic signs. Or conversely, knowing about traffic signs does not tell us how to make a car motor.
So I think that Ladyman’s attempt to justify wholesale abstraction—to say that the physical is just an illusion—is not justified by his approach of breaking everything down to quantum ambiguities and then rebuilding the world hierarchically out of abstract relationships.
structure and realism
My own take on the issues raised by Ladyman (and I am glad that he has raised them) is that structural realism does not solve the problems it is intended to solve. I feel that structural realism is troubled, first and foremost, by the lack of a single example of how it might work in practice to address science changing its mind. Ladyman offers only one example in an entire book, and it is not really an example of incommensurability. His example is that of Einstein and relativity theory replacing Newton. But a closer examination shows that that is not really a contradiction; it is a special case. Relativity theory turns into Newtonian physics in the case when the velocity is significantly less than the speed of light. They are compatible theories.
Worrsall, at least, does offer a viable example of science changing its mind incommensurably. That is the Michelson-Morley experiment and how, prior to that, it was believed that outer space consisted of an ether, and afterwards it was a void. Those are two incommensurable conclusions. Worrsall argues that, even as the description of the physical situation changed, what stayed the same was the underlying structure that made it so that science was seeing waves in the light (the waves having previously been thought to be going back and forth in the ether). Yet that might be better explained in terms of reproducibility. What science could demonstrate repeatedly before the experiment (measure the waves) it could still do over and over again, afterwards. That is what survived the advance in scientific thinking, not some abstraction.
And I think that that is the general case. Wherever it is argued that what survives scientific change is structure, it is better seen that it is reproducibility that survives.
But perhaps there is some unseen and unknowable structure lurking behind the scenes that is producing the reproducibility?
Ladyman argues that unknowable entities have no place in either science or metaphysics.
So then that brings up Ladyman’s arguments against Worrsall’s epistemic structural realism. Again, he argues mostly with obfuscation, but I think he is trying to make two main objections.
The first is that epistemic approaches (using science to try to know abstract structures) leave out how other theories are incorporated into our understanding of the theory at hand. We can’t talk about E = mc2 without knowing the other theories that define energy and mas. So when we think we are testing to find a structure, we might really be just testing the glasses (the other theories and their assumptions) through which we are looking at it. We can’t be confident of knowing the structure itself using epistemic methods.
And the second argument concerns so-called “unobservables.” When science makes theories about what we can’t actually see, then we can’t genuinely know if an abstract structure is what is being described or not.
Again, Ladyman sees himself as rescuing structural realism with his ontic version, and it needs rescuing, he thinks, because we can’t reliably know about the intrinsic natures of abstractions by utilizing the methods of measurement. Writes Ladyman, “The epistemic structural realist thinks that all we can know is structure, but it is the structure of an unknowable realm of individuals. However, we will argue that… talk of unknowable intrinsic natures and individuals is idle and has no justified place in metaphysics” (p. 131).
But just when I think I might have Ladyman figured out, he throws a curve at us. He says near the end of his book that his real objective is to defend verificationism, as in the famous Vienna Circle and logical positivism, from the arguments brought against it by Quine in his influential paper “Two Dogmas of Empiricism” (1951). Quine’s paper is frequently credited with bringing logical positivism to an end, so Ladyman’s claim would be noteworthy if correct. But I am not going to go into that, since I feel that Ladyman’s arguments are inadequate in the first place.
I should also mention that Ladyman’s arguments can be taken to extreme, as in Tegmark’s claim that the Universe simply “is” the structure of abstract mathematics. I addressed Tegmark in an earlier essay called “Is Reality a Mathematical Structure? a home experiment to find out.”
Ladyman describes a progression that starts with van Fraassen and his theory of “constructive empiricism,” and Ladyman shows how epistemic structural realism was a reaction to that, and then the ontic version was a reaction to that. Constructive empiricism holds that science only seeks “empirical adequacy,” not truth. (Van Fraassen started the talk about “unobservables” and about how they limit what science can know about ultimate truth).
Here is Ladyman quoting van Fraassen (p. 95):
Van Fraassen defines scientific realism as follows: ‘Science aims to give us, in its theories, a literally true story of what the world is like; and acceptance of a scientific theory involves the belief that it is true. Constructive empiricism is then contrastively defined thus: ‘Science aims to give us theories which are empirically adequate; and acceptance of a theory involves a belief only that it is empirically adequate.’
As I have been arguing in the preceding two posts, I see the goal of science as being neither truth nor empirical adequacy but as finding reproducibility. It is by putting together repeatable knowledge that we can achieve reliability. But reproducibility is not just about repeating an action but about finding the situations—the structures?—that foster repeatability.
So is a setup of circumstances that enables reproducibility the same as a structure?
According to Ladyman’s definitions of “structure,” they are different. That is because he argues that his structures have to be “abstract” in order to survive the changes in how science describes the physical world. That is true of both ontic and epistemic versions, the difference being only in the approach to finding the abstract structures.
In chemistry, we talk about the structure of a molecule, but we mean it in a physical sense. Or in biology, the structure of a wing is likewise physical. But we also talk about the surrounding setup of circumstances (the temperature and air pressure), and that might be considered abstract in the sense that it is about the bigger picture than just the molecule by itself. A “bigger picture” might be considered abstract. And more to the point, a bigger picture seems to be made of things going together in an arrangement that might be called a structure.
But why should we have to choose between structures being 100% abstract or 100% physical? That is why I prefer the word “arrangement” over “structure” (if structures are to be by definition purely abstract). An arrangement can have both physical qualities (we can touch it) and abstract qualities (such as potential) at the same time. And an arrangement needn’t be purely abstract to survive the changes in how science portrays the physical world because it is reproducibility, not abstraction, that survives the changes (we can still do over again what we could do over again prior to the new theory).
And a second reason that a setup of circumstances is different from a purely abstract structure is that, as I described in the essay on Tegmark, the math used to handle measurements is not, when so constrained for that purpose, the same as the pure math of abstraction. That is why physics is a different subject from mathematics. There is more to the world than its mathematical structure.
Finally, a third difference is that arrangements change—they evolve—they are not about eternally unchanging structures.
And that’s not even to get into issues such as how abstractions can make physical causality.
I think that we can safely say that each scientific discipline has its own productive use of the word “structure.” But we get into trouble when we try to argue that there is just one definition of it that applies to all subjects. The context counts.
Yet I can still see why people with realist intuitions might find structural realism of some sort to be a good place for them to land (although I wouldn’t recommend either of Ladyman’s varieties, ontic or epistemic). If someone wants to say, “Arrangements are real, so structures of some kind must be real,” that is fine with me. It isn’t how I would put it—I’d state it as reproducible rather than as real—but at least this way I can agree with Ladyman that we do not need to have Platonic pure essences or Kantian things-in-themselves. We have things-in-structures. And in Ladyman’s words, that means that things are “purged of their intrinsic natures.”
Things are at least partially defined by the structures that they are a part of (in a kind of inverse atomism).
IMAGE: these are desert mountain flowers; we’ve had rain