Theoretical Changes to Structural Parts of Scientific Theories
Another line of criticism of OSR concerns its claims that structure is so clearly discernible in scientific theories and that it is preserved across theory changes. Ontological inventories are only pragmatic devices, according to OSR, and can be replaced. The structural parts of theories, in contrast, are preserved as true or approximately true parts of mature scientific theories. This view admits at least the following objections.
In the first place, content and structure are not easily separated. It seems arbitrary to accept theoretical losses regarding non-structural components of theories and yet deny them regarding the structural components. This assumes that structure and content can be divorced: a view that we think ultimately untenable.34 To state it bluntly, structure without a bearer is a mere phantom; a bearer without a lawful structure is an alien, permanently isolated and impossible to know. Additionally, if we are looking to the inclusion of new discoveries and innovations within the explanatory domain of our theories, then we cannot be open to change in ontological inventory without also being open to a parallel change in theoretical structure.
Consider, for instance, the psychological experiment discussed by Kuhn. Subjects were asked to identify playing cards, one by one, after relatively short exposure. Most cards were ‘normal’ but a small number of them were ‘anomalous’, such as a black seven of hearts or red two of spades. Significantly, the anomalous cards were consistently but incorrectly identified as normal: as a red seven of hearts, for instance. The experiment showed that our first and natural response to new data is to be conservative and offer some resistance: to try to preserve the old theory. But once the anomaly is detected and gradually accepted, how can the ontological inventory be changed without the modal structure? One has to recognize that the structural components of the ‘normal’ theory weren’t mapping the real structure. The ontic structural realist might respond by saying that ‘cards’ are not structural elements but ontological inventory. However, you can adjust the experiment by introducing the rules of a card game as the theoretical structure that is satisfied only by the modal structure of normal cards, which are for OSR mere placeholders of the mathematical relations that constitute the game. If so, the problem returns again.
If we are really open to admitting novelty in scientific discovery we must be open to admitting placeholders like the anomalous cards within the explanatory scope of our theories. This means, assuming the commitments of OSR, not only admitting a completely new pattern of relations, but also recognizing that, after all, the old theoretical game wasn’t mapping the real game. So the question still is pressing: once the anomaly is detected, how can the ontology be changed without the modal structure? In the case of OSR the question seems more dramatic. Since it is committed to the idea that the world is not a mere extensional list of relations but an ontologically interdependent structure, it is hard to see how structural components of theories can be preserved when a new discovery is supposed to reveal a novelty in the ontic (relational) constitution of everything previously known. Theoretical holism is a natural response if one is committed to some type of relational-ontic holism. If so, old structural-theoretical components cannot be preserved and remain unaffected by new discoveries. More than any account, OSR should accept that scientific theories fail as a whole when anomalies become accepted.
In the second place, whether scientific theories suffer theoretical losses in their structural components is not something to be decided a priori or by mere postulation. It thus seems adventurous to claim, as Ladyman and Ross do, that ‘In sum, we know that well-confirmed relations among phenomena must be retained in future theories’. On the contrary, this must be an a posteriori matter to be settled by science or the history of science and not in advance by pure philosophical speculation. If we are to remain open to the possibility of novelty and discovery, then we ought to allow that theories can start to lose adequacy as a whole, which would include their structural elements. To be clear, it might well be that necessarily the world has a certain structure, which may be discovered a posteriori; just like Water=H2O might be a necessary identity which can be discovered a posteriori, in such terms that if it is the case, then it necessarily is the case. But the question is whether we have any evidence for claiming that our current mature scientific theories actually do map that allegedly really existent structure in such a way that we are also entitled to claim knowledge about what future scientific theories will and should retain (analogously, the question is whether we have any reason for believing that our current theory of water does get the essence of water right). And these latter claims seem to jar with a purported respect for the labour of natural scientists. After all, there is empirical evidence to the contrary.
Stanford offers the example of Galton’s law of inheritance, which was used to explain the proportion of contribution from ancestors. The structure of this theory has been abandoned in contemporary genetics, which takes Galton’s law as simply the wrong mapping of the structure of inheritance. Ladyman and Ross are aware of the case but evade the challenge by claiming that scientific realism is more exposed than OSR to the problem of theoretical changes.38
Stanford’s case is not a one-off. Laudan claims that the history of science reveals several instances. The general contention is that the structural components of scientific theories are just as subject to change as the ontological components. Laudan argues that laws, entities, mechanisms, and even observable regularities that figure in scientific theories can be abandoned:
Copernican astronomy did not retain all the key mechanisms of Ptolemaic astronomy (e.g., motion along an equant); Newton’s physics did not retain all (or even most of) the ‘theoretical laws’ of Cartesian mechanics, astronomy and optics; Franklin’s electrical theory did not contain its predecessor (Nollet’s) as a limiting case. Relativistic physics did not retain the aether, nor the mechanisms associated with it; statistical mechanics does not incorporate all the mechanisms of thermodynamics; modern genetics does not have Darwinian pangenesis as a limiting case; the wave theory of light did not appropriate the mechanisms of corpuscular optics; modern embryology incorporates few of the
mechanisms prominent in classical embryological theory____ [L]oss occurs at virtually
every level: the confirmed predictions of earlier theories are sometimes not explained by later ones; even the ‘observable’ laws explained by earlier theories are not always retained, not even as limiting cases; theoretical processes and mechanisms of earlier theories are, as frequently as not, treated as flotsam.
Thus, if mature scientific theories have lost structural components in the past, on what grounds can we claim that the structural components of our current mature scientific theories do map the real structure and must be retained in future theories? Why should we believe this when past history shows persistent theoretical changes at every level? The most reasonable attitude is to expect ongoing and fallible scientific progress. Why not? Persisting in the contrary claim imposes an unnecessary dogma over a type of inquiry that by its very nature resists such constraints.
Clearly we are suggesting that in defending the idea that modal structure is retained through theoretical changes, Ladyman and Ross fail to honour their own putatively naturalistic approach to metaphysics and philosophy. Paradigmatic- ally, scientific models use the tools from mathematics that are available at the time of their construction. But mathematics, like every other type of knowledge, belongs to our ‘web of belief ’ and should be revised if experience demands it. We are admittedly very conservative about mathematics and logic but this is because, as Quine thought, these disciplines are at the very centre of our web of belief. However, while we will always be reluctant to revise them if something less central can be replaced instead, they are not untouchable. Even the law of excluded middle may be revised in theory. There is no difference in principle between archaeology, for instance, and mathematical knowledge. The only difference is one of degree. All disciplines face the tribunal of experience.
If neither logic nor mathematics is immune from revision, then why should we think these imperfect tools worthy of preservation through theory change? As has been argued by Hitchcock and Mumford and Anjum,41 the tools available for representation can condition our understanding of the object represented. And if those tools for representation might change, then so might the way that we conceive and understand the structure represented. There can be no guarantee, therefore, that new mathematical tools will give us the same structure. Hence, there seems no basis for such confidence that structure will in future be retained through theoretical change.