The Tools of the Dominant Strategies
We have yet to consider whether Bernstein is right to think that some of the key concepts of powers-based ontologies are underdeveloped, such as properties and substances, and that they would therefore benefit from using the tools of the more dominant strategies. Or, in other words, would a connection to the traditional neo-Humean framework ‘open up’ greater theoretical resources for both sides? I have doubts about that. Sure, there are problems/controversies about most if not all the core concepts of powers-based accounts. However, those problems are related to an attempt to develop concepts that play some explanatory role. The neo-Humean framework avoids those problems by stripping the concepts of their explanatory roles, therefore appearing unproblematic (rather than ‘developed’). In most if not all cases, it reduces the problematic concept to some other less problematic concept, but usually at a cost with respect to explanatory power.
Take the paradigmatic neo-Humean conception of property. At the most fundamental level we have categorical properties. They are causally inert qualities, useful for not much more than serving a role as abstract identifiers in conjunction with Leibniz’ law. I say abstract, because although they were once thought to be observable (see, for instance, Goodman 1954: 40) they are now generally assumed to be unknowable. Only neo-Humeans who are also direct realists (if there are any) will disagree. The reason they are supposed to be unknowable is because they are only contingently associated with the second-order powers that affect our senses to inform us about the first-order categorical properties. Since the powers are in turn only contingently related to the sensory qualities that they give rise to, there is no way to tell whether the way a property appears in experience is the way the property really is. This is a strange situation for a type of property that is supposed to ground the identity of objects. The consequence is that we cannot in principle know the identity of anything (not even of properties), and yet we can supposedly know a priori that it is necessarily true that if a and b are identical, they have all properties in common. Leibniz’ law is thus turned into a piece of abstract trivia that can at best be applied in purely formal models.
Similar things can be said about the concept of substance. For an Aristotelian, the concept of substance, with the distinction between form and matter, has a certain explanatory role. It explains why there is always something that remains through any kind of change. The reason Chrysippus remains the same as he saunters from the gymnasium to the market is that he changes in accidents (qualitative change) but neither in essence nor matter (numerical identity). On the neo-Humean account of properties and how they ground identity, identity and change are a priori incompatible. This point might actually be one the most important things I have ever tried to argue for in print, and I owe it partly to J.M.E. McTaggart. He realised that, if the identity of things is grounded in its qualities, then whenever there is qualitative dissimilarity there is numerical distinctness, and vice versa. Which is why he thought Leibniz’ term ‘indiscernibility of identicals’ was a misnomer; it should be called ‘dissimilarity of the diverse’ instead. He didn’t connect this to the issues we are dealing with now, but the connection is straightforward. And it is very clear that the problem of temporary intrinsics, as formulated by Lewis, is a case where we arrive at the same conclusion; nothing can remain numerically the same unless it remains qualitatively the same. Ergo, change is impossible.
I think this should be enough to show that the sentiment that core concepts of powers-based ontologies are underdeveloped can be understood only as the sentiment of those who only want to describe reality but not to explain it, and are bothered by the extra complications that inevitably accompany the explanatory project. Again, the friction between the neo-Humean framework and powers-based accounts (or causal realists generally) revolves largely about meta-philosophical assumptions about the purpose and/or limits of metaphysical thought, and not specifically about the nature of causation, properties, etc.