Nominalism is a philosophical doctrine usually understood to entail a rejection of universals, in favour of the belief that only the concrete exists. Universals are understood as instantiable entities, i.e. something like types. Another flavour of nominalism involves rejection of abstracta, such as mathematical entities, propositions, fictional entities (including possible worlds). You may read about them in the SEP’s rather dull entry on Nominalism in metaphysics.
I personally think that most nominalist arguments are straightforwardly wrong, but not for the usual reasons that universals and/or abstracta are said by realists to exist, but for the opposite reason: types and abstracta are just there, even if they don’t ‘exist’, in the sense of being spatio-temporally concretised. The real problem is that we misuse the word exists at least half the time in philosophy. The way we should talk is to say things like: there are universals (clearly: we can talk of mammals, tennis matches and cabernet sauvignon), but they don’t exist as such; their qualities only exist in their instantiated particulars (your cat, a match between Federer and Djokovich in 2014, 2005 Chateau Lafon-Rochet, Saint-Estephe).
So that’s why nominalists are wrong. There are universals, but they don’t exist. Well they do, but they don’t. You know what I’m saying. (More boringly, we should all be meta-physical anti-realists, but still believe in universals. Got it now?)
Enough complaining about linguistic wrong turns. I want to show why nominalists are wrong even on their own terms, at least about most of their ideas. This came from a short email exchange with friend and colleague Barry Smith, professor of philosophy at U Buffalo, head of the National Centre for Ontological Research, co-creator of Basic Formal Ontology (BFO), and noted scientific realist.
I was reading one of Barry’s papers (with Werner Ceusters): 2010 paper entitled Ontological realism: A methodology for coordinated evolution of scientific ontologies (available here), in which they carefully sidestep any allegiance to metaphysical realism or otherwise by referring to universals or types as repeatables, meaning that numerous particulars that are instances of them may be found. I wrote the following to Barry.
I’m reading this paper, and will probably have some comments / questions, but a side thought struck me today. You and Werner summarise at the top the current general views of realists and anti-realists, including the standard description of nominalists. Your use of the term ‘repeatables’ made me think about the essence of the nominalist objection, i.e. there types are not real things, even if there are many individuals that can be grouped into extensions with very strong resemblance.
However, the very fact that there are (say) tigers and not just one tiger, and the indisputable fact of how they occur (sexual reproduction) points unavoidably to the ontic nature of their type – i.e. DNA. In general, where numerous instances of the same type occur, it is due to either biological reproduction via genotype and transcription to phenotype or to some industrial or other equivalent. Nominalists don’t disagree that phenotypes exist, but they would presumably have to deny the existence of genotypes, not to mention construction blueprints, computer chip architectures and so on, which are similarly generative of individual buildings, CPUs etc, to truly support their central claim. In other words, genotype, and anything artificial that functions as a blueprint are exactly the real-world incarnations of the type information whose existence they deny.
They might try to claim that no genotype is exactly the same as any other, even if it is 99.99% identical within a species, but this seems a weak argument, since it is very clear that a genotype’s evolutionary function is to represent type information. There is clearly a somewhat tedious argument that could be made showing how much DNA corresponds to species-level identical physiology, biochemistry, anatomical features etc, and also higher-than-species level features (e.g. chordate body plan) and also more individual allele level differences (hair colour, height etc) – in other words, there are ontic entities that correspond to ‘types’, even if they do have some more individual-level descriptive information attached to them.
I would have thought this argument would be a death-blow to nominalism, at least for biological entities – probably it is a very old argument, but I don’t follow the appropriate literature to keep track of it. However I am surprised to see it still being mentioned in ontology circles, including in papers like this one.
I have had similar thoughts, but have not seen thoughts like this written down. Two problems: They don’t fit non-biological types: planet, galaxy, sun, boson …They suggest a tricky regress: if it’s information entities like industrial designs that explain why industrial products form types, what is it that explains why information entities form types? Of course it could be information entities all the way down. In the beginning was The Word. And all that.
To which I replied:
BS: They don’t fit non-biological types: planet, galaxy, sun, boson …
TB: well these are interesting aren’t they – because they have nothing that stands for genotype; they are the result of something like particular constrained processes acting on certain types and amounts of crude matter and energy. Planets are something like small amounts of relatively heavy elements coalesced into bodies around a star; if far larger amounts of matter are involved you get a star; etc. I don’t know all the processes, but astrophysicists have standard explanations for species of galaxy, categories of star, the various kinds of black hole and so on.
Regardless of all that, I don’t see a genotype equivalent. So if anything, nominalist arguments might hold better here than for tigers, fish, tables, buildings and clinical guideline-driven workflows. There seems to be an interesting division between endogenously ‘typed’ things and exogenously ‘shaped’ or created things – possibly this is just the types v collections distinction.
I don’t have time to look into it properly, but there is clearly a qualitative difference between typed entities, such as biological organisms belonging to a species, 1938 Packards being instances of their factory blueprint (not to mention far more modern and more boring smartphones being instances of far more complex blueprints); and untyped entities, such a planets, stars, and bits of sub-atomic matter that have no blue-print, and no type from which they are instantiated. However, they are classifiable, which means that we can possibly claim they have associated classes. This would seem to imply there are at least two kinds of universal: (strong) instantiating universals, and (weak) classifying universals.
Unfortunately, there are only so many hours in the day. Well, at least I can provide a picture of a 1938 Packard. Question: is ‘1938 Packard’ real, or only real 1938 Packards?