The ontology of biology 1 - What an ontology really is

It has become common in recent years for people to use terms of philosophy in distinct contexts, as it has terms of biology. Thus, ontology has gone the way of taxonomy, being dragooned into service of database techniques, to mean something quite the opposite of what it originally meant. I have noticed this tendency of computer technology for decades, ever since I got hopelessly muddled when doing database programming in the early 80s until I realised that they were using some terminology of formal logic in exactly the wrong way (I forget what it was now). A database ontology is not an ontology, just as a database taxonomy is not a taxonomy. The homonymity is confusing and misleading. So I want to talk about what an ontology really is, from a philosophical perspective. Think of this as a Basics post for a philosophical domain and idea.

An ontology is, fundamentally, about what there is. Of course, when we think of what the world is furnished with, we tend to think of physics: the standard model, and so on. However, ontology is not about the best model of physics, exactly (there is an ontology of physics which is about things like fields, fundamental particles and so on), but about what there is in any domain. In a famous paper, entitled "On What There Is", the modern debate on ontology was crystallised by Willard Van Orman Quine, the famous American philosopher. In that paper he discusses if possibilities that are not actual are real things. He wonders if there is a mental world. He wonders if there are universals. This is classical ontology, a subset of metaphysics. It's not about specific categories into which we should slot the world's objects and properties (are there properties? That, too, is an ontological question), but about the broad classes of the furniture of the world.

When database projects call what they have an ontology, what they actually have is a data structure that suits some task or group of data collectors, like scientists. Data is not the same thing as the world (although one might ask if one's ontology of the world should include data, which I personally dispute), and so the data structure is a matter of convenience, what is serviceable for those who are to use it. It might actually reflect an ontology, but an ontology is generally something that we take from our best theories, rather than build up piecemeal according to the assays we have and the things we can measure. In short, we want our measurements to be important, and the way they are made to be important is by ensuring they are theoretically significant (and that we don't make an ontological virtue of a methodological necessity and call what we can measure now the reality). That, too, is something Quine said, although he was not the first or last to do so.

There is an extension of ontology, though, into specific fields, as I indicated when I mentioned the ontology of physics. That is, one takes a particular theory, like quantum mechanics, and asks, what are the classes of things that this theory indicates exists in the world, and what ontologies do different interpretations of that theory give us? For example, the "many worlds" interpretation of quantum theory indicates that each time a waveform collapses into a measurable event in our world, the other possibilities fork off as actual worlds. So Schrödinger's cat is dead in one world and alive in another. That is an ontology of physics. It leads to questions one cannot solve or answer by doing more physics. The theory implies that there are classes of things that exist, like parallel worlds.

But there are sciences that have theories that are pretty local and contingent. For example, to take the notion of an organism in biology, or a business in sociology, one does not think the universe is necessarily populated with organisms or businesses (there might have been no life, or no trade), but instead one must think that these are entities only of that restricted domain and the theories that explain it. And even then we might conceive of a biology in which there are no organisms, or a commerce in which there are no businesses. So the ontologies of these contingent or "special" sciences is rather limited and constrained by just the active theories of those sciences. Ontologies of these theories might be thought to be too "small" to be of interest except to the scientists themselves and those hapless souls, like myself, who study them.

I aim in this series to argue that in fact the ontology of biology is a radical challenge to the default western way of thinking and causes us to be very uncomfortable where we literally and figuratively live. In particular, the problem of change and knowledge under change, which is I think the core problem of western philosophy since Heraclitus, is radically underlined by the theories of evolution and ecology. There is a tendency for us to try to force living things into defined kinds of a standard ranking, and to make inferences as if things in the living world were A or Not-A, when in fact they are not only distributions of traits, but changing distributions of traits, and changing distributions of changing traits - not only do legs vary in length and size, they change in their variation over time, and sometimes they change from being legs, to being wings, for example, or flippers, or even disappearing.

This will be an intermittent series, done when I can find time and brainpower to prepare them. Some of the general categories of ontological objects and processes I will be covering include reproduction, lineage, and replication, adaptive niche, guild and specialisation, module, development, and stage, taxon, type and classification and so on. Each of these has, I believe, close analogues in the social sciences - in fact as a believer in cultural evolution I hold that there are exact analogues to the general ontology of biology in sociology and history. We might not expect there to be an analogue for sex in culture, nor for a Mendelian ratio or even particulate heredity, but we must expect that there is an analogue for reproduction; likewise there may not be analogies to "biological" (that is reproductively isolated) species in culture, but there may be something like lineages and phylogenies.

So, we shall see how far we can get with this. Please comment as we go.


More like this

Should biodiversity then best be understood as biodiversity of traits or phenotypes (instead of species and organisms) in any given environment? In other words, should bumper stickers instead say "Save the Baleen and Blow Holes"?

I suspect the computer science use of the term derives from the understanding that it's impossible to get past the language/logic used to set the terms, when discussing what exists, i.e., that ontology is inherently relative to language.

In particular, the problem of change and knowledge under change, which is I think the core problem of western philosophy since Heraclitus, is radically underlined by the theories of evolution and ecology.

Heraclitus is one the coolest philosophers, IMO - no permanent ontology but change itself (fire). It will be interesting to see where you take this. Some deep territory here: the "ship of Theseus", problems of identity and ontology, law of the excluded middle, etc. Identity is also a problem in QM. With change, you may also be wandering unavoidably into philosophies of time. You could also take a pragmatic route and say that ontology is relative to context, perspective, or purpose, but then philosophy of mind traps may be waiting for you.

Naturally, any use of "ontology" in the context of computers will be rather more concrete than in philosophy, but I think some sense of the original meaning is preserved, at least somewhat.

It seems to be meaningful to talk about "an ontology" in respect to someone's formalization of ontology (e.g. "Aristotle's ontology is whack"), so it seems possible to speak of a formal language and model, such as in computer science, that has its own ontology, which may not resemble what we would consider ontology as such.

One place that the concept of an ontology might be reasonable in computer science is in the definition of languages or tools. A programming language that allows only simple data to be named or manipulated is a rather different thing than one which allows things such as procedures or entire computations to be similarly handled. This is true even though every program in the one language has behaviorally equivalent programs in the other, much like a positivist account of physics, with its very poor ontology, can state the same experiments and results as a interpretation of physics as ontologically rich as the many-worlds theory.

I do agree, however, that referring to the organization of things in your database as an ontology is an embarrassment.

What of the name of Gould's book:
Ontology and Phylogeny?

See OBO. These are ontologies of biology that attempt to represent biological reality, not database structures (the fact that these ontologies have one possible role in structuring databases does not alter this fact).

The one exception in OBO would be the Ontology of Biomedical Investigations (OBI), which represents investigations rather than the largely mind-independent world we are investigating. Databases are real and are rightfully represented within this domain.

No, these are not ontologies of biology, they are ontologies of discovery. As facts are discovered, an attempt is made to put them into a data structure, but this means that whatever is first discovered determines the structure. The contingency of discovery means it is not an ontology, but a working record, and to make the mistake of conflating it with an ontology is to make a very bad mistake indeed.