Ontological Commitment, Informally

Nobody has to announce what they believe exists. Commitments arrive uninvited, carried in by ordinary claims that were made for other reasons. A biologist who says there are three subspecies has said something about the world's contents without intending an ontological statement. Working out what a body of talk has committed its speakers to, whether they noticed or not, is the central technique of the subject.

The list

An inventory is a feature of a theory, not of a person. It is settled by what the theory says, and someone can hold a theory whose inventory they would sincerely deny if it were put to them plainly. That gap is what makes the technique useful. A dispute conducted at the level of stated beliefs can go nowhere while the parties are already committed, by the rest of what they hold, to positions they have not examined.

What commits you

The test is a conditional, and it is worth stating carefully because the two obvious misreadings are both tempting.

A theory is not committed to everything it mentions. Mentioning is cheap: "unicorns do not exist" mentions unicorns and commits to none. Nor is a theory committed only to what it explicitly says exists. The commitment can be entirely implicit, and usually is. "Some subspecies are endangered" never uses the word "exists" and commits to subspecies just the same.

What settles the matter is the truth condition. Ask what the world would have to contain for the claim to be true. If every way of the claim being true involves there being things of a certain kind, the claim commits you to them.

Quine gave the classic statement of the test, in "On What There Is": a theory is committed to those and only those entities to which the bound variables of the theory must be capable of referring in order that the affirmations made in the theory be true. Making that precise requires the machinery of quantified logic, and the intermediate tier does it properly. The informal version above is the same idea with the apparatus left out.

The escape route

Faced with an unwanted commitment, one option is to accept it. The other is to deny that the sentence carries it, by supplying a different account of what the sentence says.

The clean cases are convincing. "The average British family has 1.8 children" appears to be about a family. Rewritten as a claim about the ratio of children to families, it is about children and families and nothing else, and the fractional child never had to exist. Nobody defends average families, and the paraphrase is why.

"There is a hole in the bucket" appears to commit to holes. Rewritten as "the bucket is perforated", it commits to a bucket and to a way the bucket is. That paraphrase is more contested than the first, because holes support a great deal of further talk. They can be counted, compared for size, and located relative to one another, and each such sentence needs its own rewriting. Whether every one of them can be handled is exactly the question, and it is settled case by case rather than in advance.

The second premise is where all the pressure falls, and it contains two demands that pull against each other. The paraphrase has to require less, or it achieves nothing. It also has to say the same thing, or it has changed the subject and refuted a claim nobody made.

Using it

The technique gives ontological disputes their standard shape, and once you have seen the shape it is visible everywhere.

Someone points out that a body of accepted talk quantifies over entities of some kind. Their opponent offers a paraphrase that avoids the entities. The first replies that the paraphrase fails somewhere, usually on a sentence the paraphrase was not designed for. The second refines it. The dispute proceeds as an argument about the adequacy of a rewriting, and the entities are decided by the outcome.

Two consequences of this are worth carrying into the rest of the subject.

The first is that ontology is not conducted by consulting intuitions about what is real. It is conducted by examining what our best accounts of things say, and by testing proposed rewritings of them. A view is defeated when its paraphrases can be shown to fail, and that is a demonstrable matter.

The second is that the inventory depends on which theory you start from. This is not a defect. Commitment was defined as a relation between a theory and what it needs, so a different theory can perfectly well need different things. What it does mean is that a dispute about what exists is often, underneath, a dispute about which account of some domain to accept, and settling it may require doing the science or the semantics rather than the metaphysics.

The end of the tier

Six articles have built a working vocabulary. There is the existence question and the difference between existing and merely being spoken of. There are the standard candidates and the reasons each is contested. There is the repeatability question that divides theories of properties. There are the demands of identity and counting that any proposal has to meet. There is the level shift that shows why existence is not a feature of individuals, and now the notion of commitment that turns all of it into a method.

The intermediate tier begins by making Quine's criterion exact, and by working through the objection that the whole enterprise is a choice of language rather than a discovery about the world.