Tuesday, December 20, 2011

OOO is Very Abstract, but so is KR

Over the past several months I’ve been reading around in object-oriented ontology (OOO)—I’m currently reading an interview with Levi Bryant—and I note that it’s a very abstract way of dealing with the world. Here, for example, is a passage from that Bryant interview:
Is use the term “withdrawal” in a somewhat different sense than Harman. For Harman, withdrawal means that objects are independent of all their relations such that they never touch or relate to one another. For me, by contrast, objects are capable of relating, but are also external to the relations in the sense that they can break with current relations and enter into new relations. With Harman I thus hold that objects are independent in the sense that they are not constituted by their relations, while contrary to Harman I hold that objects can enter into relations with other objects. For me, withdrawal thus means two things. On the one hand, withdrawal refers to the virtual dimension of objects. The virtual dimension of objects or their powers is forever withdrawn from other objects. Not only do objects have all sorts of powers that may or may not ever lead to manifestations or actualizations (a person might never get a tan because they live their entire life locked in a dungeon), but also powers as such are never themselves manifested. That is, the qualities an object manifests never resemble the powers that it possesses.
It’s all about JUST objects and relations, and powers, and qualities too. Very abstract.

There’s nothing surprising about that. That’s how philosophy tends to be. And I knew that going in.

What strikes me, however, is that this level of abstraction feels akin to knowledge representation (KR), the discipline in cognitive science and artificial intelligence about representing human knowledge in computational form. KR has many specific formalisms, but one can think of them as being about objects and relations, powers and qualities. If you’re building an expert system for medical diagnosis, well, what objects, relations, powers, and qualities do you need to have in your system in order to represent some body of medical diagnostics? If you want to be able to recognize stories about going into food establishments and ordering a meal, what objects, relations, powers, and qualities do you need to have in your system in order to do that? So, the study of KR is the study of how to deploy objects, relations, powers, and qualities in representing bodies of knowledge.

There are differences, obviously. KR will, at some point, involve formalized expressions of some kind—often based in set theory and predicate calculus—and may also involve diagrams. OOO seems not to involve those things at all, though there are the beginnings of such in Graham Harman’s The Quadruple Object, with its Greimasian diagrams and ontographical notation.

I mean, I’m almost tempted to develop a graphical language to represent some of Harman’s notions. For example (pp. 115-116):
First, the location of sensual objects cannot be inside the mind, since both the mind and its sensual objects are located on the interior of a more encompassing object. If I perceive a tree, this sensual object and I do not meet up inside my mind, and for a simple reason: my mind and its object are two equal partners in the intention, and the unifying term must contain both. The mind cannot serve as both part and whole simultaneously. Instead, both the mind and its object are encompassed by something larger: namely, both exist inside the object formed through the relation between me and the real tree, which may be rather different from the trees found in everyday life.
The exercise would be to come up with a graphical notion that depicts four things, the real tree, the sensual tree, the mind, and the ‘something larger,’ AND their appropriate relations. That’s not going to be a very complicated diagram. The trick, of course, is to come up with diagrammatic conventions that will perspicuously handle a lot of situations.

And THAT’s what KR is about, coming up with such schemes.

Another thing, OOO is about the world in the deepest sense. It’s metaphysics. KR is, well, just what it is IS a bit tricky. For some investigators it’s about how the mind packages knowledge, and that is very different from being about the world as it makes no claim about ultimate reality. But lots of KR is done by folks who just want to set up a working computer system. There’s no explicit claim about how the world deeply is, nor even about how the mind represents it to be; there’s just the claim that this computer system will handle some circumscribed class of knowledge-based tasks, like medical diagnosis, ordering a meal, or, more recently, competing at Jeopardy.

Metaphysics and Jeopardy are very different games. But underneath, or is it above? each is a kind of chess. You’ve got a limited number of pieces, each with highly circumscribed moves, and you’ve got to cover the waterfront only with them.

1 comment: