Summary of some key ideas in Brian Cantwell Smith,
On the Origin of Objects

A. The ontology of computation

1. is a program an abstract entity or a concrete one?  Could make a case either way.

2. need a three-way distinction between program (abstract), process (concrete but still a representer rather than something represented), and domain (the realm of things that the program is about, in the sense e.g. that states of an income tax program are about your income, your medical expenses, etc.)

B.  The ontology of the domain

(Smith argues that writing software that models a domain inevitably involves us in ontological issues about the best way to carve up the domain.  One central point he makes here is that our ontology needs to be flexible rather than absolute, in the sense that it will depend on our particular purposes and goals.)

one interesting issue in this ballpark that we have discussed (and Peter has done a presentation on) involves the ontological implications of object-oriented programming.

C.  The interaction between A and B

The way that the ontology of computation affects the ontology of the domain and vice versa.  Smith has some detailed and interesting examples here.  A general point he offers is that it is all too easy to become involved in "inscription errors," which occur when we essentially presuppose a particular ontology in constructing a theory (or a program), then take the success of that program to validate the ontology.

D. Smith's Own View ("Successor Metaphysics")

Smith goes on to describe his own metaphysical view and locate it in relation to others.  We might think of his task here as an attempt at a Hegelian synthesis of superficially competing views:

1. "modern metaphysics." As I understand it, this view combines foundationalism (the idea that there is a real world that we didn't create that underlies the categories we use in thinking about the world) with reductionism (the idea that higher-level things can be defined in terms of lower-level things; closely related to the positivist idea that all the sciences would ultimately reduce to physics:  sociology would reduce to psychology, psychology to biology, biology to chemistry, and chemistry to physics).

2. "postmodern metaphysics." This combines a rejection of foundationalism with a rejection of reductionism.

3. "successor metaphysics." Smith's own view. This seems to be a synthesis of the first to in the following sense: it combines the foundationalism of modern metaphysics, contrary to postmodernism, with the antireductionism of postmodernism, contrary to modern metaphysics. This seems to explain what Smith means in the preceding chapter when he insists that the world should be grounded, but not grounded *in* anything in particular.

E. A Form of Kantianism?

In chapters 11 and 12, Smith clarifies his own metaphysics and its relation to ontology. Here's my own quick & dirty take on what he's got in mind.  Metaphysics deals with the nature of the world "in itself" (Kant's terminology, not Smith's), about which about the best we can say is that it is unitary, not divided into things or kinds of things ("analog" rather than "digital"). Ontology, by contrast, deals with the world as digital rather than as analog, with the world as classified into kinds rather than as an indivisible whole. Ontology and representation are two sides of the same coin. In "registering" the world, we organize it into things and kinds of things. Looked at from the side of the world, we can regard these classifications as giving an ontology, an account of the basic types into which the world is organized. (But we need to keep in mind that to some extent this is pragmatic, and the classification might differ with differing goals or purposes.) Looked at from the side of the agent doing the registering, on the other hand, we have an account of the intentional content of the agent's internal states: an account of what in the world those states get attached to.

To my eye, the resulting view looks a lot like Kantianism. Metaphysics deals with the world prior to any classification or organization we might engage in:  with Kant's "things in themselves." Ontology, on the other hand, has to do with the way we organize or classify the world, and thus is relative to some extent to human categories, interests, and purposes. In Kant's scheme, ontology has to do with empirical objects, with the phenomenal world rather than the noumenal one. (Except that Kant thought that all humans would organize the world in the same way due to innate categories, while Smith seems to allow that such organizations may differ from one community to the next.)



Last update: September 15, 2004. 
Curtis Brown | Philosophy Department | Trinity University
cbrown@trinity.edu