Content area
Full Text
On the origin of objects B CANTWELL SMITH
Cambridge, MA: MIT Press, 1996
Ontology, or the study of being, has waxed and waned in popularity in the recent history of philosophy. Currently it is undergoing a resurgence of interest due, in part, to its practical use in fields such as Geographic Information Systems (GIS), artificial intelligence (AI) and in computer science in general. It is through the study of computation that Brian Cantwell Smith begins the present epic foray into ontology and metaphysics. In fact, his computational background lends a much needed multidisciplinary breadth to the study of ontology as a whole.
Ontology is of particular interest to those working in GIS, AI, and computation because each of these involve, to varying degrees, the development of expert systems. Expert systems are pieces of software which incorporate processes typically associated with human intelligence. At its most ambitious, Al seeks to develop intelligent agents which simulate the full panoply of features possessed by human intelligence. At a minimum, expert systems must be able to discern differences and similarities among objects. Although at first blush such a task seems simple, it has proven to be one of the greatest impediments to the development of useful expert systems, and it is a core problem of ontology. For some time, researchers in computation have attempted to downplay the ontological difficulties posed by AI research. Smith describes this problem as follows: Within the analytic tradition ... it is traditional to analyze semantical or intentional systems, such as computers or people, under the following presupposition: (i) that one can parse or register the relevant theoretical situation in advance into a set of objects, properties, types, relations, equivalence classes, and the like (e.g., into people, heads, sentences, real-world referents, etc.) as if this were theoretically innocuous; and then (ii) with that ontological parse in hand, go on to proclaim this or that or the other thing as an empirically justified result. (p. 16)
Such assumptions very often work in given contexts, and the world may yet be modeled in ways useful for the brute-force purposes of many worka-day problems of computation. But Smith recognizes the meta-theoretical limitations posed by an inadequate ontological description of the world, especially given his goal of developing...