Ontological Categories and How to Use Them

Amie L. Thomasson

[1] We seem to talk about and think about many different sorts of things. Numbers and baseball games, universities and electrons, laws and legends, money and battlegrounds all form part of our everyday experience. Confronted with such a bewildering array of things we seem to think about and talk about, what is an ontologist to do? One important and standard approach, following Ockham, is to see how much one can get away without and, in the interests of parsimony, accept only those sorts of entities which we need. But the need criterion alone is insufficient. What is at issue in developing an ontology is what kinds of entities one admits: Whether one postulates universals, numbers, mental states, and so on. But by "kinds" we can't mean just any group of entities falling under a simple everyday classification like cabbages, cookware, youth groups and yard sales. There are far too many such sorts of entities to address on a piecemeal basis, and so if we proceeded in that way we could not hope to offer a comprehensive and systematic appraisal of what there is.

[2] Moreover, taking a piecemeal approach to ontology is not only excessively time-consuming, but dangerous. Simply denying entities wherever possible can lead to inconsistency if we retain one kind of entity but eliminate the entities on which they're founded, or to arbitrariness and false parsimony if the kinds accepted and rejected are not relevantly different. We would certainly gain no genuine parsimony, for example, by rejecting baseball games but accepting board games into our ontology, since they share the same relevant characteristics (being events that occur over time, governed by certain publicly agreed-upon rules, engaged in by conscious agents as "players", etc.).

[3] To avoid these problems we must supplement the need criterion with a system of ontologically relevant categories in which one might claim that things exist, without prejudging the issue of whether there are things in these categories. A system of categories provides a scheme on the basis of which one can draw out different ontological pictures by determining which of these categories are non-empty, enabling one to make principled rather than piecemeal ontological decisions. Moreover, if our categories reflect relations among things in different categories, we can avoid arbitrariness, inconsistency and false parsimony in our particular decisions about what entities to accept and reject. It is the purpose of this paper to lay the groundwork for a categorial approach to ontology and to sketch some of its advantages over the piecemeal approach.


1. Ontological Categories

[4] A system of ontological categories should be natural, so that one can locate important categories and preserve central distinctions; relevant, incorporating appropriate criteria for admitting or rejecting things; and exhaustive, to insure that we haven't inadvertently left something out and that we do not pose false dichotomies. We can draw out a relevant, natural and exhaustive system of categories according to the ways in which an entity does or does not depend on intentional states and spatio-temporal entities. To do so we need clear definitions of dependence and two properties (1) being real, where x is real just in case x has a definite spatio-temporal location, and (2) being an intentional state, where x is an intentional state just in case x has an intrinsic capacity to represent something beyond itself.1 Categories are distinguished by means of the ways (if any) in which an entity depends on real things and/or intentional states.

[5] Some important kinds of dependence may be defined informally as follows:

1. Dependence: Necessarily, if a exists, then a exists.

2. Historical Dependence: Necessarily, for any time t at which a exists, b exists or existed then or at some earlier time.

3. Constant Dependence: Necessarily, for any time t at which a exists, b exists at t.


Each of these types of dependence comes in two variants:
Rigid Dependence is dependence on a particular individual.

Generic Dependence is dependence on there being something or other of a certain kind.

[6] The close relations among these definitions of dependence are summarized in a few theorems that have important consequences for what categories are possible and what ontological systems are consistent. They may be summarized as follows:

1. Constant dependence entails historical dependence

2. Historical dependence entails dependence
Provided the reasonable assumption that anything which is an intentional state is necessarily an intentional state, and that anything real is necessarily real, two more theorems may be added:
3. If a is rigidly dependent or historically dependent or constantly dependent on b, and b is real, then a is generically dependent or historically dependent or constantly dependent on there being something real.

4. If a is rigidly dependent or historically dependent or constantly dependent on b, and b is an intentional state, then a is generically dependent or historically dependent or constantly dependent on there being something which is an intentional state.
Armed with these definitions and theorems, we are now in the position to outline an exhaustive set of categories detailing the ways in which an entity depends on real things on the one hand, and on intentional states on the other.