Theories propounded under the heading of ontology have often been full of whimsical, metaphysical inventions and the source of much confusion. Ontology has been defined as the study of what is the case as contrasted with epistemology: the study of what one can know to be the case. It has also been contrasted with axiology or normative philosophy: the study of what is the case versus of what ought to be the case. 'Ontological' problems of that sort have thus been posed on the same level, and side by side, with those of epistemology and axiology or normative philosophy. Considering themselves 'specialists in the nature of being' the 'ontologists' concerned purported to be searching for what there (really) is, as opposed to what can be known to be the case, and to what should be the case. Antimetaphysicians who have always repudiated the belief in some 'nature of being' or an 'ultimate reality' have indicated that this kind of 'ontology' is an exercise in futility. With it, however, they have traditionally also repudiated all theories about kinds of existence or about the ontological status of existents. It seems like they have wrongfully equated the idea of an ultimate reality with the idea of an ultimate conceptual framework or constructional system used when speaking or thinking about reality. But it is exactly this which is the primary task of a sensible ontology: to make explicit the fundamental, conceptual or constructional categories and presuppositions of a particular system of language or thought, and to examine what are the primitives and hypothetical entities in this system. Ontology in this sense is, then, concerned in the first place with the question whether such a system does indeed make a distinction between:

The subject of ontology is therefore not what is the case as distinct from what is, can, or should be known or believed to be the case, or as distinct from what ought to be the case, but the subject of ontology is first of all the question whether the distinction between the factual and the epistemic or doxastic, and between the factual and the normative, is actually drawn in a particular system.

Some of the above distinctions correspond to a difference in ontological status, some do not, but also in 'one and the same' sphere of what was, is and will be the case some entities may have another ontological status than other entities. Immediately following is therefore the question of what is explicitly or implicitly taken to be existing in a certain language or system of thought:

The crux of the matter is, of course: what does existence mean? If we do not make use of a synonym like being , it is not possible to define this term (or these terms) without referring to entities whose (purported) existence is required for the definition itself, if only by means of giving examples of existence. This implies that the term existence may have different meanings in different systems, dependent on what kind of entities are said to 'exist' in these systems. (Compare the meaning-variance thesis in logics: the meaning of logical constants wholly depends upon the axioms or rules of the system in which they occur.) Thus solipsists who believe that they only exist themselves, or idealists who believe that only mind exists (that is, who use the word exist so that only mind 'exists') give another meaning to existence than materialists who believe that only matter exists. The same does not apply to the existence or nonexistence of the supreme being and entities with supernatural qualities like gods and demons, because the suggestion is that they have the same ontological status as human beings or people and the other things of the 'natural' world.

After having made these inventories the next step is to study the possible inconsistences in the constructional or ontological system, and the possible elimination of certain categories, postulates, primitives or hypothetical entities. Underlying this activity is the conviction that every theory or conceptual system must be free from superfluous conceptual ballast. The criterion of consistence is an essential element of the coherentist theory of truth, the criterion of simplicity or parsimony is a principle of conceptual minimization, that is, of fewest conceptual entities (categories, postulates, primitives, and so on). A problem with regard to the latter criterion is that it is not that simple to determine the general, formal simplicity of an ontological system (if possible at all), even not in only one respect. For example, the simplicity of the basis of primitive predicate expressions is not fixed by merely counting the number of primitives, or of primitive predicate places (because --as has been argued-- predicate expressions can be compounded into other predicate expressions having more places, or can be replaced by other predicate expressions having fewer places). Another question is the number of nonprimitive expressions. We shall see that our own ontological system has many more individual expressions, because it accepts the names of attributes, such as happiness, as individual expressions where other, formal systems have only -- is happy as a predicate expression. On the other hand, the latter systems need many more predicate expressions, and the total number of individual and attributive expressions remains the same in both types of system. If individual expressions should not be multiplied 'beyond necessity', then neither should attributive expressions. And altho -- is happy and the other expressions designate only one-place predicates or attributes, the primitive two-place predicate expression -- has .. needed in addition to the individual expressions is always the same one, and cannot be dismissed anyhow in any system which recognizes the relation of having-as-an-element or being-an-element-of (which is its inverse).

As to relations it is inevitable that we must use individual expressions besides the relational ones. These individual expressions are reduced or 'derelativized' one-place predicate expressions of the two- or more-place corresponding ones. (In combinatory logic such 'derelativization' is done by means of a predicate operator 'Der'.) Friendship in the sense of having (someone as) a friend, for instance, corresponds to the one-place -- has (someone as) a friend which is a reduction of the two-place -- has .. as a friend (the inverse of -- is a friend of ..). This recognition of individual relations is necessary, because when relations become the focus of attention, and when we start talking about their attributes and/or relations (with other attributes and/or relations), they become things themselves, albeit in a different domain of discourse. Relations (such as friendship) cannot have the same ontological status as the things (such as friends) they relate to each other, otherwise we would be stuck with a loose, unconnected set of objects with parts and purely nonrelational attributes at the most. Thus, when we talk about relations, we make use of individual expressions corresponding to reduced, one-place predicates, and when we talk about the things being related, we make use of two- or more-place predicate expressions. Attributes are limit cases of relations: one-place predicates as it were. They have the same ontological status and behave very much like them, especially when looked upon as things in a separate (the so-called 'secondary') domain of discourse. This is the reason that we shall employ the term predicate as common denominator of both attributes and relations, while using the phrase predicat(iv)e expression for an expression which designates an attribute or relation. (In other systems such an expression itself is often termed "a predicate", while attributes may be called "internal" or "intrinsic properties" and relations, "relational" or "extrinsic properties".)