Traditionally, ontological studies have been methodologically based on introspection and analysis of world models and abstract theories. Logic has also been used to analyse the ontological commitments in theories (Quine, 1953). Smith and Mark (1999) have reported on an experiment in the use of empirical methods, based on common-sense realism, to test aspects of an ontological theory of geographic objects. Much of the work done in categorisation is based on mathematical set theory. All objects within a set are equally representative members of a set, and it is absolutely definable as to whether an individual item is or is not a member of a set. In looking at geographical objects, this clarity is not evident. Many geographical terms such as pond, lake, sea or ocean more closely resemble ranges on a continuum rather than precisely definable items. Clearly these geographical objects do not satisfy the requirements of set theory. Equally, comparison of published papers in the field of information systems research with some of the categories proposed in schemes covering information systems research reveal a similar vagueness in distinguishing categories. For example, papers on entity relationship modelling might be categorised as systems analysis and design or as database, or might be thought to lie somewhere between the two, depending on the emphasis of the paper. In these cases we are dealing with categories that are the products of human cognition rather than products of mathematical propositions.

Rosch (1978) looked at the ability of people to differentiate between objects and to identify individual objects which fitted in specific categories better than other individuals, and found a great degree of agreement as to which were good and bad examples. Rather than discrete set-based categories, she described categories with a radial structure, with prototypical or central members surrounded by more or less typical members. This has similarities with the approach of Franz Brentano to mereology. The common-sense world is complex and is divided in different ways and at different levels. Mereology concerns the basic organising relationships of part to whole, part to part within a single whole, of identity, overlapping and discreteness.

Most of the work done by Rosch and other cognitive scientists is based on studies of entities of tabletop space such as tools, small pets or of abstract items such as colours and diseases. Smith and Mark (1999) were interested in determining whether these approaches could be applied to geographic categories. Their experimental framework consisted of two complementary phases: traditional ontological work (largely deductive, introspective and formal) and research with human subjects (empirical, inductive). The ontological theories were used as the starting points for the design of experimental protocols to test the degree of fit of the ontological theories. This data can then be used to refine the ontological theories to form the basis for further iterations.