Conceptual spaces as a framework for cognitive semantics

Peter Gärdenfors

Revue Internationale de Philosophie 44, pp. 24-46, 1990. Reprinted in Knowledge, belief and strategic interaction, ed. by C. Bicchieri and M. L. dalla Chiara, Cambridge University Press, Cambridge, 1992, pp. 377-396.

Abstract

There are two main approaches to modelling epistemic states. One is the foundations theory which holds that one needs to keep track of the justifications for one's beliefs: Propositions that have no justification should not be accepted as beliefs. The other is the coherence theory which holds that one need not consider the pedigree of one's beliefs. The focus is instead on the logical structure of the beliefs -- what matters is how a belief coheres with the other beliefs that are accepted in the present state. After presenting the two teories, I also introduce some of the empirical evidence relating to how people in fact do update their epistemic states and discuss its relevance for the two theories. The foundations and the coherence theories have very different implications for what should count as rational changes of belief systems. According to the foundations theory, belief revision should consist, first, in giving up all beliefs that do no longer have a satisfactory justification and, second, in adding new beliefs that have become justified. On the other hand, according to the coherence theory the objectives are, first, to maintain consistency in the revised epistemic state and, second, to make minimal changes of the old state that guarantee sufficient overall coherence.

In order to illustrate the conflicts between the theories, I present two modelings of belief revisions. The first is Doyle's Truth Maintenance System (TMS), which is a system for keeping track of justifications in belief revisions, and thus it directly follows the foundations theory. The second is the so called AGM model of belief revision that has been developed by myself in collaboration with Carlos Alchourron and David Makinson. This modelling operates essentially with minimal changes of epistemic states and is, in its simplest form, thus in accordance with the coherence theory. The advantages and drawbacks of each each modelling will be discussed.

One of the main criticisms directed against coherence models is that they cannot be used to express that some beliefs are justifications or reasons for other beliefs. This is true, of course, for single states of belief that only contain information about which beliefs are accepted and which are not. However, I argue that if one has further information about how such a state will potentially be revised under various forms of input, then it is possible to formulate criteria for considering one belief to be a reason for another. In particular, the notion of the epistemic entrenchment of beliefs, which plays a central role in a development of the AGM model, is useful here. Adding this kind of information about the beliefs in an epistemic state will thus produce a model of belief revision which satisfies most of the requirements of the foundations as well as the coherence theory.

The paper is available in PostScript format (get).

To LUCS Home Page