No label defined (Q4236370)

From MaRDI portal

!

This is the item page for this Wikibase entity, intended for internal use and editing purposes. Please use the normal view instead:

scientific article; zbMATH DE number 1269571
Language Label Description Also known as
default for all languages
No label defined
    English
    No label defined
    scientific article; zbMATH DE number 1269571

      Statements

      0 references
      0 references
      0 references
      14 September 1999
      0 references
      belief change
      0 references
      belief revision
      0 references
      representation of epistemic states
      0 references
      focusing
      0 references
      conditioning
      0 references
      cognitive state
      0 references
      The main thrust of this chapter is to suggest, as Gärdenfors (1988) already did in more restricted frameworks, that beyond the diversity of formal models of cognitive states, and their corresponding belief change rules, there is some agreement upon what belief change means. The complementarity of semantic and syntactic representations of cognitive states has been emphasized: the use of numbers and orderings in the representation greatly facilitates the calculation of belief revision, be it by making the revision rule unique in a given setting. However semantic representations are often poorer than syntactic ones because pieces of information contained in the cognitive state are lumped into a single uncertainty measure. The increased expressivity introduced by ordinal or quantitative representations of cognitive states can go along with the refinements in the belief change modes allowed by syntactic representations in a logic, provided that one moves from classical logic representations to non-classical ones. This paper considers different numerical settings for the representation of epistemic states and surveys the existing rules for taking into account new information either in a belief revision perspective in the spirit of Alchourrón, Gärdenfors and Makinson postulates, or in an updating perspective in the sense of Katsuno and Mendelzon's postulates, or in a conditioning perspective referred to as `focusing', in the spirit of probabilistic reasoning. The possibilistic framework for conditioning leads to a more complete agreement with the AGM postulates (first stated for propositional logic) than the Bayesian probabilistic setting. We suggest that probabilistic change rules can also be described in terms of the standard axiomatic approaches to belief change of the logical framework, if the probabilistic approach is cast in the setting of convex sets of probabilities. In that respect, the crucial distinction between focusing and revision laid bare in generalized probabilistic settings, is pointed out. This distinction refers to the difference between generic knowledge concerning a (more ore less well-defined) population, and pieces of evidence, or observations on a case under study. A cognitive state can consist of generic knowledge plus factual observations that pertain to a case. Then, deriving plausible conclusions is done by focusing on the reference class pointed out by the observations, thus revising previous plausible specific conclusions, not the generic knowledge. Revision only takes place when the cognitive state and the inputs are of the same nature: observations (e.g. data fusion), or generic knowledge (e.g. adding rules to an expert system). This distinction looks crucial for discussing belief change in conditional knowledge bases, where conditionals are generic while observations are propositional. This distinction is purely conceptual in the Bayesian setting because focusing and revision are usually expressed by Bayes rule. However, the unique probability assumption in the Bayesian setting is often the result of default assumptions (conditional independence, maximal entropy). It is not clear that revising a probability distribution \(P\) induced by a family \({\mathcal P}\) thereof, is the same as revising the probability set \({\mathcal P}\) and constructing a single probability from this revised set. Generally, it will not be so, and the latter procedure looks more reasonable. Similarly revising the rational closure of a conditional belief base may differ from computing the rational closure of the previously revised conditional belief base. Lastly, we have not considered the problem of syntax-based methods for changing probabilistic belief bases. This question seems not to have been addressed, but in the setting of Bayes networks, for the local computation of Bayes rule. This is one more topic for further investigations. For the entire collection see [Zbl 0908.90003].
      0 references

      Identifiers

      0 references
      0 references
      0 references
      0 references