On the revision of probabilistic belief states (Q1903584)

From MaRDI portal
scientific article
Language Label Description Also known as
English
On the revision of probabilistic belief states
scientific article

    Statements

    On the revision of probabilistic belief states (English)
    0 references
    0 references
    27 August 1996
    0 references
    The author first investigates revision of probabilistic states of belief that are represented by nonstandard conditional probability functions (Popper functions). Although plausible accounts of belief revision are available in this framework, iterated belief revision does not seem possible to achieve. Let \(P(/)\) be a Popper function. Then the revised factual probability function \(P^*_A\) is given by \(P^*_A(B)= P(B/A)\). However, no revised conditional probability function corresponding to \(P^*_A\) is obtainable, and therefore it is not possible to revise \(P^*_A\). The remedy proposed for this is to introduce probabilistic ordinal conditional functions, POCFs (an extension of Spohn's ordinal conditional functions, OCFs). \(\langle \kappa, P\rangle\) is a POCF if and only if (1) \(\kappa\) is a function assigning a non-negative integer to each world, and assigning \(0\) to at least one world, and (2) \(P\) is a function that assigns to each world a real number in the interval \((0, 1]\). (Worlds assigned the value \(0\) by \(\kappa\) are those considered possible by the agent.) Each POCF determines a Popper function as follows: \[ P(B/A)= {\sum \{P(w)\mid w\in \min(\kappa, A)\text{ and } w\models B\}\over \sum\{P(w)\mid w\in \min(\kappa, A)\}}. \] Many different POCFs may induce the same Popper function. Instead of conditionalizing by a sentence \(A\), conditionalization refers to a pair \(\langle A, k\rangle\), where \(k\) is an integer representing the strength with which \(A\) is given. The outcome \(\kappa_{A, k}\) of \(A, k\)-conditionalizing \(\kappa\) is determined in the way proposed by Spohn for OCFs. This model has the advantage of allowing for iterated revision. A series of results are proved that confirm its intuitive plausibility. Its major disadvantage, acknowledged by the author, is that it requires more information than most other belief revision models. Related work by \textit{Abhaya Nayak} [Erkenntnis 41, 353-390 (1994)] should have been mentioned and discussed.
    0 references
    belief change
    0 references
    conditionalization
    0 references
    Popper functions
    0 references
    revision of probabilistic states of belief
    0 references
    nonstandard conditional probability functions
    0 references
    belief revision
    0 references
    probabilistic ordinal conditional functions
    0 references
    iterated revision
    0 references

    Identifiers

    0 references
    0 references
    0 references
    0 references
    0 references