A Basis for AGM Revision in Bayesian Probability Revision

Abstract In standard Bayesian probability revision, the adoption of full beliefs (propositions with 1) is irreversible. Once an agent has belief in a proposition, no subsequent revision can remove that belief. This unrealistic feature, and it also makes incompatible change theory, which focuses on how set modified through both additions retractions. problem theory be solved model (i) lets codomain function hyperreal-valued rather than real-valued closed interval [0, 1], (ii) identifies as propositions whose either 1 or infinitesimally smaller 1. this model, changes will result (belief set), constitutes submodel conceived “tip iceberg” within larger contains lower levels probability. The patterns coincides corresponding pattern slightly version AGM commonly gold (dichotomous) change. modification only concerns marginal case by inconsistent input sentence. These results show dichotomous unified one same framework, – if we so wish subsumed under allows for iterated removal beliefs.