Comparing two Models on Preference and Preference Change Fenrong Liu February 28, 2007 1 Some general points about preference There are different notions of preference, expressed in various ways in natural language. It occurs in many research areas as well. Typically, we use preference to draw comparison between two things explicitly. Depending on the real situations, things under comparison can be possible states of affairs, objects, actions, means, and so on, as listed in [Wri63]. Logical modelling is an abstraction, we can only focus some aspects. It is dangerous to always try to match the formal preference with our intuitions of preference. [Hal57] described several aspects of preference usually ignored. In the following we are going to compare the recent two approaches of modelling preference and preference change, proposed in [BL07] (R-preference for short) and in [JL06] (C-preference for short), respectively. We start with reviewing the basics. Then we will discuss several issues, to show how these two ways of modelling can benefit from each other. 2 Review on R-Preference: the basics 2.1 Basic static system Definition 2.1 (model) An epistemic preference model is a tuple M=(S, { i | i I }, { i | i I }, V), S, i and V are standard, i is a reflexive and transitive relation over the worlds. We read s i t as ‘t is at least as good as s’ for agent i. Intuitively, worlds stand for situations that are compared. Definition 2.2 (language) Take a set of propositional variables P and a set of agents I , with p ranging over P and i over I . The epistemic preference language is given by: ϕ ::= ⊥| p ϕ | ϕ ψ |〈K i ϕ |〈pref i ϕ | Eϕ. pref i ϕ says that some worlds which the agent considers as least as good as the current one satisfy ϕ. E is an auxiliary existential modality. U is the dual of E. Definition 2.3 (truth conditions) We define M,s | = ϕ (formula ϕ is true in M at s) by induction on ϕ, here are two new clauses: 1. M,s | = pref i ϕ iff for some t : s i t and M,t | = ϕ 2. M,s | = iff for some t: M,t | = ϕ. 1