What is ceteris paribus preference? (Q1918914)

The author wants to show that more plausible logical models for ceteris paribus preference can be obtained if absolutely complete alternatives, i.e. possible worlds, are replaced by contextually complete alternatives. A set \({\mathcal A} = \{A_1, \dots, A_n\}\) of sets of sentences is called a set of (contextually) complete alternatives if (1) \({\mathcal A}\) has at least two elements, (2) every \(A_i\) in \({\mathcal A}\) is consistent, and (3) for every \(A\in {\mathcal A}\), \(A'\in {\mathcal A}\) and \(p\in A\), either \(p\in A'\) or \(\neg p \in A'\). As a first approximation, the informal statement ``\(p\) is better than \(q\)'' is translated into \((p \& \neg q) P\) \((q \& \neg p)\), where \(P\) stands for ``Preferred''. However, the author argues that the translation should rather be \((p\;{/_{\mathcal A}} q)P(q\;{/_{\mathcal A}} p)\), where \(p\;{/_{\mathcal A}} q\) equals \(p \& \neg q\) if \(q\) does not follow from \(p\) in the context \({\mathcal A}\) and \(p\;{/_{\mathcal A}} q\) equals \(p\) otherwise. A pair \(\langle A,B \rangle\) of complete alternatives, such that \(p\;{/_{\mathcal A}} q\) is true in \(A\) and \(q\;{/_{\mathcal A}} p\) is true in \(B\), is called a representation of the pair \(\langle p\;{/_{\mathcal A}} q\), \(q\;{/_{\mathcal A}} p \rangle\). Not all representations of \(\langle p\;{/_{\mathcal A}} q,\;q\;{/_{\mathcal A}} p \rangle\) need to be relevant to the comparison between \(p\) and \(q\); those that are relevant are picked out by a representation function \(f\). Now, let \(R\) (at least as good as) be a basic preference relation in \({\mathcal A}\) and \(f\) a representation function for \({\mathcal A}\). Then the weak preference relation \(R_f\) is defined by \(pR_fq\) iff \(ARB\) for all representations \(\langle A,B \rangle\) of \(\langle p\;{/_{\mathcal A}} q,\;q\;{/_{\mathcal A}} p \rangle\), selected by \(f\). So, the author has shown how ceteris paribus preferences can be derived from basic preferences that refer to a set of contextually complete alternatives. In section 5 the author argues that further refinement may be attainable by the introduction of a similarity relation on which the representation function can be based. In section 6 it is studied for which properties it holds that if \(R\) has this property, then so does \(R_f\). And in section 7 the author discusses principles of contraposition, such as if \(pR_fq\), then \(\neg q R_f \neg p\), conjunctive and disjunctive preferences. By examples he illustrates that the positive and negative results of his theorems are well in tune with pre-analytical intuitions.