Abstract. Basin, Clavel, and Meseguer showed in [1] that membership equational logic is a good metalogical framework because of its initial models and support of reflective reasoning. A development and an application of those ideas was presented later in [4]. Here we further extend the metalogical reasoning principles proposed there to consider classes of parameterized theories and apply this reflective methodology to the proof of different parameterized versions of the deduction theorem for minimal logic of implication.
MotivationA reflective logic is a logic in which important aspects of its metalogic can be represented at the object level in a consistent way, so that the object-level representations correctly simulate the relevant metalogical aspects. As a consequence, in a reflective logic, metatheorems involving families of theories can be represented and logically proved as theorems about its universal theory. Basin, Clavel, and Meseguer showed in [1] that logical frameworks can be good metalogical frameworks when their theories always have initial models and they support reflective and parameterized reasoning; they also showed that membership equational logic is a particular logical framework that satisfies these requirements. In this paper, we extend their ideas and apply them to the (parameterized) deduction theorem.Basin and Matthews have shown in [2] how metatheories based on inductive definitions can be used to formalize metatheorems that are parameterized with their scope of application. As a case study, they formalize different parameterized versions of the deduction theorem in the theory FS 0 [8]; we will use the same case study to motivate the developments of the following sections.We can use membership equational logic (described in more detail in Section 2) to represent theoremhood in a logic as a sort in a theory. Conditional membership axioms then directly support the representation of rules as schemas, which is typically used in presenting logics and formal systems. Similarly, we can