On the basis of Poincaré and Weyl’s view of predicativity as invariance, we develop an extensive framework for predicative, type-free first-order set theory in which $\Gamma _0$ and much bigger ordinals can be defined as von Neumann ordinals. This refutes the accepted view of $\Gamma _0$ as the “limit of predicativity”.

Hermann Weyl was one of the greatest mathematicians of the 20th century, with contributions to many branches of mathematics and physics. In 1918, he wrote a famous book, “Das Kontinuum”, on the foundations of mathematics. In that book, he described mathematical analysis as a ‘house built on sand’, and tried to ‘replace this shifting foundation with pillars of enduring strength’.

In this paper, we reexamine and explain the philosophical and mathematical ideas that underly Weyl’s system in “Das Kontinuum”, and show that they are still useful and relevant. We propose a precise formalization of that system, which is the first to be completely faithful to what is written in the book. Finally, we suggest that a certain set-theoretical modern system reflects better Weyl’s ideas than previous attempts (most notably by Feferman) of achieving this goal.

The operations of expansion and refinement on nondeterministic matrices (Nmatrices) are composed to form a new operation called rexpansion. Properties of this operation are investigated, together with their effects on the induced consequence relations. Using rexpansions, a semantic method for obtaining conservative extensions of (N)matrix-defined logics is introduced and applied to fragments of the classical two-valued matrix, as well as to other many-valued matrices and Nmatrices. The main application of this method is the construction and investigation of truth-preserving ¬-paraconsistent conservative extensions of Gödel fuzzy logic, in which ¬ has several desired properties. This is followed by some results regarding the relations between the constructed logics.

Many-valued logics in general and 3-valued logic in particular is an old subject which had its beginning in the work of Łukasiewicz [Łuk]. Recently there is a revived interest in this topic, both for its own sake (see, for example, [Ho]), and also because of its potential applications in several areas of computer science, such as proving correctness of programs [Jo], knowledge bases [CP] and artificial intelligence [Tu]. There are, however, a huge number of 3-valued systems which logicians have studied throughout the years. The motivation behind them and their properties are not always clear, and their proof theory is frequently not well developed. This state of affairs makes both the use of 3-valued logics and doing fruitful research on them rather difficult.

Our first goal in this work is, accordingly, to identify and characterize a class of 3-valued logics which might be called natural. For this we use the general framework for characterizing and investigating logics which we have developed in [Av1]. Not many 3-valued logics appear as natural within this framework, but it turns out that those that do include some of the best known ones. These include the 3-valued logics of Łukasiewicz, Kleene and Sobociński, the logic LPF used in the VDM project, the logic RM3 from the relevance family and the paraconsistent 3-valued logic of [dCA]. Our presentation provides justifications for the introduction of certain connectives in these logics which are often regarded as ad hoc. It also shows that they are all closely related to each other. It is shown, for example, that Łukasiewicz 3-valued logic and RM3 (the strongest logic in the family of relevance logics) are in a strong sense dual to each other, and that both are derivable by the same general construction from, respectively, Kleene 3-valued logic and the 3-valued paraconsistent logic.

In this work we describe a new approach to the notions of relevance and paraconsistency. Unlike the works of Anderson and Belnap or da Costa (see [2], [8] and [7]) we shall mainly be guided in it by semantical intuitions. In the first two sections we introduce and investigate the algebraic structures that reflect those intuitions. The corresponding formal systems are briefly described in the third section (a more detailed treatment of these systems, including full proofs, will be given in another paper).

Our basic intuitive idea is that of “domains of discourse” or “relevance domains”. Classical logic, so we think, is valid in as much as sentences get values inside one domain; limitations on its use can be imposed only with respect to inferences in which more than one domain is involved. There are two basic binary relations over the collection of domains. One is relevance. It is reflexive and symmetric (but not necessarily transitive). Under a given interpretation two sentences are relevant to each other when their values are in relevant domains. Another basic relation between domains, no less important, is that of grading according to “degrees of reality”. The idea behind it is not new. Gentzen, for example, divided in [9] the world of mathematics into three grades, representing three “levels of reality”. The elementary theory of numbers has the highest degree or level of reality; set theory has the smallest degree and mathematical analysis occupies the intermediate level. In the theory of types, or in the accumulative von Neumann universe for set theory, we can find indication of a richer hierarchy.

The system RM is the most well-understood (and to our opinion, also the most important) system among the logics developed by the Anderson and Belnap school. In this paper we investigate RM from a constructive point of view. For example, we give a new proof of the completeness of RM relative to the Sugihara matrix (first shown by Meyer), a proof in which a p.r. procedure is presented, applying which to a sentence A in RM language yields either a proof of it in RM or a refuting valuation for it in the Sugihara matrix SZ.

Two topics dealt with in this work deserve a special attention.

a) The admissibility of γ. This is a famous theorem of Meyer and Dunn. In [1] Anderson and Belnap emphasize that “the Meyer-Dunn argument … guarantees the existence of a proof of B, but there is no guarantee that the proof of B is related in any sort of plausible way to the proofs of A and Ā ∨ B.” In §2 we provide such a guarantee for the RM-case. In fact, we give there a direct method of obtaining a proof of B from given proofs of A and Ā ∨ B.

b) The relationships betweenRMand its full negation-implication fragment. RM is known ([1, pp. 148–149], and [3]) to be a conservative extension of (Sobociński 3-valued logic; see [4]). Anderson and Belnap admit [1, p. 149] that this fact came to them as a distinct surprise, since RM as a whole is far from being three-valued. In this paper, however, this “surprising” fact appears quite natural (see III.3). In fact, we show that , is the “hard core” of RM, since our proof of the completeness of RM is based in an essential way on the completeness of relative to the Sobociński matrix, and since the Gentzen-type calculus we develop for RM is a direct extension of a similar (but much simpler) calculus for . Because of the importance has in this work, we devote the first section to a constructive investigation of it.

We note, finally, that the Gentzen-type calculus mentioned above admits cut-elimination and normal-form techniques. (Such calculi were found till now only for RM without distribution.)

We deal here with two modal logics, GL and Grz, that are known to have interesting arithmetical interpretations connected with the notion of provability. GL is the extensiom of K (or K4) by the schema □(□ A → A) → □ A, and Grz is the extension of S4 by □(□(A → □A) →A) → □A. GL is also known to be sound and complete with respect to the class of all Kripke models that are transitive, irreflexive and well founded. Grz bears the same relation to the corresponding reflexive models. We refer the reader to [1] for a full exposition of the subject. (See also [4], [2], [6].)

In §I we develop a sequential calculus for both GL and Grz and give a semantical proof that both systems admit cut-elimination. (Incidentally, this provides an easy proof of the semantical completeness of the two systems.) With respect to GL this yields a correction of an error in [2].

In §II we show that cut-elimination fails for QGL (the extension of GL to a language with quantifiers). We further show that, despite this failure, QGL still has some of GL's interesting properties (e.g., the disjunction property). We also show, using fixed-point techniques, that similar properties obtain if we take as semantics for QGL the arithmetical interpretation extended in the obvious way.

We want to thank Professor H. Gaifman for his help while working on the subject.

This work results from an attempt to give the vague notion of relevance a concrete semantical interpretation. The idea is that propositions may be divided into different “domains of relevance”. Each “domain” has its own “T” and “F” values, and propositions “belonging” to one domain can never entail propositions “belonging” to another, unconnected one.

The semantics we have developed were found to correspond to an already known system, which we call here RMI⥲. Its axioms are the implication-negation axioms of the system RM ([1, Chapter 5]). However, as Meyer has shown, RM is not a conservative extension of RMI⥲, since RMI⥲ has the sharing-of-variables property ([5], and [1, pp. 148–149[), which the implication-negation fragment of RM has not.

RMI⥲ has four advantages in comparison to its more famous sister R⥲ (the pure intentional fragment of the system R; see [1]):

a) It has a very natural (from a relevance point of view) many-valued semantics, the simple form of which we describe here.

b) RMI⥲, ⊢ A1 → [A2 → (… → (An → A) …)] iff there is a proof of A from the set {A1, …, An} that actually uses all the members of this set. In R⥲, this holds only if we talk about “sequences” instead of “sets”. This is somewhat less intuitive (see [1, pp. 394–395]).

c) RMI⥲ is a maximal “natural” relevance logic, in the sense that every proper extension of it limits the number of “domains of relevance” (§III).