Abstract

This paper discusses the probabilities of inductive generalizations in languages containing two-place predicates. The depth of the sentences considered here is restricted to two, that is, they contain at most two layers of quantifiers. The analysis of relational hypotheses presented below is based on the theory of distributive normal forms in first-order logic. The main purpose of this paper is not to present methods of calculating unique probability-values for various generalizations, but rather to clarify the general conceptual situation and concentrate on issues of philosophical interest.