Relevant consequence and empirical inquiry (Q2366530)

The inference from \(A\) to \(A\vee X\), whilst classically valid, has the pragmatic disadvantage of permitting the introduction of ``irrelevant'' subject matter, and the formal peculiarity that we may substitute an arbitrary sentence \(X'\) for \(X\) in the consequent along without loss of validity. The authors' general question is how far suspect inferences of this kind are necessary for scientific investigation. They show that under a certain formal model of scientific inquiry, such inferences are not needed when making successive conjectural approximations to a true sentence, provided that true sentence itself satisfies certain relevance conditions.