A Computational Definition of 'Consilience’

Abstract

This paper defines in a formal and computational way the notion of ‘consilience’, a term introduced by Whewell in 1847 for the evaluation of scientific theories. Informally, as it has been used to date, a model or theory is ‘consilient’ if it is predictive, explanatory and unifies the evidence. Centred in a constructive framework, where new terms can be introduced, an initial and more classical approach based on the inverse idea of separation soon manifests the problem of the introduction of fantastic concepts to unify disparate sub-theories. Our second approach is constructed by using a detailed evaluation of the relationship between the theory and the evidence by means of reinforcement propagation. With the use of reinforcement, fantastic concepts can be better detected and the role of consilience for theory construction and revision can be specialised for different inference mechanisms like explanatory induction, abduction, deduction and analogy.