Structural rules for abduction

Abstract

Atocha Aliseda's Abductive Reasoning (2006) gives a structural characterization of the "forward" explanatory reasoning from a theory to observational data. This paper asks whether there are any interesting structural rules for the "backward" abductive reasoning from observations to explanatory theories. Ignoring statistical cases, a partial explication of abduction is converse deductive explanation: h is abducible from e iff h deductively explains e. This relation of abducibility trivially satisfies Converse Entailment (if h entails e, then h is abducible from e), but it does not generally satisfy Converse Consequence (if h is abducible from e and g entails h, then g is abducible from e), since deductive explanation is not always transitive.