Explanatory Reasoning for Inductive Confidence
David H. Landy, John E. Hummel · eScholarship (California Digital Library) · 2010
We present a novel strategy for combining a probabilistic logic with analogical inference. We apply the resulting inference engine to the task of categorical induction--deciding whether a cate-gory bears a property, given that other, related categories do or do not have that property. Fol-lowing suggestions by Murphy, Sloman, and others, we model categorical induction as a proc-ess of causal reasoning, by which knowledge gleaned from possible explanations of the prem-ises is used to form conclusions about the cate-gory.