The material theory of object-induction and the universal optimality of meta-induction: Two complementary accounts.

This paper brings together two accounts of induction that appear to be in opposition: John Norton's material account of induction (2003, 2010, manuscript) and Schurz' account of the universal optimality of meta-induction (2008, 2017, 2019). According to the material account of induction, all reliable rules of 'induction' are local and context-dependent. Here "induction" is understood in the sense of object-induction, i.e., induction applied at the object-level of events. In contrast, Schurz' account proceeds from the demonstration that there are universally optimal rules of meta-induction, i.e., rules of induction applied at the level of competing methods of prediction, including methods of object-induction. The two accounts are not in opposition; on the contrary, they agree on most questions related to the problem of induction. Beyond this agreement the two accounts are complementary: the material account suffers from a justificational circularity or regress problem that the meta-induction account can solve. On the other hand, the meta-inductive account abstracts from domain-specific aspects of object-induction that are supplied by the material account.