The Unnaturalness of Grue

Abstract

A category of non-standard predicates was introduced by Goodman (1954) while attempting to recast the old riddle of induction in terms amenable to solution within confirmation theory. The New Riddle proved as intractable as the old one but the category of predicates, “mutant” ones, may assist us in understanding cognitive development from neonate vacuity to linguistically-informed rational inquiry. This paper proposes a naturalistic explanation of why we tend to reject grue-type predicates as proper bases for induction. Its conclusion is that such predicates violate requirements on normal predicates of languages that are capable of being learned by humans. The explanation does not itself directly address standard epistemological questions associated with mutant predicates but instead focusses on the pragmatic bases of such epistemic practices as induction and finds them unfulfilled by mutant predicates.

Earlier versions of this paper were read at Uppsala, Sweden, and Rutgers University. Special thanks to Susan Haack and John Bickle for helpful comments. Many of the topics discussed here are considered at greater length in Cognitive Practices (1994).