Skip to content

Justification logic

This page is being written. It will be expanded with a formal treatment and the relationship to the four-warrant taxonomy. In the interim, the CACM paper §“Justification logic” is the bridge from Artemov’s calculus to the chain-resident form.

Summary

The four-warrant taxonomy of Concepts is a chain-instantiated form of Artemov’s justification logic — a modal calculus that makes explicit which evidence justifies each modality. Each chain-resident proposition is paired with a typed justification term: an explicit, structural accounting of how the proposition was reached. The kernel’s type-checker verifies the composition of justification terms; what gets committed is the agent’s reasoning made structural, not the agent’s prose.

The reasoning institution’s JustifiedBy inductive type is the chain-resident formalisation: it pairs a justification term (DeclaredEvidence(iri), ObservedEvidence(iri), DerivedEvidence(iri), VerifiedEvidence(iri), and their App and SpecStr compositions) with the proposition the term justifies. The kernel verifies that the pairing is well-formed.

References

  • Artemov, S. The logic of justification. The Review of Symbolic Logic 1(4):477–513, 2008.
  • The CACM paper, §“The Compiler for AI Thought” → “Justification logic”

See also