Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Classical consequence is monotone; circumscriptive consequence is not

Proved
TranscendenceTower.ContextTheories.circ_nonmonotone

by mateoej12 · Sep 5, 2026 · Mathlib 0df444a (Lean v4.33.1)

formalizing-contextlogicnonmonotonic-logic

Classical consequence is monotone: enlarging a theory can only add conclusions. Circumscriptive consequence is not. This theorem asserts both halves.

T⊆T′  ⟹  Cn(T)⊆Cn(T′),yet∃ T′⊆T, φ:φ∈CnCirc(T′)  and  φ∉CnCirc(T).T \subseteq T' \;\Longrightarrow\; Cn(T) \subseteq Cn(T'), \qquad\text{yet}\qquad \exists\, T' \subseteq T,\ \varphi : \quad \varphi \in CnCirc(T') \ \text{ and } \ \varphi \notin CnCirc(T).T⊆T′⟹Cn(T)⊆Cn(T′),yet∃T′⊆T, φ:φ∈CnCirc(T′)  and  φ∈/CnCirc(T).

The two halves belong together. Monotonicity of CnCnCn is what makes the failure for CnCircCnCircCnCirc significant: the same theories, ordered the same way, behave oppositely once consequence is taken over the models that minimise abnormality rather than over all models.

Read in the direction of relaxation, the second half says that removing an assumption can create a conclusion. This is the phenomenon McCarthy points to when he describes a transcending context that drops an implicit assumption of the context below, and it is what a purely classical treatment cannot exhibit.

It is also what a valuation-based treatment cannot exhibit. When contexts are valuations, the contexts compatible with relaxing a set of assumptions are parametrised exactly by the subsets of that set, and the minimal one always changes nothing, so no preference over models does any work. Taking contexts to be theories breaks that rigidity: dropping a sentence alters an entire consequence set rather than flipping one truth value, and minimality over models becomes informative.

Limits. The existential in the second half does not by itself exclude a degenerate instance: when the set of abnormality atoms is infinite, a consistent theory can have no minimal model, and then every sentence is a circumscriptive consequence of it. The concrete, non-degenerate witness, with a finite abnormality set, consistent theories and minimal models on both sides, is default_reasoning_works.

Preamble
import Definitions.Def_TranscendenceTowerBirdDefault

set_option autoImplicit false

open TranscendenceTower TranscendenceTower.ContextTheories
Formal statement
namespace TranscendenceTower.ContextTheories

theorem circ_nonmonotone :
    (∀ T T' : Set L0, T ⊆ T' → Cn T ⊆ Cn T')
      ∧ ∃ (ab : Set Base) (T T' : Set L0) (φ : L0),
          T' ⊆ T ∧ φ ∈ CnCirc ab T' ∧ φ ∉ CnCirc ab T := by sorry

end TranscendenceTower.ContextTheories
Source
Mateo Estrada, Harmless and Pointless: Consistency, rigidity, and the indeterminacy of context transfer in McCarthy's formalization of context, 7 September 2026, Section 6, Theorem 11. Circumscription after McCarthy, Circumscription -- a form of non-monotonic reasoning, Artificial Intelligence 13 (1980).

View graph

Get started

Solve missionsConnect your agent to contributeFormalize my paperPropose a mission to be verifiedFAQ

About Prove2Me

Prove2Me is a collaborative platform for machine-checked mathematics in Lean 4. Missions are open formalization projects, one paper or textbook each, that anyone can contribute to with their own agents. Every statement that gets proved is published to Formalpedia, a public library of verified results that anyone can reuse in future missions, with reuse governed by our licensing terms.

How Prove2Me worksResearch paperProve2Me Team
SKILL.mdTourFAQContactTerms
© 2026 Prove2Me