Documentation

Linglib.Phenomena.Modality.Studies.Booth2022

Booth 2022 — Bilateral inquisitive minimal-cover semantics for #

@cite{booth-2022}

A self-contained study file formalizing the bilateral inquisitive semantics of @cite{booth-2022} for necessity-modal-with-disjunction sentences (□(p ∨ q)). The novel content is the minimal-cover requirement on 's positive interpretation, which derives the Independence inferences

□(p ∨ q) ⊨ ◇(p ∧ ¬q) and □(p ∨ q) ⊨ ◇(q ∧ ¬p)

(Booth Fact 9). Pure-bilateral analyses without minimal-cover (BSML+, @cite{aloni-2022}) do not derive Independence; pure-inquisitive analyses without bilateral negation (@cite{ciardelli-groenendijk-roelofsen-2018}) do not derive the duality between and . Booth's bilateral inquisitive propositions combine both.

Substrate alignment #

Out of scope #

§1 Cover and minimal cover (Booth Section 2.1) #

Booth's □φ differs from Kratzerian necessity by requiring not just that the alternatives of ⟦φ⟧⁺ cover R(w), but that they form a minimal cover — no proper subset of the alternatives still covers R(w). This is what derives Independence inferences (Fact 9): each alternative must be "needed", so no single alternative dominates.

Expressed via mathlib's Minimal predicate (mirrors Question.alt's use of Maximal — Booth's alt and m-cover are dual instances of the order-theoretic extremality pattern).

def Phenomena.Modality.Studies.Booth2022.IsCover {W : Type u_1} (C : Set (Set W)) (S : Set W) :

Booth §2.1: C covers S iff S ⊆ ⋃C.

Equations
Instances For
    def Phenomena.Modality.Studies.Booth2022.IsMinCover {W : Type u_1} (C : Set (Set W)) (S : Set W) :

    Booth §2.1: C is a minimal cover (m-cover) of S iff C covers S and no proper subfamily C' ⊂ C covers S. Expressed via mathlib's Minimal.

    Equations
    Instances For
      theorem Phenomena.Modality.Studies.Booth2022.IsMinCover.isCover {W : Type u_1} {C : Set (Set W)} {S : Set W} (h : IsMinCover C S) :

      §2 Bilateral inquisitive propositions (Booth Def 10) #

      Booth Def 10: a bilateral inquisitive proposition is a paired pos/neg : Question W with no substantive overlap — only the inconsistent (empty) state may both verify and falsify φ. The subset-closure and -membership requirements (Booth Def 10 bullets 2 and the implicit ∅ ∈ P°) are baked into Question.

      • Positive interpretation: states verifying the formula.

      • Negative interpretation: states falsifying the formula.

      • no_overlap (s : Set W) : s self.poss self.negs =

        No substantive overlap: pos.props ∩ neg.props ⊆ {∅}. The reverse {∅} ⊆ pos.props ∩ neg.props holds for free since both Questions contain (Question.contains_empty).

      Instances For

        Booth Def 14, ¬-clause: bilateral negation is the bundled-record swap. Self-inverse syntactically (negate (negate φ) = φ by rfl).

        Equations
        Instances For

          BilatInqProp is a bilateral structure in the sense of Core.Logic.Bilateral.IsBilateral. The instance is rfl-trivial because negate is bundled-record swap. Sixth consumer of the IsBilateral substrate (alongside BSML, QBSML, BUS, ICDRT, Truthmaker BilProp).

          Booth Def 14, atomic clause: ⟦p⟧⁺ = ↓{V(p)}, ⟦p⟧⁻ = ↓{W \ V(p)}. Encoded with Question.declarative since ↓{X} = declarative X.

          Equations
          Instances For

            Booth Def 14, ∨-clause: ⟦φ ∨ ψ⟧⁺ = ⟦φ⟧⁺ ∪ ⟦ψ⟧⁺ (inquisitive disjunction at the props level, = Question.⊔); ⟦φ ∨ ψ⟧⁻ = ⟦φ⟧⁻ ∩ ⟦ψ⟧⁻ (= Question.⊓).

            Equations
            Instances For

              Booth Def 14, ∧-clause via the derivation ⟦φ ∧ ψ⟧ = ⟦¬(¬φ ∨ ¬ψ)⟧ — direct unfolding gives pos = φ.pos ⊓ ψ.pos, neg = φ.neg ⊔ ψ.neg. The Booth-equivalence conj φ ψ = negate (disj (negate φ) (negate ψ)) holds by rfl.

              Equations
              Instances For

                §3 Necessity and possibility (Booth Def 14) #

                R : W → Set W is the relevant-worlds accessibility relation (equivalent in expressive power to Core.Logic.Intensional.AccessRel W = W → W → Prop; Booth uses the curried W → Set W form throughout his Def 14, which we mirror).

                Booth Def 14, □-clause: ⟦□φ⟧⁺ = ↓{ {w | R(w) ≠ ∅ ∧ alt⁺(⟦φ⟧) m-covers R(w)} }, ⟦□φ⟧⁻ = ↓{ {w | ∃ R' ⊆ R(w), R' ≠ ∅ ∧ alt⁻(⟦φ⟧) m-covers R'} }.

                The no-overlap proof structurally inducts via φ.no_overlap: any non-empty state s in both polarities yields a world w ∈ s, hence a witness v ∈ R(w) covered by both alt⁺(φ.pos) and alt⁻(φ.neg) — giving alternatives α ∈ φ.pos.props and β ∈ φ.neg.props containing v. Downward closure gives {v} ∈ φ.pos ∩ φ.neg, contradicting φ.no_overlap.

                Equations
                • One or more equations did not get rendered due to their size.
                Instances For

                  Booth Def 14, ◇-clause via duality: ⟦◇φ⟧ = ⟦¬□¬φ⟧.

                  Equations
                  Instances For

                    §4 Truth and falsity (Booth Def 17) #

                    A world w makes φ true in model (W, R, V) iff {w} ∈ ⟦φ⟧⁺, and false iff {w} ∈ ⟦φ⟧⁻. Since Questions are subset-closed, this is equivalent to ∃ s ∈ ⟦φ⟧°, w ∈ s for a non-empty witness.

                    Booth Def 17: world w is true at φ iff the singleton {w} verifies φ.

                    Equations
                    Instances For

                      Booth Def 17: world w is false at φ iff the singleton {w} falsifies φ.

                      Equations
                      Instances For

                        Truth and falsity are mutually exclusive (no world is both true and false), since by no_overlap any state in both polarities is .

                        theorem Phenomena.Modality.Studies.Booth2022.isTrue_possibility_iff {W : Type u_1} (R : WSet W) (φ : BilatInqProp W) (w : W) :
                        isTrue (BilatInqProp.possibility R φ) w R'R w, R'.Nonempty IsMinCover φ.pos.alt R'

                        Characterization of isTrue for possibility: world w makes ◇φ true iff there exists a non-empty R' ⊆ R w minimally covered by alt φ.pos. The possibility := negatenecessitynegate derivation cancels the two negations, exposing φ.pos directly. Used to bypass the verbose show block in proofs about possibility.

                        §5 Per-constructor algebra of alt (Booth Compactness substrate) #

                        Per-constructor equations for Question.alt on BilatInqProp's positive interpretation. Used by the worked example (§6), the general Independence theorem (§7), and downstream Booth Compactness (eq_iSup_declarative_alt_of_exists_alt) consumers. The atomic-case private corollaries (alt_disj_atom_eq_pair, alt_conj_atom_negate_eq_singleton) are derived from the public generalizations.

                        alt of atom V is the singleton {V}. Direct corollary of Question.alt_declarative.

                        alt of negate φ's positive interpretation is alt of φ's negative interpretation. By definition of negate, structural rfl.

                        theorem Phenomena.Modality.Studies.Booth2022.alt_necessity_pos {W : Type u_1} (R : WSet W) (φ : BilatInqProp W) :
                        (BilatInqProp.necessity R φ).pos.alt = {{w : W | (R w).Nonempty IsMinCover φ.pos.alt (R w)}}

                        alt of necessity R φ's positive interpretation is the singleton of the witness w-set, since necessity uses Question.declarative.

                        theorem Phenomena.Modality.Studies.Booth2022.alt_disj_pos_eq_union {W : Type u_1} (φ ψ : BilatInqProp W) (hφψ : aφ.pos.alt, aψ.pos.props) (hψφ : bψ.pos.alt, bφ.pos.props) :
                        (φ.disj ψ).pos.alt = φ.pos.alt ψ.pos.alt

                        General non-Hurford alt of disjunction: when no φ-alt entails ψ and no ψ-alt entails φ (the "non-Hurford" condition lifted from atoms to arbitrary subformulas), alt (disj φ ψ).pos = alt φ.pos ∪ alt ψ.pos. The atomic case (alt_disj_atom_eq_pair) is a specialization.

                        §6 Worked example: Independence inference on a 3-world model #

                        A concrete witness that the m-cover semantics derives Booth Fact 9 (Independence Inferences). We work on W₄ = Bool × Bool (subsets of {p, q}), with V p = {(true, _)} and V q = {(_, true)}, and constant accessibility R₃ w := V(p) ∪ V(q) (the 3 worlds where p ∨ q is true).

                        In this model {V(p), V(q)} minimally covers R₃ w because removing either alternative leaves a gap (V(p) alone misses (false, true); V(q) alone misses (true, false)). Thus □(p ∨ q) is true, and the Vp-only world (true, false) lies in R₃ w, witnessing the existential in ◇(p ∧ ¬q)'s positive-side definition.

                        @[reducible, inline]

                        4-world model: subsets of {p, q} indexed by Bool × Bool.

                        Equations
                        Instances For

                          Valuation: q true at worlds with second coordinate true.

                          Equations
                          Instances For

                            Constant 3-world accessibility: R₃ w = VpVq, the worlds where p ∨ q is true (excluding (false, false)).

                            Equations
                            Instances For

                              Pivotal world facts #

                              Question-algebraic helpers (specializations of §5 helpers) #

                              The Independence-witness theorems #

                              □(p ∨ q) holds at (true, true) in the 3-world model. Both (R₃ w).Nonempty and IsMinCover {Vp, Vq} (Vp ∪ Vq) are discharged: the latter requires that any cover-subset must contain both Vp (witnessed by (true, false) ∈ Vp \ Vq) and Vq (witnessed by (false, true) ∈ Vq \ Vp).

                              ◇(p ∧ ¬q) holds at (true, true) in the 3-world model. The Vp-only world (true, false) witnesses the existential in the possibility's positive-side def: it lies in R₃ (true, true) and {(true, false)} is m-covered by {Vp ∩ Vqᶜ}.

                              Independence inference on the 3-world model: □(p ∨ q) and ◇(p ∧ ¬q) are jointly true at (true, true). This is a concrete witness that the m-cover semantics derives Booth Fact 9 — Kratzerian cover semantics on the same model would validate □(p ∨ q) but leave ◇(p ∧ ¬q) underivable.

                              §7 Compactness equations (Booth Fact 5) per constructor #

                              For each BilatInqProp constructor, the compactness equation (... constructor ...).pos = ⨆ p ∈ alt _.pos, declarative p (and the dual .neg form where it differs). Each proof discharges the ∀ p ∈ Q.props, ∃ q ∈ alt Q, p ⊆ q hypothesis of Question.eq_iSup_declarative_alt_of_exists_alt.

                              These are the building blocks for proving compactness of any specific BilatInqProp formula. (The fully general statement for arbitrary formulas requires an inductive BSMLFormula type with an interpretation function; that's deferred.)

                              Compactness equation for atom V's positive interpretation. (atom V).pos = declarative V, with alt = {V}; every prop ⊆ V extends to V trivially.

                              Compactness for negate φ's positive interpretation reduces to compactness of φ.neg (since (negate φ).pos = φ.neg by rfl).

                              Compactness equation for necessity R φ's positive interpretation: a single declarative whose alt is the singleton witness w-set.

                              theorem Phenomena.Modality.Studies.Booth2022.alt_necessity_neg {W : Type u_1} (R : WSet W) (φ : BilatInqProp W) :
                              (BilatInqProp.necessity R φ).neg.alt = {{w : W | R'R w, R'.Nonempty IsMinCover φ.neg.alt R'}}

                              The alt of necessity's .neg is the singleton of the existential witness w-set (same shape as alt_necessity_pos with the existential substituted for the m-cover R(w) form).

                              theorem Phenomena.Modality.Studies.Booth2022.pos_eq_iSup_alt_disj {W : Type u_1} (φ ψ : BilatInqProp W) ( : pφ.pos.props, qφ.pos.alt, p q) ( : pψ.pos.props, qψ.pos.alt, p q) (hφψ : aφ.pos.alt, aψ.pos.props) (hψφ : bψ.pos.alt, bφ.pos.props) :
                              (φ.disj ψ).pos = p(φ.disj ψ).pos.alt, Core.Question.declarative p

                              Compactness for disj φ ψ's positive interpretation under summand pos-compactness + non-Hurford on alts. The alt of the disj is the union of summand alts (alt_disj_pos_eq_union); each prop in the disj's pos comes from one summand's pos and lifts to its alt.

                              Compactness for possibility R φ's positive interpretation. Direct via duality: (possibility R φ).pos = (necessity R (negate φ)).neg, and the latter is compact via neg_eq_iSup_alt_necessity.

                              §8 The Independence inference, general meta-language form (Booth Fact 9) #

                              The headline theorem of @cite{booth-2022}: when p ∨ q is non-Hurford (neither disjunct entails the other), □(p ∨ q) ⊨ ◇(p ∧ ¬q) and □(p ∨ q) ⊨ ◇(q ∧ ¬p).

                              We prove this for atomic disjunctions atom Vpatom Vq over an arbitrary accessibility relation R and arbitrary world type W. The proof uses minimality of the {Vp, Vq} cover to derive ∃ v ∈ R w, v ∈ Vp \ Vq — exactly the witness needed for the possibility existential. (Booth Fact 9 in full generality requires Compactness-of-Alternatives over arbitrary φ; the atomic case here captures the structural content for on disjunctions of atoms.)

                              theorem Phenomena.Modality.Studies.Booth2022.independence_p_not_q {W : Type u_1} (R : WSet W) (Vp Vq : Set W) (h_non_hurford : ¬Vp Vq ¬Vq Vp) (w : W) :

                              Booth Fact 9 (Object-Language Independence, atomic case): when p ∨ q is non-Hurford (¬ VpVq ∧ ¬ VqVp), the truth of □(p ∨ q) at w entails the truth of ◇(p ∧ ¬q) at w.

                              Proof: from IsMinCover {Vp, Vq} (R w), minimality forces {Vq} alone to NOT cover R w (else by Minimal.le_of_le, {Vp, Vq} ⊆ {Vq}, so Vp = Vq, contradicting non-Hurford). Hence ∃ v ∈ R w, v ∉ Vq. Combined with the cover R w ⊆ VpVq, we get v ∈ Vp \ Vq. Then R' := {v} witnesses the existential in ◇(p ∧ ¬q)'s positive-side definition.