Documentation

Studies.Yan2023

Yan (2023): Monotonicity in Intensional Contexts #

This file formalizes Chapter 4 of [Yan23], which defends an upward-monotonic semantics for desire verbs against three puzzles: Asher's puzzle and the teach-on-Tuesdays example reported in [Hei92], and Ross's paradox under want from [Crn11] after [Ros44]. Want is the necessity modal over a bouletic accessibility relation and it is ok its dual, §4.4.1, in the quantified bilateral state-based modal logic of [AvO23]. Each puzzle reduces to a free-choice inference: the monotonic step is valid on the NE-free fragment (nec_disj_intro, nec_exi_within_mono); the inference that the agent is ok with the unwanted alternative follows by □-free choice, Fact 13, from the pragmatically enriched disjunctive conclusion (ross_fc, fc_of_reinterpret); and at a desire state that supports the enriched premise the enriched conclusion is unsupported (ross_blocked, asher_blocked, heim_blocked), so the three puzzles are pragmatic failures, Table 4.3. Where no disjunction is overt, the reinterpretation function of Definition 32 supplies one: a predicate with a contextually salient sub-predicate is read as the disjunction of the predicate within the sub-predicate and outside it (reinterpret), which stays NE-free and is bilaterally equivalent to the original (reinterpret_neFree, eval_reinterpret_iff) yet parts from it under enrichment (asher_concl_enriched, asher_blocked).

Implementation notes #

References #

The reinterpretation function, Definition 32 #

def Yan2023.within {Const : Type u_3} {Pred : Type u_4} {Var : Type u_5} (P Q : Pred) (x : Var) :
QBSML.Formula Var Const Pred

within P Q x, Q within P, is P x ∧ Q x, the paper's P_Q x.

Equations
Instances For
    def Yan2023.without {Const : Type u_3} {Pred : Type u_4} {Var : Type u_5} (P Q : Pred) (x : Var) :
    QBSML.Formula Var Const Pred

    without P Q x, Q outside P, is ¬P x ∧ Q x, the paper's ¬P_Q x.

    Equations
    Instances For
      theorem Yan2023.within_neFree {Const : Type u_3} {Pred : Type u_4} {Var : Type u_5} (P Q : Pred) (x : Var) :
      (within P Q x).NEFree
      theorem Yan2023.without_neFree {Const : Type u_3} {Pred : Type u_4} {Var : Type u_5} (P Q : Pred) (x : Var) :
      (without P Q x).NEFree
      def Yan2023.reinterpret {Const : Type u_3} {Pred : Type u_4} {Var : Type u_5} (sub : Pred → Pred → Prop) [DecidableRel sub] (P : Pred) :
      QBSML.Formula Var Const Pred → QBSML.Formula Var Const Pred

      Reinterpretation by P replaces each atom whose predicate Q has P as a contextually salient sub-predicate, sub P Q, by the disjunction of Q within P and Q outside P, and commutes with every connective.

      Equations
      • One or more equations did not get rendered due to their size.
      Instances For
        theorem Yan2023.reinterpret_neFree {Const : Type u_3} {Pred : Type u_4} {Var : Type u_5} (sub : Pred → Pred → Prop) [DecidableRel sub] (P : Pred) {φ : QBSML.Formula Var Const Pred} (h : φ.NEFree) :
        (reinterpret sub P φ).NEFree

        Reinterpretation stays in the NE-free fragment.

        theorem Yan2023.reinterpret_nec_exi {Const : Type u_3} {Pred : Type u_4} {Var : Type u_5} (sub : Pred → Pred → Prop) [DecidableRel sub] (P : Pred) {Q : Pred} (h : sub P Q) (x : Var) :

        The reinterpreted want □∃x‖Qx‖_P is the disjunctive want.

        Substitution salva veritate, §4.3.2 #

        The classical equivalence of Q x and (P x ∧ Q x) ∨ (¬P x ∧ Q x) that licenses reinterpretation holds bilaterally in team semantics for unenriched formulas.

        theorem Yan2023.eval_reinterpret_iff {W : Type u_1} {Domain : Type u_2} {Const : Type u_3} {Pred : Type u_4} {Var : Type u_5} [DecidableEq W] [DecidableEq Var] [Fintype Var] [DecidableEq Domain] [Fintype Domain] (M : QBSML.Model W Domain Const Pred) (sub : Pred → Pred → Prop) [DecidableRel sub] (P : Pred) (φ : QBSML.Formula Var Const Pred) (s : Finset (QBSML.Index W Var Domain)) (b : Bool) :
        QBSML.eval M b (reinterpret sub P φ) s ↔ QBSML.eval M b φ s

        Reinterpretation is bilaterally equivalent to the original. ‖φ‖_P and φ are supported and anti-supported by the same states, for any sub, since the sub-predicate and salience conditions govern felicity rather than truth.

        theorem Yan2023.support_reinterpret_iff {W : Type u_1} {Domain : Type u_2} {Const : Type u_3} {Pred : Type u_4} {Var : Type u_5} [DecidableEq W] [DecidableEq Var] [Fintype Var] [DecidableEq Domain] [Fintype Domain] (M : QBSML.Model W Domain Const Pred) (sub : Pred → Pred → Prop) [DecidableRel sub] (P : Pred) (φ : QBSML.Formula Var Const Pred) (s : Finset (QBSML.Index W Var Domain)) :
        QBSML.support M (reinterpret sub P φ) s ↔ QBSML.support M φ s
        theorem Yan2023.antiSupport_reinterpret_iff {W : Type u_1} {Domain : Type u_2} {Const : Type u_3} {Pred : Type u_4} {Var : Type u_5} [DecidableEq W] [DecidableEq Var] [Fintype Var] [DecidableEq Domain] [Fintype Domain] (M : QBSML.Model W Domain Const Pred) (sub : Pred → Pred → Prop) [DecidableRel sub] (P : Pred) (φ : QBSML.Formula Var Const Pred) (s : Finset (QBSML.Index W Var Domain)) :

        Monotonicity, free choice and blocking, §4.4.3 #

        The pattern of Figures 4.2 and 4.3 on an arbitrary model: the monotonic step is valid, the enriched conclusion yields the unwanted possibility by □-free choice, and a desire state whose accessible worlds contain no instance of the unwanted disjunct supports the enriched premise but not the enriched conclusion.

        theorem Yan2023.nec_disj_intro {W : Type u_1} {Domain : Type u_2} {Const : Type u_3} {Pred : Type u_4} {Var : Type u_5} [DecidableEq W] [DecidableEq Var] [Fintype Var] [DecidableEq Domain] [Fintype Domain] (M : QBSML.Model W Domain Const Pred) {s : Finset (QBSML.Index W Var Domain)} {α β : QBSML.Formula Var Const Pred} (hβ : β.NEFree) (h : QBSML.support M α.nec s) :
        QBSML.support M (α.disj β).nec s

        Disjunction introduction under want, the mode of Ross's paradox in Table 4.1, is valid for an NE-free added disjunct.

        theorem Yan2023.nec_exi_within_mono {W : Type u_1} {Domain : Type u_2} {Const : Type u_3} {Pred : Type u_4} {Var : Type u_5} [DecidableEq W] [DecidableEq Var] [Fintype Var] [DecidableEq Domain] [Fintype Domain] (M : QBSML.Model W Domain Const Pred) {s : Finset (QBSML.Index W Var Domain)} {P Q : Pred} {x : Var} (h : QBSML.support M (QBSML.Formula.exi x (within P Q x)).nec s) :

        Conjunction elimination under want and ∃, the mode of Asher's and Heim's puzzles in Table 4.1, is valid.

        theorem Yan2023.fc_of_reinterpret {W : Type u_1} {Domain : Type u_2} {Const : Type u_3} {Pred : Type u_4} {Var : Type u_5} [DecidableEq W] [DecidableEq Var] [Fintype Var] [DecidableEq Domain] [Fintype Domain] (M : QBSML.Model W Domain Const Pred) {s : Finset (QBSML.Index W Var Domain)} {sub : Pred → Pred → Prop} [DecidableRel sub] {P Q : Pred} {x : Var} (hsub : sub P Q) (h : QBSML.support M (reinterpret sub P (QBSML.Formula.exi x (QBSML.Formula.pred Q x)).nec).enrich s) :

        By free choice from the enriched reinterpreted want, (14) and (27), the agent is ok with Q within P and with Q outside P.

        theorem Yan2023.support_enrich_nec_predc {W : Type u_1} {Domain : Type u_2} {Const : Type u_3} {Pred : Type u_4} {Var : Type u_5} [DecidableEq W] [DecidableEq Var] [Fintype Var] [DecidableEq Domain] [Fintype Domain] (M : QBSML.Model W Domain Const Pred) {s : Finset (QBSML.Index W Var Domain)} {A : Pred} {c : Const} (hs : s.Nonempty) (hacc : ∀ i ∈ s, (M.access i.world).Nonempty) (hA : ∀ i ∈ s, ∀ w ∈ M.access i.world, FirstOrder.Language.ModalStructure.relInterp₁ M (QBSML.predSymb A) w (FirstOrder.Language.ModalStructure.constInterp M ((FirstOrder.Language.monadic Pred).con c) w)) :

        The enriched want of a constant atom is supported at a nonempty state each of whose indices sees some world, all of them verifying the atom.

        theorem Yan2023.not_support_enrich_nec_disj {W : Type u_1} {Domain : Type u_2} {Const : Type u_3} {Pred : Type u_4} {Var : Type u_5} [DecidableEq W] [DecidableEq Var] [Fintype Var] [DecidableEq Domain] [Fintype Domain] (M : QBSML.Model W Domain Const Pred) {s : Finset (QBSML.Index W Var Domain)} {A B : Pred} {c : Const} (hs : s.Nonempty) (hB : ∀ i ∈ s, ∀ w ∈ M.access i.world, ¬FirstOrder.Language.ModalStructure.relInterp₁ M (QBSML.predSymb B) w (FirstOrder.Language.ModalStructure.constInterp M ((FirstOrder.Language.monadic Pred).con c) w)) :

        The enriched disjunctive want [□(A c ∨ B c)]⁺ is unsupported at a nonempty state none of whose accessible worlds verifies B c.

        theorem Yan2023.support_enrich_nec_exi_within {W : Type u_1} {Domain : Type u_2} {Const : Type u_3} {Pred : Type u_4} {Var : Type u_5} [DecidableEq W] [DecidableEq Var] [Fintype Var] [DecidableEq Domain] [Fintype Domain] (M : QBSML.Model W Domain Const Pred) {s : Finset (QBSML.Index W Var Domain)} {P Q : Pred} {x : Var} (hs : s.Nonempty) (hacc : ∀ i ∈ s, (M.access i.world).Nonempty) (hPQ : ∀ i ∈ s, ∀ w ∈ M.access i.world, ∃ (d : Domain), FirstOrder.Language.ModalStructure.relInterp₁ M (QBSML.predSymb P) w d ∧ FirstOrder.Language.ModalStructure.relInterp₁ M (QBSML.predSymb Q) w d) :

        The enriched want [□∃x(P x ∧ Q x)]⁺ is supported at a nonempty state each of whose indices sees some world, all of them holding an individual that is P and Q.

        theorem Yan2023.support_enrich_nec_exi_pred_of_within {W : Type u_1} {Domain : Type u_2} {Const : Type u_3} {Pred : Type u_4} {Var : Type u_5} [DecidableEq W] [DecidableEq Var] [Fintype Var] [DecidableEq Domain] [Fintype Domain] (M : QBSML.Model W Domain Const Pred) {s : Finset (QBSML.Index W Var Domain)} {P Q : Pred} {x : Var} (h : QBSML.support M (QBSML.Formula.exi x (within P Q x)).nec.enrich s) :

        The enriched premise supports the enriched unreinterpreted conclusion, since the same witnesses serve [□∃x Q x]⁺.

        theorem Yan2023.not_support_enrich_reinterpret {W : Type u_1} {Domain : Type u_2} {Const : Type u_3} {Pred : Type u_4} {Var : Type u_5} [DecidableEq W] [DecidableEq Var] [Fintype Var] [DecidableEq Domain] [Fintype Domain] (M : QBSML.Model W Domain Const Pred) {s : Finset (QBSML.Index W Var Domain)} {sub : Pred → Pred → Prop} [DecidableRel sub] {P Q : Pred} {x : Var} (hsub : sub P Q) (hs : s.Nonempty) (hPQ : ∀ i ∈ s, ∀ w ∈ M.access i.world, ∀ (d : Domain), FirstOrder.Language.ModalStructure.relInterp₁ M (QBSML.predSymb Q) w d → FirstOrder.Language.ModalStructure.relInterp₁ M (QBSML.predSymb P) w d) :

        The enriched reinterpreted want [□∃x‖Q x‖_P]⁺ is unsupported at a nonempty state whose accessible worlds have every Q within P: by free choice it would need an accessible Q outside P.

        inductive Yan2023.Var :
        Instances For
          @[instance_reducible]
          instance Yan2023.instDecidableEqVar :
          DecidableEq Var
          Equations
          @[instance_reducible]
          instance Yan2023.instFintypeVar :
          Fintype Var
          Equations

          Ross's paradox under desire, (3), (26), Figure 4.2 #

          The predicates of Ross's paradox are sending and burning the letter.

          Instances For
            @[instance_reducible]
            Equations
            def Yan2023.instReprRossPred.repr :
            RossPred → ℕ → Std.Format
            Equations
            Instances For
              @[instance_reducible]
              Equations

              sendL is SEND a, the constant a being the letter.

              Equations
              Instances For

                John's desire model has one desire world, reflexively accessible, where the letter is sent and not burnt.

                Equations
                Instances For
                  def Yan2023.rossState :
                  Finset (QBSML.Index Unit Var Unit)

                  John's desire state is the desire world with the empty assignment.

                  Equations
                  Instances For

                    The monotonic step (26a) to (26b) is semantically valid.

                    From the enriched disjunctive want, it is ok to send, (26c), and ok to burn, (26d).

                    The enriched premise [□SEND a]⁺ is supported at John's desire state.

                    The enriched disjunctive want [□(SEND a ∨ BURN a)]⁺ is not supported there, since free choice would require an accessible burn world.

                    A desire model for a salient sub-predicate #

                    Nicholas wants a free trip, (1); I want to teach on Tuesdays, (2). At the desire world every predicate holds of the one individual; at the other world only the wanted predicate, so the sub-predicate's denotation is a proper subset globally.

                    def Yan2023.IsSubPred {W : Type u_1} {Domain : Type u_2} {Const : Type u_3} {Pred : Type u_4} (M : QBSML.Model W Domain Const Pred) (P Q : Pred) :

                    P is a sub-predicate of Q in M when its denotation is a proper subset of Q's.

                    Equations
                    • One or more equations did not get rendered due to their size.
                    Instances For
                      def Yan2023.desireModel {Pred : Type u_4} (Q : Pred) :
                      QBSML.Model Bool Unit Unit Pred

                      In the two-world desire model for the wanted predicate Q, only the desire world true is accessible; every predicate holds there and only Q at false.

                      Equations
                      Instances For
                        def Yan2023.desireState :
                        Finset (QBSML.Index Bool Var Unit)

                        The desire state is the desire world with the empty assignment.

                        Equations
                        Instances For
                          theorem Yan2023.isSubPred_desireModel {Pred : Type u_4} {P Q : Pred} (h : P ≠ Q) :

                          Any other predicate is a sub-predicate of Q in desireModel Q.

                          The enriched premise [□∃x(P x ∧ Q x)]⁺ is supported at the desire state.

                          The enriched unreinterpreted conclusion [□∃x Q x]⁺ is supported at the desire state.

                          theorem Yan2023.desireModel_blocked {Pred : Type u_4} {sub : Pred → Pred → Prop} [DecidableRel sub] {P Q : Pred} (hsub : sub P Q) :

                          The enriched reinterpreted conclusion [□∃x‖Q x‖_P]⁺ is not supported at the desire state: every Q in the desire world is P. The two classically equivalent conclusions part under enrichment.

                          Asher's puzzle, (1), (27), Figure 4.3 #

                          The predicates of Asher's puzzle are being free and being a trip on the Concorde.

                          Instances For
                            @[instance_reducible]
                            Equations
                            @[instance_reducible]
                            Equations
                            def Yan2023.instReprAsherPred.repr :
                            AsherPred → ℕ → Std.Format
                            Equations
                            Instances For

                              FREE is the contextually salient sub-predicate of TRIP.

                              Equations
                              Instances For
                                @[instance_reducible]
                                Equations
                                @[reducible, inline]

                                Nicholas wants a free trip, □∃x(FREE x ∧ TRIP x), (27a).

                                Equations
                                Instances For
                                  @[reducible, inline]

                                  Nicholas wants a trip, □∃x TRIP x, (27b).

                                  Equations
                                  Instances For

                                    The side condition of Definition 32 holds globally.

                                    theorem Yan2023.asher_monotone {W : Type u_5} {Domain : Type u_6} [DecidableEq W] [DecidableEq Domain] [Fintype Domain] (M : QBSML.Model W Domain Unit AsherPred) {s : Finset (QBSML.Index W Var Domain)} (h : QBSML.support M asherPremise s) :

                                    The monotonic step (27a) to (27b) is semantically valid on any model.

                                    theorem Yan2023.asher_fc {W : Type u_5} {Domain : Type u_6} [DecidableEq W] [DecidableEq Domain] [Fintype Domain] (M : QBSML.Model W Domain Unit AsherPred) {s : Finset (QBSML.Index W Var Domain)} (h : QBSML.support M (reinterpret subFree AsherPred.free asherConcl).enrich s) :

                                    The enriched reinterpreted conclusion licenses being ok with a free trip, (27c), and with a non-free trip, (27d).

                                    Nicholas's desire state supports the enriched premise but not the enriched reinterpreted conclusion: being ok with a non-free trip is licensed only by a premise never granted.

                                    Heim's example, (2), (15)–(16) #

                                    TEACH is reinterpreted by its salient sub-predicate teaching on Tuesday; the paper leaves the derivation as parallel to Asher's.

                                    The predicates of Heim's example are being on Tuesday and being a teaching next semester.

                                    Instances For
                                      @[instance_reducible]
                                      Equations
                                      @[instance_reducible]
                                      Equations
                                      def Yan2023.instReprHeimPred.repr :
                                      HeimPred → ℕ → Std.Format
                                      Equations
                                      Instances For

                                        TUESDAY is the contextually salient sub-predicate of TEACH.

                                        Equations
                                        Instances For
                                          @[instance_reducible]
                                          Equations
                                          @[reducible, inline]

                                          I want to teach next semester, □∃x TEACH x, (15b).

                                          Equations
                                          Instances For
                                            theorem Yan2023.heim_fc {W : Type u_5} {Domain : Type u_6} [DecidableEq W] [DecidableEq Domain] [Fintype Domain] (M : QBSML.Model W Domain Unit HeimPred) {s : Finset (QBSML.Index W Var Domain)} (h : QBSML.support M (reinterpret subTuesday HeimPred.tuesday heimConcl).enrich s) :

                                            The enriched reinterpreted conclusion licenses being ok with teaching on a non-Tuesday, (16c).

                                            The desire state of a speaker who teaches only on Tuesdays supports the enriched premise (15a) but not the enriched reinterpreted conclusion.