Documentation

Studies.McKayVanInwagen1977

McKay and van Inwagen (1977): Counterfactuals with Disjunctive Antecedents #

This file formalizes the defence in [McKVI77] of the variably strict conditional of [Lew73b] against simplification of disjunctive antecedents, the inference from a counterfactual with a disjunctive antecedent to the counterfactual from either disjunct. Two arguments are stated on the substrate's variably strict conditional. The bumper-crop argument: the English sentence is false while the disjunctive-closure reading is true, so the sentence is regimented as the conjunction of the per-disjunct counterfactuals, which is false as required. The Spain counterexample: If Spain had fought on the Axis side or the Allied side, Spain would have fought on the Axis side is acceptable, yet simplification would yield the absurd counterfactual from the Allied disjunct.

Implementation notes #

Both readings reduce to Conditional.variablyStrictImp, evaluated on the disjunction or conjoined over the disjuncts, with propositions as sets of worlds of enumerated world types. Similarity is a ranking of the worlds, so it is total as [Lew73b]'s comparative similarity requires, and with finitely many worlds each verdict is computed on the closest antecedent-worlds (holds).

References #

theorem McKayVanInwagen1977.holds {W : Type u_1} [Finite W] {α : Type u_2} [LinearOrder α] (d : W → W → α) (p q : Set W) :
Conditional.variablyStrictImp (fun (w : W) => Preorder.lift (d w)) p q = Conditional.closestImp (fun (w : W) => Preorder.lift (d w)) p q

On a ranking of finitely many worlds the variably strict conditional is the conditional of the closest antecedent-worlds.

The Bumper Crop Argument #

The critics argue that Lewis's semantics is wrong using the sentence:

S: "If we were to have good weather this summer or if the sun were to grow cold, we would have a bumper crop."

They claim S is equivalent to the regimented counterfactual S* = (goodWeather ∨ sunCold) > bumperCrop. Since Lewis's semantics makes S* true (the closest (goodWeather ∨ sunCold)-world has good weather, hence bumper crop) but S is clearly false, Lewis must be wrong.

McKay & Van Inwagen's rebuttal: premise (2) is false — S is NOT equivalent to S*. The correct regimentation of S is the conjunction (goodWeather > bumperCrop) ∧ (sunCold > bumperCrop), which IS false on Lewis's semantics, matching the English judgment.

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

      A good-weather world is more like the actual world than any world in which the sun grows cold, which the paper takes to be "surely true".

      Equations
      Instances For
        @[reducible, inline]

        Similarity to the actual world, by rank.

        Equations
        Instances For
          @[reducible, inline]

          The good-weather worlds.

          Equations
          Instances For
            @[reducible, inline]

            The worlds where the sun grows cold.

            Equations
            Instances For
              @[reducible, inline]

              The bumper-crop worlds: the good-weather world.

              Equations
              Instances For

                S* (Lewis disjunctive closure) is TRUE: the closest (goodWeather ∨ sunCold)-world is a good-weather world. This is premise (3) of the critics' argument.

                The conjunction regimentation is FALSE: "if the sun grew cold, we'd have a bumper crop" is false. This matches the English judgment that S is false.

                Lewis's disjunctive closure is true while the conjunction regimentation is false. Since the English sentence S is false (matching the conjunction) while S* is true (matching Lewis), S ≠ S*: premise (2) is false.

                The Spain Example #

                "Neither. Spain did not enter the war. But if she had fought on one side or the other, it would have been the Axis."

                That is, we assert: (Axis ∨ Allies) > Axis. This is true on Lewis's semantics (Spain was ideologically closer to the Axis).

                But if SDA were valid, it would follow that: Allies > Axis — "If Spain had fought on the Allied side, Spain would have fought on the Axis side." This is absurd.

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

                    Axis-worlds rank closer to actual than Allied ones: the paper asserts the counterfactual without stating an ordering, and this is the ranking that makes it true.

                    Equations
                    Instances For
                      @[reducible, inline]

                      Similarity to the actual world, by rank.

                      Equations
                      Instances For
                        @[reducible, inline]

                        The worlds where Spain fought with the Axis.

                        Equations
                        Instances For
                          @[reducible, inline]

                          The worlds where Spain fought with the Allies.

                          Equations
                          Instances For

                            Lewis's disjunctive-closure reading is TRUE: the closest (Axis ∨ Allies)-world is the Axis-world. The English sentence "if she had fought on one side or the other, it would have been the Axis" is acceptable.

                            The absurd SDA simplification: "If Spain had fought on the Allied side, Spain would have fought on the Axis side" is false. This is what the SDA schema would derive from spain_lewis_true.

                            theorem McKayVanInwagen1977.sda_invalid :
                            ∃ (W : Type) (ord : W → Preorder W) (A : Set W) (B : Set W) (C : Set W), ∃ w ∈ Conditional.variablyStrictImp ord (A ∪ B) C, w ∉ Conditional.variablyStrictImp ord B C

                            SDA is not a valid schema for counterfactuals. The Spain example: (Axis ∨ Allies) > Axis is true, but Allies > Axis is false.