Asher & Pelletier (2013): More Truths about Generic Truth #
[AP13] (ch. 12 of Genericity, Mari, Beyssade & Del Prete eds., OUP)
defends the modal-quantifier analysis of generics of [AM95] and
[PA97] — "φs ψ" is ∀x(φ(x) > ψ(x)), with > the weak conditional of
[AM91]'s commonsense entailment — refined so that the consequent is
evaluated per individual: for each a, in the worlds where a is a normal φ.
Here that is rendered as restrictor-indexed normality orderings — each restrictor
class carries its own Preorder, and a generic holds iff its scope holds
throughout Normality.optimal of its restrictor's ordering. Covers §12.3–12.4;
the chapter's accommodation and double-genericity machinery (§12.6, §12.8) is
not formalized.
Per-individual evaluation (§12.3, exx. 7–8) #
For ∀x(φ(x) > ψ(x)), the consequent ψ is evaluated for each individual a at the worlds where a is a normal φ — under the normality ordering of a's restrictor class.
Worlds in which Opus is a bird (all of them, by taxonomy).
Equations
Instances For
Worlds in which Opus is a penguin.
Equations
Instances For
Bird-normality: the bird default "birds fly" processed into the ordering.
Equations
Instances For
Penguin-normality: the penguin default "penguins don't fly" processed into the ordering.
Equations
Instances For
Ex. (8): "Birds fly" — Opus flies throughout the normal Opus-bird worlds.
Ex. (7): "Penguins don't fly" — Opus is grounded throughout the normal
Opus-penguin worlds. With birds_fly, both generics hold simultaneously:
they are evaluated at different normal-world sets.
"The normal Opus-penguin worlds are not normal Opus-bird worlds": the two evaluation sets are disjoint, so neither generic disturbs the other.
Contextual construals of normality (§12.3, ex. 9) #
"Turtles live to be 100" is true under the Aristotelian/teleological construal and false under the statistical one: the "give" in what counts as a normal world sits in the choice of ordering, fixed by context and discourse.
Equations
- AsherPelletier2013.instDecidableEqTurtleWorld x✝ y✝ = if h : x✝.ctorIdx = y✝.ctorIdx then isTrue ⋯ else isFalse ⋯
Tim lives to be 100 in this world.
Equations
Instances For
Teleological construal: promotes telos-fulfilling worlds.
Equations
Instances For
Statistical construal: promotes the statistically typical early death.
Equations
Instances For
Under the teleological construal the generic is true.
Under the statistical construal the same generic is false.
Against the probabilistic account (§12.4) #
The "too weak" argument against [Coh99]'s Pr(ψ | φ) > 1/2 semantics: a margin just over half — the chapter's cats have tails in 50.05% of the normal cases, its coin comes up heads 50.000000001% of the time — verifies a generic the modal account rejects, since the normal cases do not settle the scope.
Tails in 11 of the 20 equally normal cat-cases: just over half.
Equations
- AsherPelletier2013.tailed w = (↑w < 11)
Instances For
Equations
- AsherPelletier2013.instDecidablePredFinOfNatNatTailed w = (↑w).decLt 11
Cohen's majority GEN verifies "Cats have tails", but the generic fails on the modal account: some normal case lacks tails.