von Fintel & Iatridou (2005): What to Do If You Want to Go to Harlem #
This file formalizes [vFI05]'s analysis of anankastic conditionals, the
Harlem sentence if you want to go to Harlem, you have to take the A train (1). Three analyses
in [Kra91]'s framework fail. The obvious one, the if-clause restricting the modal base
of a teleological have to ordered by the actual goals (11), founders on the Hoboken problem:
when you actually want to go to Hoboken the best want-Harlem worlds take the PATH train, so the
sentence is predicted false. [Sab01]'s analysis adds the hypothetical goal to the ordering
source instead (14), and founders on conflicting goals: with Hoboken and Harlem inconsistent,
the best worlds achieve either, so the sentence is again false (13). The nested analysis, a
covert closeness modal restricted by the if-clause over the teleological one (15), (16),
fails where the closest want-Harlem worlds keep the Hoboken goal (§5). The paper's proposal
follows [Slo70]: a teleological modal takes a designated goal, which every world
quantified over achieves, while the remaining goals are ancillary considerations that rank
those worlds; have to quantifies over all goal-achieving accessible worlds and ought to
over the best of them (24), so have to entails ought to and not conversely (haveTo,
oughtTo). The designated goal overrides the actual goals, which dissolves the Hoboken and
mayor problems (25), (26) and predicts the have to and ought to verdicts of Huitink's van
Nistelrooy scenario (27). Necessary conditions come out trivially true under have to (34),
and [Nis05]'s to go to Harlem, you ought to kiss Pedro Martinez (36) comes out true,
against which the paper conjectures that the prejacent must be an essential part of a way of
achieving the goal (42), IsEssentialPart.
Implementation notes #
The obvious analysis is Conditional.Restrictor.conditionalNecessity, and the designated-goal
modals are Kratzer necessity over the base restricted to the goal, with the ancillary ordering
source for ought to and the empty one for have to. The scenarios are finite world types
with the relevant propositions as predicates and the refutations decided. The condition (42)
quantifies over arbitrary propositions, and so holds of any prejacent as soon as some accessible
world lacks both the prejacent and the goal (isEssentialPart_of_exists); the paper leaves the
status of the condition open. The paper's examples are the rows of
Data.Examples.VonFintelIatridou2005.
References #
The obvious analysis and the Hoboken problem (§3) #
In the Hoboken scenario you take the A train or the PATH train, and hypothetically want to go to Harlem or not; the A train reaches Harlem, the PATH train Hoboken.
- wantA : Hoboken
Wants Harlem, takes the A train.
- wantPath : Hoboken
Wants Harlem, takes the PATH train.
- otherA : Hoboken
Does not want Harlem, takes the A train.
- otherPath : Hoboken
Does not want Harlem, takes the PATH train.
Instances For
Equations
- VonFintelIatridou2005.instDecidableEqHoboken x✝ y✝ = if h : x✝.ctorIdx = y✝.ctorIdx then isTrue ⋯ else isFalse ⋯
Equations
- One or more equations did not get rendered due to their size.
Equations
Instances For
Equations
Instances For
Equations
Instances For
Instances For
The actual goal in every world is to go to Hoboken.
Equations
Instances For
On (11), the obvious analysis, the if-clause restricting the circumstantial base of a modal ordered by the actual goals, makes the Harlem sentence false, since the best want-Harlem world takes the PATH train.
Sæbø's analysis and conflicting goals (§4) #
The conflicting-goals scenario (13): this afternoon you take the A train to Harlem or the PATH train to Hoboken, not both.
Instances For
Equations
- VonFintelIatridou2005.instDecidableEqConflict x✝ y✝ = if h : x✝.ctorIdx = y✝.ctorIdx then isTrue ⋯ else isFalse ⋯
Equations
- One or more equations did not get rendered due to their size.
Instances For
Equations
- x✝.goHoboken = (x✝ = VonFintelIatridou2005.Conflict.pathTrain)
Instances For
Sæbø's ordering source (14) is the actual goal of going to Hoboken with the hypothetical goal of going to Harlem added.
Equations
Instances For
Sæbø's analysis makes the Harlem sentence false, since with the two goals inconsistent the best worlds achieve either, and not all take the A train.
Nested modality (§5) #
In the scenario of §5 the closest world in which you want to go to Harlem is one in which you still want to go to Hoboken as well.
- actual : Nested
The actual world: wants Hoboken, takes the PATH train.
- bothPath : Nested
Wants both, takes the PATH train.
- bothA : Nested
Wants both, takes the A train.
Instances For
Equations
- VonFintelIatridou2005.instDecidableEqNested x✝ y✝ = if h : x✝.ctorIdx = y✝.ctorIdx then isTrue ⋯ else isFalse ⋯
Equations
- One or more equations did not get rendered due to their size.
Equations
Instances For
Equations
Instances For
Instances For
The goals of each world are Hoboken in the actual world and both in the others.
Equations
Instances For
The closeness ordering of the higher modal is agreement with the actual world on wanting Hoboken.
Equations
Instances For
On (16), the nested analysis, a closeness modal restricted by the if-clause over the teleological modal, makes the Harlem sentence false, since in the closest want-Harlem world the Hoboken goal survives and the PATH train is among the best worlds.
Designated goals (§6) #
To p, have to q (24b) holds when every accessible world achieving the designated goal p
is a q-world.
Equations
- VonFintelIatridou2005.haveTo f p q w = Conditional.Restrictor.conditionalNecessity f ⊥ p q w
Instances For
To p, ought to q (24a) holds when every accessible world achieving the designated goal p
that is best by the ancillary considerations g is a q-world.
Equations
- VonFintelIatridou2005.oughtTo f g p q w = Conditional.Restrictor.conditionalNecessity f g p q w
Instances For
Sloman's insight is that have to entails ought to whatever the ancillary considerations.
A necessary condition of the goal is trivially something you have to do (34).
In Sloman's London example (23) the train is the best means without being the only one, so
you ought to take the train, but you don't have to. World true goes by train, false by
bus; both arrive by noon, and only the train is comfortable.
In Huitink's van Nistelrooy scenario (27) both the A and the C train reach Harlem, and Ruud rides the A train.
Instances For
Equations
- VonFintelIatridou2005.instDecidableEqRuud x✝ y✝ = if h : x✝.ctorIdx = y✝.ctorIdx then isTrue ⋯ else isFalse ⋯
Equations
- One or more equations did not get rendered due to their size.
Instances For
Kissing Pedro Martinez (§7.2) #
Equations
- VonFintelIatridou2005.instDecidableEqPedro x✝ y✝ = if h : x✝.ctorIdx = y✝.ctorIdx then isTrue ⋯ else isFalse ⋯
Equations
- One or more equations did not get rendered due to their size.
Equations
Instances For
Equations
- x✝.kissPedro = (x✝ = VonFintelIatridou2005.Pedro.cTrainKiss)
Instances For
By (42), q is an essential part of a way of achieving p when some premises together with
q entail p over the modal base while without q they do not.
Equations
- VonFintelIatridou2005.IsEssentialPart f p q w = ∃ (P : Set (W → Prop)), sInf (insert q (f w ∪ P)) ≤ p ∧ ¬sInf (f w ∪ P) ≤ p
Instances For
Over arbitrary premises the condition (42) is undiscriminating, since any prejacent is an essential part as soon as some accessible world lacks both the prejacent and the goal, the premise that either the goal holds or the prejacent fails doing the work.
In Nissenbaum's scenario, staying home neither kisses Pedro nor reaches Harlem, so kissing Pedro counts as essential by (42).