Tamina Stephenson (MIT) [email protected] Oxford, March 26, 2009

Imagining Contradictions Tamina Stephenson MIT [email protected]

1. Introduction 1.1. A contrast: Epistemic modals vs. know Yalcin (2007) compares standard Moore-paradoxical sentences to similar examples with epistemic modals: (1)

# It’s raining but I don’t know that it’s raining.

(2)

# It isn’t raining but it might be raining.

[Moore’s paradox] [Yalcin’s cases]

Both are unassertable and contradictory-sounding BUT standard Moore-paradoxical sentences become acceptable when embedded under predicates like imagine (3)

Imagine that it’s raining but you don’t know that it’s raining.

(4)

# Imagine that it isn’t raining but it might be raining.

(The pronoun is changed to you to match the imperative in the embedded cases.) 1.2. Outline •

[Sec. 2] Introduce relativist system where propositions = sets of centered worlds

•

[Sec. 3] Generalization: Moore-paradoxical type sentences become acceptable under imagine only if they express non-subjective propositions [do not depend on the “center”]

•

Proposal: There are two uses of imagine; the use that can “rescue” Moore-paradoxical sentences is only compatible with non-subjective propositions.

2. Background: Relativist Semantics Standard view: • proposition = set of possible worlds [sentences true / false at possible worlds] Relativist view: • proposition = set of world-individual pairs (centered worlds) [sentences true / false at world-individual pair] (Lasersohn, 2005, Stephenson, 2007) Note: I’ll sometimes call the individual center of a world the “judge”

Tamina Stephenson

March 2009

2.1. Relativist semantics for epistemic modals (5)

(a)

might p is true at iff p is compatible with x’s knowledge in w

(b)

must p is true at iff not p is not compatible with x’s knowledge in w (Stephenson, 2005, Stephenson, 2007)

Motivated by two main phenomena… 2.2. Phenomenon 1: Subjective disagreement (6)

Sam: Sue:

Bill might be in his office. No, he can’t be!

[≈ not compatible with Sue’s knowledge that Bill is in his office] Compare to: (7)

Sam: Sue:

#

I don’t know that Bill isn’t in his office. Yes, you do! / Nuh-uh / No he isn’t! I know that he’s at home!

2.3. Phenomenon 2: Linking in attitude reports (8)

Sam thinks that the cook might be the murderer. ≈ it’s compatible with Sam’s knowledge/beliefs that the cook is the murderer

With a quantified subject: [Every boy]i thinks hei might be stupid.

(9)

≈ for every boy x, it’s compatible with x’s beliefs/knowledge that x is stupid

3. The Generalization Predicted meanings for Moore-paradoxical cases (as set of centered worlds): (10)

# It’s raining but I don’t know that it’s raining. [=(1)] { : it’s raining in w and the speaker doesn’t know in w that it’s raining }

(11)

# It isn’t raining but it might be raining. [=(2)] { : it isn’t raining in w and it’s compatible with x’s knowledge in w that it’s raining }

Notice: • (10) does not depend on the individual center let’s call this a non-subjective proposition (= non-judge-dependent, non-center-dependent) •

2

(11) depends on the individual center let’s call this a subjective proposition (= judge-dependent, center-dependent)

Oxford

Recall: (1) becomes acceptable under imagine but (2) does not: (12)

Imagine that it’s raining but you don’t know that it’s raining.

(13)

[=(3)]

# Imagine that it isn’t raining but it might be raining.

[=(4)]

3.1. Hypothesis Empirical Generalization: • Non-subjective propositions become acceptable under imagine • Subjective propositions remain unacceptable 3.2. Further evidence Sentences with taste predicates such as tasty / fun are also subjective (Lasersohn, 2005 and others) (14) (a) (b)

y is tasty is true at y tastes good to x in w y is fun is true at iff y is fun for x in w (Lasersohn, 2005)

Moore-paradoxical type case seems to pattern with Yalcin’s cases: (15)

?/# The cake is tasty but it doesn’t taste good to me. { : the cake tastes good to x but doesn’t taste good to the speaker> Æ Subjective proposition

(16)

?/# Imagine that the cake is tasty but it doesn’t taste good to you.1

3.3. Side note: Two readings for taste predicates Taste predicates can be linked to a salient third-person referent: (17)

[Context: Sam is watching his cat eat a new brand of cat food. After a few hesitant bites the cat devours it eagerly. Sam says:] Oh good, the new cat food is tasty. [tasty = tastes good to the cat]

Along with this, linked readings in attitude reports are not obligatory: (18)

Mary: Sam:

How’s that new brand of cat food you bought? I think it’s tasty, because the cat has eaten a lot of it.

[tasty = tastes good to the cat] 1

I’ve put “?/#” instead of “#” because these might both be okay if tasty is interpreted truly generically – as something like ‘tastes good to most people.’ Speakers seem to vary in how much they accept this kind of reading for predicate of personal taste (a fact which ought to be investigated further). The crucial point here is that (15) and (16) are equally acceptable on this reading.

3

Tamina Stephenson

March 2009

I have argued elsewhere that this is a separate use Key evidence: (19) (a) (b)

A: The cat food is tasty. B:(#/OK) No, it’s isn’t!

•

Scenario 1: A tastes the cat food and goes “Yum!” B tastes the cat food and goes “Yuck!” Æ OK

•

Scenario 2: A tastes the cat food and goes “Yum!” B watches the cat turn up its nose at it. Æ # (odd; or misunderstanding) Scenario 3: A watches the cat gobble down the cat food; B tastes it and goes “Yuck!” Æ # (odd; or misunderstanding) Scenario 4: A watches the cat gobble down the cat food; B sees the cat turn up its nose. [Or B knows that the cat is just starving.] Æ OK

• •

Note: It’s not generally possible to link epistemic modals to a salient referent: (20)

[Context: Sam is watching his dog eat dog food. The dog is eating the food with enthusiasm that he usually saves for table scraps. Sam says:] # Oh good, that might be table scraps. [might ≠ compatible with the dog’s knowledge – i.e., (20) is infelicitous if Sam knows what he fed the dog]

4. Proposal 4.1. Outline •

Two senses of imagine: “subjective” and “objective”

•

Subjective propositions are only compatible with subjective imagine

•

Moore-paradoxical propositions (broadly construed) become acceptable only under objective imagine; under subjective imagine they yield (near-)contradictions

•

Result: Non-subjective Moore-paradoxical cases can be rescued by embedding under objective imagine, but subjective Moore-paradoxical cases cannot

4.2. Two Ways of Imagining Imagining “from the outside” vs. “from the inside”2 (21)

I imagined skiing down the hill.

(Ninan, 2007, no. 2a)

Typically: speaker imagines the subjective experience of moving down the hill.

2

4

See recent work of Ninan (2007) and Recanati (2007), and work cited there.

Oxford

(22)

I imagined that I was skiing down the hill.

(Ninan, 2007, no. 3a)

Scenario 1: as in (21) Scenario 2: speaker imagines something like what they would look like skiing down the hill (in a movie, e.g.) The terminology I will use: •

Subjective imagine / imagineSUBJ = imagining “from the inside”

•

Objective imagine / imagineOBJ = imagining “from the outside”

One reason (from a linguistic perspective) to find this distinction plausible: the grammatical category of the argument of imagine seems to make a difference: (23)

I imagined a cat. [Speaker built up mental imagine of a cat → objective / “from the outside”]

(24)

Imagine climbing up a tree. [Speaker thought of the experience of climbing → subjective / “from the inside”]

When the argument is a regular sentence with a finite verb, it can go either way: (25)

I imagined that I was climbing up a tree. [(i) Speaker built up mental image of themselves climbing → objective] [(ii) Speaker thought about the experience of climbing → subjective]

4.3. Objective imagine The basic idea: objective imagine is fundamentally a relation between the imaginer and a thing For example: (26)

I imagined a cat. = A certain relation holds between me and a particular possible cat [Note: “possible cat” = possible individual (i.e., individual in some possible world) that satisfies the property of being a cat.]

(27)

I imaginedOBJ that I was skiing down the hill. = A certain relation holds between me and a particular possible situation3 s (where s is a situation in which I am skiing down the hill)

1st try: (28)

x imaginesOBJ p is true iff x has a mental image of some situation where p is true

3

I’m assuming something similar to Kratzer’s notion of situations as parts of possible worlds (Kratzer, 1989, Kratzer, 2007). I am not adopting a full-scale situation semantics here, but I assume this would be compatible.

5

Tamina Stephenson

March 2009

Problem: on my view propositions are not true in “situations” but rather something like “centered situations” [In other words, we need to convert propositions from type to type ] There’s an easy and natural way to do this with non-subjective propositions. One simple option: (29)

Objectivized Propositions: For a proposition p, pOBJ = is defined iff ∀s,x,y, p has the same truth value at and If defined, pOBJ = { s : for all individuals x, p is true at }

2nd try: (30)

x imaginesOBJ p is defined iff pOBJ is defined If defined, x imaginesOBJ p is true iff x has a mental image of some situation in the set pOBJ

Consequence: objective imagine is only possible with non-subjective propositions 4.4. Subjective imagine The basic idea [2 parts]: •

Subjective imagining is a relation between an individual and a proposition [understood as a set of world-individual or situation-individual pairs]

•

Subjectively imagining p = putting oneself hypothetically in the position of having a certain kind of perceptual evidence for p [I’ll return to the kind of evidence below]

[“Oneself” is crucial here – subjective imagining is essentially a type of de se attitude] For example: (31)

I imaginedSUBJ that it was raining. = the speaker put themselves hypothetically in the position of having direct perceptual evidence for rain – (e.g., the feeling of water on my head, the sound of raindrops on the roof, …)

(32)

I imaginedSUBJ that the cake was tasty. = the speaker put themselves hypothetically in the position of having direct perceptual evidence that the cake was tasty – (e.g., having particular experiences associated with activation of reward centers in the brain)

6

Oxford

In general: (33)

x imaginesSUBJ p is true iff x imagines having direct perceptual evidence for p

4.5. Explaining the Contrast Recall: (34)

Imagine that it’s raining but you don’t know that it’s raining.

Compare to: (35) (36)

# Imagine that it isn’t raining but it might be raining. ?? Imagine that the cake is tasty but it doesn’t taste good to you.

Explanation: Traditional Moore-paradoxical case: •

(34) with objective imagine: requires addressee to build up image of a situation where it’s raining and the addressee (that particular individual) doesn’t know it → OK

•

(34) with subjective imagine: requires addressee to think about the experience of having direct perceptual evidence that it’s raining but also having evidence that they themselves don’t know this → (nearly) impossible to have this kind of evidence for both things simultaneously

→ Result: (34) OK (though restricted to objective interpretation of imagine) Epistemic modal case: •

(35) with objective imagine: not possible because the embedded proposition is subjective

•

(35) with subjective imagine: requires addressee to think about the experience of having direct perceptual evidence that it isn’t raining while also having evidence that it might be raining → (nearly) impossible to have this kind of evidence for both things simultaneously

→ Result: (35) is impossible

7

Tamina Stephenson

March 2009

Taste predicate case: •

(36) with objective imagine: not possible because the embedded proposition is judgedependent

•

(36) with subjective imagine: requires addressee to think about the experience of having direct perceptual evidence that the cake is tasty while also having evidence that it doesn’t taste good to them themselves → (nearly) impossible to have this kind of evidence for both things

→ Result: (36) is impossible. 4.6. Some remaining puzzles (37)

Imagine a tasty cake.

Can it make sense to talk about building up a mental image (“from the outside”) of a tasty cake? (38)

Imagine being betrayed by your best friend and not knowing it.

[I. Heim, p.c.]

Are you imagining a larger stretch of time (where you don’t know it and then you do)?

References Kratzer, Angelika. 1989. An Investigation of the Lumps of Thought. Linguistics and Philosophy 2:607-653. Kratzer, Angelika. 2007. Situations in Natural Language Semantics. In The Stanford Encyclopedia of Philosophy (Spring 2007 Edition), ed. Edward N. Zalta. http://plato.stanford.edu/archives/spr2007/entries/situations-semantics Lasersohn, Peter. 2005. Context Dependence, Disagreement, and Predicates of Personal Taste. Linguistics and Philosophy 28:643-686. Ninan, Dilip. 2007. Imagination, Inside and Out. Manuscript, MIT. Recanati, François. 2007. Imagining de se. Paper presented at Mimesis, Metaphysics, and Make-Believe: A Conference in Honour of Kendall Walton, University of Leeds, Leeds, U.K. Stephenson, Tamina. 2005. Assessor Sensitivity: Epistemic Modals and Predicates of Personal Taste. In New Work on Modality, eds. Jon Gajewski, Valentine Hacquard, Bernard Nickel and Seth Yalcin, 179-206. Cambridge, Massachusetts: MITWPL. Stephenson, Tamina. 2007. Judge Dependence, Epistemic Modals, and Predicates of Personal Taste. Linguistics and Philosophy 30:487-525. Yalcin, Seth. 2007. Epistemic Modals. Mind 116:983-1026.

8