Laboratoire Cognition, Langues, Langage, Ergonomie (CLLE)
UMR 5263

Accueil > Agenda

Knowledge ascriptions, fiction, and the historical present

le 8 novembre 2018
14h - 16h

Giuseppe Spolaore (Università di Padova)- Séminaire ERSS

Moore’s paradox is usually presented in connection with statements like:

(1)  It’s raining, but I don’t believe/know that it’s raining.

What is special about these claims is that they sound absurd even though they are coherent and possibly true. Let us call them Moore-paradoxical sentences.
It is commonly held that Moore-paradoxical sentences are assertable in no context whatsoever. In the talk I show that this idea is actually false: it is easy to cook up linguistic contexts in which (1) is assertable. Moreover, I present and discuss a problem involving self-knowledge reports about fiction.  I argue that this problem is to be dealt with in analogy with a natural treatment of the historical present, that is, by exploiting a double-context semantics.
Tenses play a key role in Moore’s paradox. E.g., the past-tensed version of (1):

(1t)  It was raining, but I didnt believe/know that it was raining,

is perfectly sensible.  Moreover, as a matter of fact, the standard examples of Moore-paradoxical sentences all involve present-tensed sentences. However, it is well known that there exist present- tensed discourses that actually concern past events or states. These include the narrations in the (so-called) historical present. Here is an example, which I borrow from [4]:

(2) Fifty  eight years ago to this day, on January 22, 1944, just as the Americans are about to invade Europe, the Germans attack Vercors.

Clearly, this present-tensed statement conveys essentially the same content as “Fifty eight years ago to this day, on January 22, 1944, just as the Americans were about to invade Europe, the Germans attacked Vercors.”  In light of this, it is hardly surprising that Moore-paradoxical sentences can be perfectly assertible when embedded within  narrations in the historical present. Here is one example.

 (3) It’s July 23, 2012. I’m locked up in an underground dark room, with no holes or windows. It’s raining, but I don’t know that it’s raining.

Suppose that, during a debate about Milos Forman’s movie Amadeus, I assert:

(4) Mozart’s Requiem was commissioned by Salieri.

Claims like (4), which are understood to be true in a work of fiction, are called fictive statements. It is worth noting that  also fictive statements can give raise to ‘counter-examples’ to Moore’s paradox. E.g., I can use statement (1) during a debate about narration (3), assuming (3) is a piece of fiction about me. According to a natural approach, fictive sentences call for a context shift: the world of the context is not to be identified  with  the world of use but with  ‘the world of the story’ (see [3]).  There are different ways to make  sense of the notion of world of a story.  In accordance with a recent proposal [omissis ], I shall equate a world of a story with a fictional setting, defined  as a function from possible worlds to propositions – intuitively,  the propositions that are true in the story relative to that world.  In any event, if the context-shift view is adopted, then (4) turns out to be true simpliciter, for it is clearly true relative to the relevant ‘world of the story’.  It is thus far from mysterious that I can feel entitled to assert it.
There is a problem, though. Moorean considerations suggest that, if (4) is assertable by me in a context, then also the following is:

 (5) I know that Mozart’s Requiem was commissioned by Salieri.

 Unfortunately, it is very dubious that the word-shift approach can be extended to (5).  Firstly, because there is no guarantee that I exist in the world of Amadeus.  Secondly, because (5) intuitively entails that I know something in the actual world and not just in some fictional world.  These problems motivate the quest for alternative treatments. In the talk,  I briefly  discuss another natural  approach to fictive  statements and to sentences like  (5),  which appeals to a covert “in  fiction”  operator (see [2]).

In this section,  I argue that there is a close analogy between knowledge self-ascriptions  like (5) and certain knowledge ascriptions involving the historical present. My proposal is to extend to the former ones a natural semantic approach to the latter.
There are good reasons to think  that  a Kaplan-style semantic analysis of narrations in the historical present requires two contexts, one representing the (typically present) situation in which the story is told,  the other representing the (typically  past) situation  the narration is about; Schlenker [4] calls them context  of thought and context  of utterance,  respectively. E.g., in the case of (2), the context of thought ct  involves (possibly among other things) the actual world and the present time (in 2018), while the context of utterance cu involves the actual world and a certain non-present time (in 1944). Indexical expressions such as “now”  and “this  day” tend to be sensitive to the context of thought, whereas the historical present tends to be sensitive to the context of utterance. Thus, e.g., in (2) “this  day” picks up the day of the time of ct  (in 2018), while the present tense is relative to the time of cu (in 1944). Now let us consider:

 (6)  It’s August 1914. Many German soldiers believe they are sent to a blitzkrieg.  But now we know that they are wrong, and that a long, long war is forthcoming.

 The italicized knowledge ascription at the end of (6) (in its more natural understanding) concerns our present (in 2018) epistemic situation. However, interestingly, the historical present tense in the complement sentence is relative to a past time (in 1914). Within  a double-context semantics, it is possible to provide adequate truth-conditions for such intricate statement. Assuming a Hintikka- style approach to knowledge ascriptions, we can say (very roughly) that  the statement  is true relative  to (ct, cu) iff,  in all the worlds compatible with  what we know at the time of ct,  the embedded sentence is true at the time of cu.
In the talk, I argue that nonfictive knowledge ascriptions that, like (5), involve fictive statements in the complement are to be dealt with by appeal to a double context, in analogy with the above treatment of (6). This time, we can distinguish between a real context cr  and a fictional context cf . Simply put, sentence (5) is true in (cr , cf ) iff, in all the worlds compatible with what I know in the world of cr  (viz., in the actual world), the embedded sentence is true in the fictional setting of cf  (viz. in the ‘world of the story’). This solution requires that one same story can have different contents relative to different epistemic alternatives. I take  this requirement  to be very plausible within  a Hintikka-style epistemic logic, as it just  boils down to the platitude  that people can have partial  knowledge of the content of a story, or even no knowledge at all. This solution can be extended to other problems involving fictive statements and factive constructions,  such as the ones discussed in [1, § 3.1].


[1] Everett, A. (2013). The Nonexistent. Oxford University Press, Oxford.
[2] Lewis, D. (1978). Truth  in fiction. American Philosophical Quarterly, 15(1):37–46.
[3] Predelli, S. (1997). Talk about fiction. Erkenntnis, 46(1):69–77.
[4] Schlenker, P. (2004). Context of thought and context of utterance: A note on free indirect discourse and the historical present. Mind & Language, 19(3):279–304.



Lieu(x) :
Salle E412


Version PDF | Mentions légales | Conseils d'utilisation | Lien vers RSSSuivre les actualités | haut de la page