Chapter 1: Introduction
2. Against the Knowledge Requirement
2.3.1. Case Three
COMPETING INTENTIONS is less far fetched than MAD SCIENTIST. However, it is still rather complex. Thus, I imagine that the dedicated proponent of the knowledge condition might well remain unconvinced24. Additionally, the question of the audience's knowledge of what is said is of
24 Another strategy my opponent could take would be to question whether the knowledge gained in this case is
genuinely testimonial, since it depends on so many outside factors (thanks to Brian Weatherson for this point). Here is how I think we should conceive of testimonial knowledge: testimonial knowledge is a testimonial belief which counts as knowledge. A testimonial belief is a belief formed on the basis of testimony, which means something like the following: Someone has said that p, or made an utterance with the content p, or intended to say that p and had their intention recognized, or said something sufficiently similar to p and the audience reasonably recovered p etc.
interest due to worries raised in the philosophy of language about the successful recovery of the speaker's intended meaning. Yet, neither of the cases so outlined so far bears any similarity to the problematic cases discussed in the philosophy of language. Thus it is worth considering a final case which bears directly on problems of context-sensitivity25.
The reader will recall the problem from section one. The problem is that in cases of context- sensitivity there will often be several similar values which can be assigned to a context-sensitive term, each determining a different proposition26. The audience's evidence will often be insufficient
to grant them knowledge that the speaker intended one of these propositions rather than another. However, as the final case indicates, these epistemic difficulties do not always rule out testimonial knowledge
As noted in the introduction, Wettstein (1981) provides the following argument against the Russellean view of definite descriptions: in most attributive uses of a definite description the explicit content underdetermines the referent. That is, it is not sufficient to determine a unique referent. For example, take the sentence 'the statue is broken' - the world contains many statues, thus 'the statue' is not enough by itself to select a unique statue. Those words could be used in different contexts to identify very different statues. The description must be supplemented in order to designate a unique statue. However, there will be many potential supplementations which would uniquely identify the intended statue. For example the description might be supplemented as 'the statue belonging to Alex', or 'the statue of a frog'. Wettstein argues that this will make the proposition expressed indeterminate on the Russellean view. He takes this to be an unacceptable consequence and provides an alternative view of definite descriptions on which the meaning is determinate. However, the problem seems to arise for a wide array of context sensitive terms, not just descriptions. Similar cases can be given involving quantifier domain restriction, non-sentential assertion, and a whole host of other context sensitive terms. Thus Wettstein's response seems unsatisfactory, for it is unlikely that a unified account can be given which secures determinacy for
and on the basis of this fact the audience came to believe p. On this conception of testimonial knowledge Matt's knowledge is genuinely testimonial.
25 The cases discussed so far have also been cases of context sensitivity. However, they are cases of reference
assignment which seems less epistemically problematic than many other forms of context sensitivity. In cases of reference assignment there are usually fewer candidate meanings for the audience to choose between. In the cases which are generally seen to be problematic there are a large number of similar meanings for the audience to choose between. It is seen as mysterious how the audience could select the correct meaning from amongst the many available candidates.
26 This will not generally be the case when context merely supplies a referent (as is the case with, for example,
demonstratives or referential uses of definite descriptions). However, it will be the case when context must supply a more complex value such as a property, comparison class, modal base, ordering relation on worlds, standard of taste etc, and in cases of loose talk or non-sentential assertion.
a feature of context sensitive terms. This can lead to problems concernig knowledge of what is said however. Consider the following case (adapted from Donnellan (1966)):
MURDERER: Sally and Matt are investigating a chain of murders they know to be related. Sally is standing over the mangled remains of Frank - the latest victim. Frank has clearly been murdered in a rather brutal way, and the scene has been covered in insane etchings and other clear indicators that the murderer is insane. Additionally, both Sally and Matt know that Frank was murdered by the murderer they are investigating (that is, the individual who committed the previous murders). Sally phones Matt and says 'the murderer is clearly insane'. The description used here is incomplete, and could be rendered complete in a number of different ways. For example, Sally could have intended the proposition 'the murderer (of Frank) is clearly insane', or 'the murderer (who we are investigating) is clearly insane'. Matt recovers the proposition 'the murderer (who we are investigating) is clearly insane' (which happened to be the proposition intended by Sally), and comes to believe it on the basis of Sally's testimony.
Once again I think that it is clear that Matt gains testimonial knowledge in this case. The belief he forms is true, safe, sensitive, and caused by its truth maker. Additionally, Sally undertook all the usual commitments to the proposition that Matt comes to believe. However, since as far as Matt can tell Sally might have intended the other (very similar) proposition, Matt does not know that Sally intended the proposition he recovered.