In
Bamboozled by our own words, Keith DeRose gives a defene
of epistemologial
ontextualism, the position that the semanti ontent of the word `know' varies with the
ontext in whih is uttered. This defene is made by attempting to dismiss four dierent
objetions to ontextualism. In this paper I will present and evaluate DeRose's defene.
The theory of epistemologial ontextualism interests philosophers of language
beause, if proven to be true, it also proves the position known as linguisti
ontextualism. This position laims that the semanti ontent of many words (that is:
more words than just the basi indexials) vary with the ontext in whih they are
uttered. If the word `know' does, this would prove linguisti ontextualism to be true.
In setion I, I present the rst two objetions to ontextualism DeRose outlines,
together with his answers to them. In setion II, I present the objetion from so-alled
belief reports, and argue that DeRose's reply fails. In setion III I will look at the
onept of single soreboard semantis, and argue that the onept is implausible.
I will not disuss the fourth objetion, Shier's objetion to ontextualist solutions
to skeptiism. I believe that this objetion does not onstitute an attak on
ontextualism as suh, like the other objetions do. Although an evaluation of DeRose's
defene ertainly would be interesting, it falls outside the sope of this artile.
I
The objetions from truth-value onfusion
The rst two objetions stem from our intuitions of what is true and false onerning
the word `know'. DeRose resolve them in muh the same way. There are important
dierenes between the reasons why DeRose dismiss the objetions, but they share one
ommon feature: they are more sophistiated than mere appeal to semanti blindness.
Semanti blindness is the position that language-users are unaware of some or all of the
semanti ontent they are expressing.
In both of the objetions from truth-value dierenes, we meet two subjets (SH
and
SL ),
who are in two dierent ontexts.
SH
is in
H,
where the epistemi standards
are high. Suh high standards mean that the amount of evidene needed to `know' some
proposition is large.
SL
is in
L,
where it is easier to laim knowledge about a
proposition. Then both utter a sentene:
(1)
SL :
(2)
SH : SL
I know that
p
does not know that
p
The rst objetion is that from judgements of omparative ontent. There seems to
be a ontradition between (1) and (2). But ontextualism is (naïvely) understood to
1
laim that
and
SL
SH 's
`know' has a dierent semanti ontent than
SL 's.
But then both
SH
ould be right in saying what they are saying, and the ontradition disappears.
This is a problem to ontextualism beause it is unintuitive. It seems to suggest that
language users who see a ontradition here do not know that the word `know' is
ontext-sensitivethey are blind to the semantis of the words they use. (DeRose
2005a, p. 2)
DeRose is eager to avoid semanti blindness, and begins by noting that the
ontextualist an maintain that there is a ontradition. In DeRose's ontextualism,
there is a single (or ommon) soreboard of semantis for all partiipants in a
disussion. This means that in a disussion, `know' an only have one semanti ontent.
The laims
do
ontradit eah other, and the ontextualist is spared from the
embarrassment of semanti blindness. But who is right, and who is wrong? It seems to
depend on how the single soreboard works. I will return to that in setion III, but I
note here that DeRose laims that neither of the laims are true. This seems to all for
a new defene against another unintuitive result.
But there is no suh ontradition, laims DeRose, unless they are atively engaging
eah other in an argument. If they are not, they are making ompatible laims beause
the semanti ontent of `know' diers between the two ontexts, and they have their
own soreboard. And in this ase, our intuition that they are ontraditing eah other is
weaker. (DeRose 2005a, p. 4)
So, when
SL
and
SH
are in truly dierent ontexts and onversations, and assert
some surfae-ontraditory laims about knowledge (like (1) and (2)), they are not
ontraditing eah other. DeRose is very areful to stress that the standards of
H
are
not neessarily those of philosophial epistemology, but heightened due to some
pratial reason, for instane a polie investigation. Nevertheless, when the standards
are signiantly dierent, and
SL
will laim to know a lot more than
SH
will. In this
situation, they are both orret, and aording to DeRose our intuition of a
ontradition is weaker than the objetion presumes.
The latter part of DeRose's argument is as onvining as the original objetion was.
They both appeal to our intuition, and I think DeRose manages to show that our
intuition is not as lear-ut as the objetion states. The rst part of the argument, on
the other hand, rests entirely on the theory of single soreboard semantis. I will return
to this in setion III.
The seond objetion is that from metalinguisti laims. We are still using our
previous example with utteranes (1) and (2). The objetion is then based on the
2
observation that our subjets would be willing to state the following:
(3)
SL :
(2) is false
(4)
SH :
(1) is false
This is a problem for ontextualism, beause aording to ontextualism, both (3) and
(4) are wrong, and it seems unlikely that language works in a way that makes people
utter falsehoods routinely. (DeRose 2005a, p. 8)
As in the ase with the omparative judgement, DeRose stresses that the intuition
of a problem only seems to arise when onsidering the wrong ases; the ases where
SL
and
SH
are arguing with eah other. Therefore, we primarily look at those ases.
Here we need not introdue single soreboard semantis: Contextualism says exatly
that
SL
is allowed to assert his laim (1) without expressing a falsehood, beause `know'
is ontext-sensitive. But yet the intuition is strong that
ontextualism,
SH
SH
ould say (4). Given
would be wrong when asserting (4), yet it seems to be likely that
SH
would say it. Does this mean that people in general do not know what they say, and
an we make sense of a language where rules of truth exist independently of usage?
The rst two objetions share muh of the same essential problem. With a ertain
interpretation of ontextualism (a naïve one, maybe), the rules of the language dier
demonstrably from the rules the language-users seem to subsribe to. DeRose attempts
to show that this is not true in every ase, but he admits that the ase of disagreement
in a onversation has this problem. On the other hand, invariantism leads to the same
problems the other way around.
DeRose's defene is oensive on this objetion. He shows how also invariantism
must admit some `semanti blindness' in other ases involving metalinguisti laims,
and that language-users' intuitions are unlear on matters involving this kind of
metalinguisti laims. Therefore, this is not a problem for ontextualism alone. There is
something else than ontextualism or invariantism that auses this onfusion. I believe
DeRose's argument is sound on this matter, unless someone an provide an
epistemologial theory whih does not share these problems.
II
Belief reports
The third objetion is the objetion from belief reports. The objetion states that if a
ertain shema of disquotation is orret, it is a problem for ontextualism beause it
shows how the verb `know' an be used aross dierent ontexts, and therefore aross
dierent epistemologial standards.
3
The partiular shema is from John Hawthorne (Hawthorne 2004, p. 101), and is
known as the disquotational shema of `know' (DSK, for short). DeRose has modied
the formulation of the shema slightly, and the modiation makes its appliation
slightly simpler to use in our disussion. DeRose's shema is formulated like this:
If a speaker
S1
sinerely utters a sentene of the form `A knows that
sentene in the that-lause means that
refers to
a,
then a speaker
to mean that
p,
S2 ,
p,
and the
and `A' is a name or an indexial that
also using `A' to refer to
an truthfully state that
p',
1
S1
a
and using the that-lause
believes that
A
knows that
p.
(DeRose 2005a, p. 10)
Although DeRose's modied formulation of the priniple is meant to make the
onsequent meta-linguisti, he fails to do so orretly. The last part of the last period
should read . . . , an truthfully state that `S1 believes that
The problem to ontextualism appears when
S1
and
S2
A
knows that
p'.
are in dierent ontexts,
and those ontexts assign dierent semanti values to `know'. If ontextualism is
orret, there are ases where
that
A
knows that
p,
S2
would not be truthful when stating that
even after hearing
following: S1 said `A knows that
p '.
S1
utter `A knows that
p'. S2
S1
believes
an only state the
But if the DSK is orret, there are no suh
ases. Therefore, ontextualism must be wrong.
This objetion rests entirely on the orretness of the disquotational shema, and
DeRose presents an argument for its inorretness. His argument is an example where
we apparently would not disquote in the way the shema allows us to.
2
low) and a polie interview (high).
The example has two ontexts, a bar (
A
person (S2 ) overhears another person (S1 ) making some laim in the bar. The laim has
the form of
S1
knows that
Immediately afterwards,
p,
S2
for instane I know that Jim was at the oe today.
is interviewed by the polie, who explain that a serious
rime has been ommitted and then ask whether
If the disquotational shema is orret,
that
S1
S2
S1
knows where Jim was today.
will always truthfully be able to report
believes that she knows that Jim was at the oe. But DeRose argues that the
low is very dierent from that in
high. For instane, DeRose says S1 would say, while in low, that she knows Jim was
amount of evidene needed to laim knowledge in
at the oe if the reeptionist told her and she has seen his hat hanging in the hall.
But
S2
wouldn't tell the polie that
S1
believes that she knows Jim was at the oe
1I
have modied the formatting in this itation by replaing underlined text with italiised and using
subsripts and italiised text for variables.
2 The word `would' is of great importane here. It seems too strit to say that we `ould not' disquote
(beause we an do so just to prove it). Yet `would' might be too sloppy. Another possibility is that we
would not be orret if we disquote in the way the shema allows us, but this seems to lead attention
away from the point DeRose is trying to make, beause he appeals to our intuition of what would be
`natural' for the ators in our story.
4
based on this evidene. Beause of this, the disquotational shema is inorret and does
not onstitute a real objetion to ontextualism. Furthermore, DeRose suggests that the
shema is used only when the epistemologial standards are appropriately similar.
Presumably it also applies when
But it isn't lear that
S1
S2
has a lower standard than
S1 ,
but not the other way.
atually would laim that she knows. At least my intuition
is unlear on the matter. In DeRose's example
S1
is arguing that she knows beause of
a $2 bet. But there is a dierene between saying that she knows and doing so
truthfully: she might seretly have her doubts, and only say that she knows to win the
bet. In fat, if she is asked how do you know that?, she ould easily answer something
like well, I don't atually know, but the reeptionist told me, or a shortut like the
reeptionist told me. If that is the ase, `know' works more like a rhetorial shortut
(and way to win the bet) than atually expressing any knowledge-relation in
has
artiially
S1 .
Low
low standards in this example, and DeRose's laim that suh a ontext
exists seems dubious. A properly designed example should let its ators express
essentially truthful statements about their mental states, and not position them in
plaes where lies and rhetorial triks are abundant.
A defender of the disquotational shema an simply point to the word `sinerely' in
the shema, and laim that DeRose has not shown it to be the ase that we would not
disquote if
S1
is sinere. By stressing this point, we are reduing the usability of the
shema. How an
S2
know whether
shema ould be to hange
sinere in saying so,
S1
S2 's
S1
is sinere? A potential amendment to the
potential statement to: To the degree that
believes that A knows that
p.
S1
was
Note that this amendment is not
due to some inonsisteny in the original shema, only that it requires the statement to
be sinere. When this annot be guaranteed,
seems highly likely that
S2
S2
must make a dislaimer. And then it
will tell the polie exatly what the disquotational shema
allows her: To the degree that
S1
Jim was in the oe. The reason
was sinere in saying so, she believes that she knows
S2
won't say anything about
example is due to the very fat that she (S2 ) doesn't know that
S1 's beliefs in DeRose's
S1 was sinere when she
talked at the pub. But that is something altogether dierent from not disquoting
beause of the dierene in epistemologial standard. Maybe
S2
suspets that
involved as well, and telling the story on the pub to give Jim an alibi?
know that
S1
S2
S1
was
might even
is known to tell small lies from time to time? I believe it is this kind of
unertainty about sinerity that makes our intuition unlear in the example, not the
dierene in ontext.
Could we onstrut an example of a similar kind, that avoids the problems of
insinerity? DeRose says that the best ases for this general priniple, DSK, are ones
where the ontextualist will think there is a marked dierene in standards (DeRose
5
2005a, p. 11). DeRose is also wary of using test ases where one of the ators is in a
ontext with very tough standards, like an epistemologial disussion. It makes the
relevant pro-ontextualist intuitions stronger and more stable if the elevated epistemi
standards in
high are tied to a pressing and very pratial onern. . . (DeRose 2005b,
p. 176). Therefore, we opt for Cohen's airport example. This is used by Cohen and
DeRose elsewhere.
3
Mary and John are at the L.A. airport ontemplating taking a ertain ight to
New York. They want to know whether the ight has a layover in Chiago. They
overhear someone ask a passenger Smith if he knows whether the ight stops in
Chiago. Smith looks at the ight itinerary he got from the travel agent and
responds, Yes I knowit does stop in Chiago. It turns out that Mary and John
have a very important business ontat they have to make at the Chiago airport.
Mary says, How reliable is that itinerary? It ould ontain a misprint. They ould
have hanged the shedule at the last minute. Mary and John agree that Smith
doesn't really
know
that the plane will stop in Chiago. They deide to hek with
the airline agent. (Cohen 1999, p. 58)
In this example, Smith is in a dierent ontext than Mary and John. Mary and
John are in
high, and Smith is in low.
To illustrate an example of the disquotational
shema, we need some further dialogue. Let's all the person who asks Smith the
question Keith. Keith is our
S1 .
Mary or John plays
S2 ,
and Smith is
A.
Would Mary
say that Keith believes that Smith knows that the ight stops in Chiago? I think the
answer is fairly lear: she would. This kind of example provides no attak on the
shema, and the objetion to ontextualism stands.
I have shown that DeRose's refutation of the belief report argument against
ontextualism is at best inonlusive. Nevertheless, there seems to be something strange
about the onept of disquotation. It seems to fore the epistemologial standards of
the disquotee onto the disquoter. I am not onvined that the DSK is orret, although
DeRose's objetion to it does not prove it wrong.
This belief report argument reets a general problem of ontextualism about
knowledge. Given ontextualism, it seems to be impossible for two people to have
knowledge about eah others knowledge without being in the same ontext. Even if
knows
p
both
X
in ontext
and
Y
C1 ,
know
and
p.
Y
knows
p
in ontext
C2 ,
it is not possible for
Z
X
to say that
There is apparently no way to ompare knowledge between
ontexts, and this is at odds with the way we normally use the onept of knowledge.
3 Although
DeRose has re-written it to a disussion on the opening hours of a bank. I nd this
peuliar, beause the intuition of not knowing in high is signiantly weaker in DeRose's example, due
to the relative trustworthiness of a bank.
6
III
Single soreboard semantis
If I am orret in asserting that the DeRose's defene of ontextualism rests on his
theory of single soreboard semantis, it is interesting to look at this premise as well. In
this setion, I argue that the onept of single soreboard semantis is dubious and far
from proven to be plausible. If this is the ase, DeRose's defene of ontextualism is
equally weak.
In
Single Soreboard Semantis
(DeRose 2004), DeRose presents his theory of
sore-keeping in a onversation. The metaphor of sore-keeping is adapted from Lewis
(Lewis 1979), but there seems to be important dierenes between DeRose's and
Lewis's soreboards. In an interesting note in
Single Soreboard Semantis, DeRose
seems to suggest that his soreboard exists outside of the minds of the speakers:
Here my use of the metaphor soreboard diers from Lewis's. For Lewis, there is
a soreboard in the head of eah of these partiipants, and what the sore is an
be a funtion in part of what all these dierent soreboards say the sore is. As I
use soreboard here, it by denition gives the right sore. (DeRose 2004, p. 19)
Even though DeRose delares his use of soreboard to be a metaphor, it seems
unlear what this is a metaphor
for.
In Lewis's ase, it is fairly obvious that the
soreboard is a metaphor for the partiipants' internal states of mind. In DeRose's ase,
we get to know that it
by denition
gives the right `sore'whih is the orret tally of
semanti ontent for the sentenes in a onversation. It is also lear that DeRose's
ontextualism requires this metaphor to be sound, and for a metaphor to be sound it
must refer to something that atually exists. The question then beomes what the
metaphor of a single, ommon soreboard refers to in the ontext of a onversation.
It is tempting to begin with Lewis's soreboards. A single, ommon soreboard
ould possibly be shown to be a metaphor for a set of individual soreboards
+
a set of
rules or funtions governing the ombination of these soreboards. If these rules or
funtions are onstruted the right way, it ould be possible to redue the metaphor
into mental states, and therefore redue DeRose's soreboard to Lewis's soreboards.
But DeRose expliitly obstruts this straightforward manner of redution, and I will
therefore not attempt to perform it. We must look elsewhere.
Another attempt ould be a metaphysial one. We ould postulate the atual
existene of a soreboarda real entity that somehow keeps trak of the semanti
ontent of utteranes in onversations. Whether partiipants in the onversation are
allowed to tap into this entity to get an update on the sore is not important. This
would obviously solve DeRose's problem, but seems to be grossly implausible. Suh an
entity has never been observed, and based on all the rest of our knowledge of the world
7
it seems unlikely that we will disover it soon. Yet, one suh entity ould be God, and if
DeRose is willing to make that laim, we would have to leave the matter there.
The most forgiving interpretation is that DeRose objets to Lewis's laim that
eah
partiipant has a soreboard, period. Therefore, he onstruts his metaphor of a
ommon soreboard, whih is meant to be a metaphor of the
linguisti pratie
employed by the partiipants. In other words, the single soreboard is a set of inferene
rules, funtions and/or just onventions used to infer the semanti ontent from what
was said. This gives the right sorein other words the orret semanti ontentby
denition, simply beause the denition of semanti ontent is what the partiipants'
pratie rules it to be.
If this interpretation is orret, it seems to be ompletely without explanatory
power. Taken together with DeRose's argument that the semanti ontent of utteranes
made by partiipants in the same onversation is xed through the use of the same
semanti soreboard, this onstitutes a ompletely trivial laim. Inserting our
interpretation into the argument gives us the semanti ontent of sentenes made by
partiipants in the same onversation is xed through the use of the inferene rules,
funtions or onventions used to infer the semanti ontent from what was said. This is
probably true, but it doesn't provide us with muh of a defene of ontextualism.
The appeal of the single soreboard seems to be a lot stronger when kept as a
metaphor instead of being explained fully, in muh the same way as the metaphor of a
nervous stok market is apparently stronger than its expansion: nervous stok
brokers.
Single soreboard semantis, instead of being proven through its existene, is being
proven through its eets. To infer from eet to ause leaves one at risk of
epiphenomenalism. The eets of DeRose's single soreboard might well be explained
otherwise: for instane by disarding ontextualism as suh. If only indexials are
ontext-sensitive, there seems to be little need for a soreboard to settle semanti value.
In this artile, I have shown how DeRose's defene of ontextualism leaves new
questions to be answered. In the rst objetion (that from judgements of omparative
ontent), DeRose's defene rests on the notion of single soreboard semantis, whih I
argue is a less than plausible position. In the seond objetion, DeRose argues that
ontextualism is only struk by this objetion if invariantism is, and therefore it
onstitutes no argument against ontextualism only. In the matter of the objetions
from belief reports, I have shown how DeRose's defene fails to dismiss the
disquotational shema. As far as I'm onerned, two out of three objetions to
ontextualism still stand.
8
Referenes
Cohen, S. (1999). Contextualism, skeptiism, and the struture of reasons.
Philosophial Perspetives 33 (13), 5789.
DeRose, K. (2004, May). Single soreboard semantis.
Philosophial Studies 119 (1-2),
121.
DeRose, K. (2005a, September). Bamboozled by our own words. (draft)
http://pantheon.yale.edu/~kd47/.
DeRose, K. (2005b). The ordinary language basis for ontextualism, and the new
invariantism.
The Philosophial Quarterly 55 (219), 172,198.
Hawthorne, J. (2004).
Knowledge and lotteries. Oxford:
Lewis, D. (1979). Sorekeeping in a language game.
339359.
9
Clarendon Press.
Journal of Philosophial Logi 8,