Re: More about questions and the like (was:What I have for dinner...")
John Cowan
> Problem 1: Given "for x, if x went to the party, then John knows thatx went
> to the party" and that Paul went to the party, we might infer "John knows
> that Paul went to the party." This sentence is ambiguous and the most likely
> reading (it is usually said) may well be false. since John may never have
> heard of Paul as such and may have him under a totally wrong-headed
> description, so that we might never find out from John that Paul was there,
So on this view, "George knows that Tully was a Roman orator" is false
even if George assents to the proposition "Cicero was a Roman orator",
given that Cicero is Tully. This seems a perverse reading to me;
I would take it as true.
> Problem 2. From "Pegasus was the winged horse captured by Bellerophon" being
> true,
I grant the rest of your argument, but I deny this premise; I can't
accept that "P. was the winged horse" etc is just uncontroversially
true. It needs to be qualified by something like "In the world of
Greek myth", and even the use of "world" is questionable, because it
is not clear that a mythical "world" might not contain logically
contradictory propositions. In which case we need to talk about what
the Greek myths *say* in which case all bets are off, logically speaking.
--
John Cowan co...@ccil.org
I am a member of a civilization. --David Brin
Py...@aol.com
> "Pegasus was the winged horse captured by Bellerophon"
> is-the-winged-horse-captured-by-Bellerophon"
I like it!>>
Why? It doesn't solve this problem, for now we cannot infer from "John
Kennedy was a Democratic President" to "Some Democrat was Predisent." It also
makes all sentences about Pegasus true -- even though they are no longer
about Pegasus, for it screws up semantics, replacing reference to objects by
reference to singletons, and giving senses to words that don't got none,
like"Pegasus."
pc
<<> -- and the universal quantification doesn't license the
> inferences "There was a winged horse" and "Winged horses have
> existed."
The problem is that we want to imply that there *are* winged horses, in a
certain context. Using the above, "Bellerophon was the winged horse
captured by Pegasus" would be equally true. In reality, of course, both
sentences do have equal truth values, but we want to indicate that we're
actually in a very particular fiction.>>
The point of xu'a. (I am not at all sure that ther two sentences needs must
have the same truth value in reality, but that is a product of some
uncertainty about what truth value either has in reality.)
And Rosta
>
> Problem 1: Given "for x, if x went to the party, then
> John knows thatx went to the party" and that Paul went
> to the party, we might infer "John knows that Paul
> went to the party." This sentence is ambiguous and the
> most likely reading (it is usually said) may well be
> false. since John may never have heard of Paul as such
> and may have him under a totally wrong-headed
> description, so that we might never find out from John
> who is in fact Paul - whoever John may think him to be
> -- that he went to the party. The set of answers
> solution for questions needs quite a bit of extra work
> work to be addapted to indirect questions (and
> propositional attitudes generally). Like including mappings
> from the world to the belief worlds involved, for this one.
> And several other things for the other ones.
> Problem 2. From "Pegasus was the winged horse captured by
> Bellerophon" being true, it is automatic to infer "There
> was a winged horse" and thence "Winged horses have
> existed." But they haven't. The role of xu'a or whatever
> is simply to prevent these inferences in the cases where
> context does not (and so should always be used, just in
> case). It does not say which performative is involved,
> only that it is opaquifying and that the ordinary rules
> thus do not apply -- in particular, names need not
> denote. The alternatives -- in a logical language --
> are to make obvious truths false or to allow truth value
> gaps or to deny the usual rules; none of these are
> impossible but all are unpleasant.
It seems to me that both problems are avoidable by treating
names as predicates (which IMO is right & proper).
Hence
"Paul went to the party"
= "x is-Paul & x went-to-the-party"
And the formula "for every x, if x went to the party, then
John knows that x went to the party" thus no more entitles
one to conclude
"john knows that x is-Paul & x went-to-the-party"
than to conclude
"John knows that x is-overweight and x went-to-the-party"
on the basis of
"x is-overweight and x went-to-the-party"
Likewise, for the second problem,
"Pegasus was the winged horse captured by Bellerophon"
= "for every x, if x is-Pegasus then x
is-the-winged-horse-captured-by-Bellerophon"
-- and the universal quantification doesn't license the
inferences "There was a winged horse" and "Winged horses have
existed."
--And.
Py...@aol.com
even if George assents to the proposition "Cicero was a Roman orator",
given that Cicero is Tully. This seems a perverse reading to me;
I would take it as true.>>
"George knows that Tully was a Roman orator" is ambiguous between our two
interpretations, and the one you dislike is usually said to be the normal
interpretation. That however may be because it is the one that highlights
the usual intensional problems. The trick is, does George assent to the
claim "Tully was a Roman orator"? If he says, "I never heard of Tully" (as
he well might in spite of his remark about Cicero), then it is hard to see
him as knowing anything about Tully, for the rule is that mere actual
identities don't guarantee substitution in intensional contexts (essentially
a definition of same, in fact). It is true that of the man who is in fact
Tully, George knows that h was a Roman orator, but that is somehting else
again (your -- less likely, they say -- reading).
<< >Problem 2. From "Pegasus was the winged horse captured by Bellerophon"
being true,
I grant the rest of your argument, but I deny this premise; I can't
accept that "P. was the winged horse" etc is just uncontroversially
true. It needs to be qualified by something like "In the world of
Greek myth", and even the use of "world" is questionable, because it
is not clear that a mythical "world" might not contain logically
contradictory propositions. In which case we need to talk about what
the Greek myths *say* in which case all bets are off, logically speaking.>>
That is the point of xu'a, to remind us that we are in some intensional
context like "Greek myths say." We do treat such sentences as true and ones
like "Pegasus is a unicorn" as false without the warning, so, in a logical
language, we need the warning, either contextually or explicitly.
pc
Py...@aol.com
party (perhaps with a relative completeness guarantee or something like "and
if x did not go to the party then John does not believe that x went to the
party") has to be used with care. One should not infer from it that, if Paul
went to the party, John knows that Paul went to the party, at least in the
sense that if John is asked "Did Paul go to the party?" he will say yes, even
if he is being as cooperative as possible. the problem is the intensionality
of "know," for John may not know Paul under the name "Paul" and may even know
him under the mistaken guise "Bill's father" (when he is actually Joan's
father). To be sure, the quantifier outside the context guarantees that the
generality is indifferent to the disguises, but the individual cases have no
such obvious external connection. We need either explicitly write "under
some concepty" somewhere here or flag the cases with the opposite of subject
raising (which I think Lojban has). The set-of-answers version of questions
does not work so well in this case, unless the questions are subdivided into
identity classes and all that is required is that John know at least one
member of each of the appropriate classes.
A couple pages on from this point in the Handbook of Philosophic (i.e.,
freaky) Logic is the reminder that every natural language sentence is a
dependent of a (usually unexpressed) performative, usually "I tell you that"
or some such. However, some of these performatives may also be intensional,
in which case every term in the surface sentence is in that cloud-cuckoo-land
where Leibnitz's law fails along with existential generalization and
universal instantiation. In particular, sentences mentioning non-existent
objects which are nonetheless held to be true are under performative like "I
now recite to you a bit of myth that..." This being the case (and it sure
solves a lot of problems), Lojban needs to dig into its small stock of unused
cmavo for a flag of this sort for when context is not enough. Remembering
that the term-length flag of this sort is something like tu'a, I suggest the
corresponding x form, xu'a. this refers to a different enduring problem form
ages past.
pc
Py...@aol.com
names are from the point of view of the speaker (bearing in mind the
listener), so that if I use "la djan" in a sentence, the only thing that
matters is whether I and the listener know who John is, not whether le
djuno uses that name (or description) as part of lenu le djuno cu djuno.>>
I wonder if we could have decided that and then could make it stick for lo se
djuno. In a lot of cases, it is clearly important what concept/name is
involved in the clause: John knows that the number of planets is larger than
seven has to be about the number of planets, not some other name of nine
(especially since John may not know it is nine) . If John thinks that the
number of planets is eleven and knows that the number of players on a
football team (which he has right) is larger than seven, that will not count
as his knowing that the number of planets is larger than seven. Similarly,
if John knows Paul under some wrongheaded description but knows that the
person he knows under that description went to the party, that may well count
for knowing that Paul went to the party. My and your concepts don't seem to
count in any consistent way.
<<This sounds like our ancient overly-discussed veridical unicorn
problem. You know - whether "lo iunikorn" is equivalent to "da poi
iunikorn" thereby claiming that at least one unicorn exists. Are you now
suggesting xu'a as a resolution to this issue, or are you suggesting xu'a
to kick us into some imaginary world when no otherwise stated.>>
Yes, that is, both: solving the probolem by kicking it into an imaginary
world not otherwise stated.
<< I'm not sure how any of this ties into performatives, or whether we need
to
start seeking boxes again zo'o.>>
Performatives are just a way of getting in the intensional contexts that
"seek" already provides for the boxes, but that are not apparent for loose
talk about Pegasus and unicorns.
pc
And Rosta
>
> The "for all x, if x went to the party then John knows that
> x went to the party (perhaps with a relative completeness
> guarantee or something like "and if x did not go to the
> party then John does not believe that x went to the party")
> has to be used with care. One should not infer from it
> that, if Paul went to the party, John knows that Paul went
> to the party, at least in the sense that if John is asked
> "Did Paul go to the party?" he will say yes, even if he is
> being as cooperative as possible. the problem is the
> intensionality of "know," for John may not know Paul under
> the name "Paul" and may even know him under the mistaken
> guise "Bill's father" (when he is actually Joan's
> father). To be sure, the quantifier outside the context
> guarantees that the generality is indifferent to the
> disguises, but the individual cases have no such obvious
> external connection. We need either explicitly write "under
> some concepty" somewhere here or flag the cases with the opposite
> of subject raising (which I think Lojban has). The
> set-of-answers version of questions does not work so well in
> this case, unless the questions are subdivided into identity
> classes and all that is required is that John know at least one
> member of each of the appropriate classes.
I don't see that we're falling into this trap, at least on
the "for all x, John knows that x is..." treatment as opposed to
the set of answers one. This is precisely because the criterion
by which x is to be identified from whatever is not x is not
included within the proposition that John knows.
> A couple pages on from this point in the Handbook of Philosophic (i.e.,
> freaky) Logic is the reminder that every natural language sentence is a
> dependent of a (usually unexpressed) performative, usually "I
> tell you that" or some such. However, some of these performatives
> may also be intensional, in which case every term in the surface
> sentence is in that cloud-cuckoo-land where Leibnitz's law fails along
> with existential generalization and universal instantiation. In
> particular, sentences mentioning non-existent objects which are
> nonetheless held to be true are under performative like "I
> now recite to you a bit of myth that..." This being the case
> (and it sure solves a lot of problems), Lojban needs to dig into its
> small stock of unused cmavo for a flag of this sort for when context is
> not enough. Remembering that the term-length flag of this sort is
> something like tu'a, I suggest the corresponding x form, xu'a.
> this refers to a different enduring problem form ages past.
I don't understand the problem or the proposed solution. Or at least,
I am aware that performatives are necessary, but assuming they are
available (linguistically), then I see no outstanding problems. And
I don't see how you envisage {xu'a} working. What we need is a way
to identify (a) something as a performative, and (b) its scope. And
since the baseline has been set, all we can actually do is ask "do
(a) and (b) exist?".
--And.
John Cowan
> One names-as-predicates. It is odd semantically: names usually (certainly in
> English) don't have a sense, just a referent
Everybody says this since Frege, including And, but I still think that
the sense of "Fido" is "dog".
> With the xu'a approach, this sort can take place
> without odd readings of names and as part of a general rule about intensional
> operators, which we will need anyhow.
I don't think this works in general. Consider the following statements:
The _Arabian Nights_ was translated by Sir Richard Francis Burton.
Scheherezade told a story about a genie and a fisherman.
The genie threatened to kill the fisherman.
If the first sentence is true in the real world (it is), and the second
sentence demands xu'a, what does the third sentence demand? xu'axu'a?
No one simple trick will work for all cases.
--
Schlingt dreifach einen Kreis vom dies! || John Cowan <jco...@reutershealth.com>
Schliesst euer Aug vor heiliger Schau, || http://www.reutershealth.com
Denn er genoss vom Honig-Tau, || http://www.ccil.org/~cowan
Und trank die Milch vom Paradies. -- Coleridge (tr. Politzer)
John Cowan
>
> From: Py...@aol.com
>
> << I think at one point that we decided that intentional descriptions and
> names are from the point of view of the speaker (bearing in mind the
> listener), so that if I use "la djan" in a sentence, the only thing that
> matters is whether I and the listener know who John is, not whether le
> djuno uses that name (or description) as part of lenu le djuno cu djuno.>>
>
> I wonder if we could have decided that and then could make it stick for lo se
> djuno. In a lot of cases, it is clearly important what concept/name is
> involved in the clause: John knows that the number of planets is larger than
> seven has to be about the number of planets, not some other name of nine
But this is not a name in the sense meant above; it is a veridical description
of nine, not a name of nine. I can say "John knows that George is greater than
seven" if by "George" I mean "the number of planets" (quotes are mandatory here).
> (especially since John may not know it is nine) . If John thinks that the
> number of planets is eleven and knows that the number of players on a
> football team (which he has right) is larger than seven, that will not count
> as his knowing that the number of planets is larger than seven.
I agree with this.
> Similarly,
> if John knows Paul under some wrongheaded description but knows that the
> person he knows under that description went to the party, that may well count
> for knowing that Paul went to the party.
This sounds like Bernard J. Ortcutt again.
reci...@freenet.edmonton.ab.ca
> Likewise, for the second problem,
>
> "Pegasus was the winged horse captured by Bellerophon"
> = "for every x, if x is-Pegasus then x
> is-the-winged-horse-captured-by-Bellerophon"
I like it!
> -- and the universal quantification doesn't license the
> inferences "There was a winged horse" and "Winged horses have
> existed."
The problem is that we want to imply that there *are* winged horses, in a
certain context. Using the above, "Bellerophon was the winged horse
captured by Pegasus" would be equally true. In reality, of course, both
sentences do have equal truth values, but we want to indicate that we're
actually in a very particular fiction.
co'omi'e xarmuj.
Bob LeChevalier (lojbab)
>From: Py...@aol.com
>
>Problem 1: Given "for x, if x went to the party, then John knows thatx went
>to the party" and that Paul went to the party, we might infer "John knows
>that Paul went to the party." This sentence is ambiguous and the most likely
>reading (it is usually said) may well be false. since John may never have
>heard of Paul as such and may have him under a totally wrong-headed
>description, so that we might never find out from John that Paul was there,
>even though he knows of the man who is in fact Paul - whoever John may think
>him to be -- that he went to the party. The set of answers solution for
>questions (and propositional attitudes generally). Like including mappings
>from the world to the belief worlds involved, for this one. And several
>other things for the other ones.
I think at one point that we decided that intentional descriptions and
names are from the point of view of the speaker (bearing in mind the
listener), so that if I use "la djan" in a sentence, the only thing that
matters is whether I and the listener know who John is, not whether le
djuno uses that name (or description) as part of lenu le djuno cu djuno.
>Problem 2. From "Pegasus was the winged horse captured by Bellerophon" being
>true, it is automatic to infer "There was a winged horse" and thence "Winged
>horses have existed." But they haven't.
This sounds like our ancient overly-discussed veridical unicorn
problem. You know - whether "lo iunikorn" is equivalent to "da poi
iunikorn" thereby claiming that at least one unicorn exists. Are you now
suggesting xu'a as a resolution to this issue, or are you suggesting xu'a
to kick us into some imaginary world when no otherwise stated.
I'm not sure how any of this ties into performatives, or whether we need to
start seeking boxes again zo'o.
lojbab
----
lojbab ***NOTE NEW ADDRESS*** loj...@lojban.org
Bob LeChevalier, President, The Logical Language Group, Inc.
2904 Beau Lane, Fairfax VA 22031-1303 USA 703-385-0273
Artificial language Loglan/Lojban:
see Lojban WWW Server: href=" http://xiron.pc.helsinki.fi/lojban/ "
Order _The Complete Lojban Language_ - see our Web pages or ask me.
Py...@aol.com
to the party" and that Paul went to the party, we might infer "John knows
that Paul went to the party." This sentence is ambiguous and the most likely
reading (it is usually said) may well be false. since John may never have
heard of Paul as such and may have him under a totally wrong-headed
description, so that we might never find out from John that Paul was there,
even though he knows of the man who is in fact Paul - whoever John may think
him to be -- that he went to the party. The set of answers solution for
questions needs quite a bit of extra work work to be addapted to indirect
questions (and propositional attitudes generally). Like including mappings
from the world to the belief worlds involved, for this one. And several
other things for the other ones.
true, it is automatic to infer "There was a winged horse" and thence "Winged
simply to prevent these inferences in the cases where context does not (and
so should always be used, just in case). It does not say which performative
is involved, only that it is opaquifying and that the ordinary rules thus do
not apply -- in particular, names need not denote. The alternatives -- in a
logical language -- are to make obvious truths false or to allow truth value
gaps or to deny the usual rules; none of these are impossible but all are
unpleasant.
John Cowan
> The trick is, does George assent to the
> claim "Tully was a Roman orator"? If he says, "I never heard of Tully" (as
> he well might in spite of his remark about Cicero), then it is hard to see
> him as knowing anything about Tully,
This seems to contradict your other claim. A monolingual German cannot assent
to the sentence "Snow is white", for it is mere gibberish for him, but that
does not mean that he does not know that snow is white. Similarly, if Gheorghe knows
that the man talking on the TV just now has brown hair, and the man in question
is (all unknown to Gheorghe) Bill Clinton, then it seems to me extremely
arbitrary to deny that Gheorghe knows that Bill Clinton has brown hair, even though
George would presumably (if he were a cautious logician type) not assent to the
sentence "Bill Clinton has brown hair". So what you would assent to is only
an indirect indication of what you believe or know.
> That is the point of xu'a, to remind us that we are in some intensional
> context like "Greek myths say." We do treat such sentences as true and ones
> like "Pegasus is a unicorn" as false without the warning, so, in a logical
> language, we need the warning, either contextually or explicitly.
I grasp that now, but I think that we need full semantic world-setting, not just
a syntactic marker. Ray Smullyan's skeptic, after all, believes that the mental
states he is experiencing now (while awake) are the same in kind as those he
experiences while dreaming, merely at a different level --- he would not be
surprised to "wake up" from this current life.
Jorge Llambias
Here's another argument against the expansion,
"for all x, John knows that x is...", this time not involving
names.
Let's say this morning John bought two bottles of
milk and put them in the fridge, and just to make it
simpler let's say the fridge was otherwise empty.
Now, Mary tells John that she took one of the bottles
out of the fridge. She obviously doesn't tell him which
one, because nobody cares which one she took.
Does John know what is now in the fridge?
Yes, he knows that there is a bottle of milk in the fridge.
Is it true that for all x, John knows whether x is in the fridge?
No, he only knows that one of the two bottles that he
put in the fridge is there, but he doesn't know which
one (nor does he care).
So, in this case at least, we cannot take the quantifier out
of the intensional context.
co'o mi'e xorxes
Py...@aol.com
both -- and the referent is an object, not a set -- as it is for predicates.
On problem 2, using the nap approach makes all sentences about Pegasus true,
which is as objectionable in context as the non-denoting names approach that
makes them all false (or undefined). Pegasus was the winged horse but
Pegasus was not a unicorn. With the xu'a approach, this sort can take place
operators, which we will need anyhow.
As for problem one, if John doesn't know Paul as Paul, he probably does not
know that he has the property is-Paul either and so not that something both
is Paul and went to the party, i.e., that Paul went to the party on the one
reading. On the other hand, it probably does cover the other reading, that
there is something which is-Paul and John knows that it went to the party.
But this does not require the odd predicate is-Paul (rather reads it as "=
Paul") to work. Indeed, that is one general solution for these cases, treat
"Paul" or whatever as an external "quantifier" to work in: in Lojban, set
some variable to "Paul" in the prefix: "For x = Paul, John knows that x went
to the party." The problem is with the name inside the the intenly waysional
context, not outside. The trick is always to disambiguate in the less likely
way -- when the thing involved is real. So the real trick is to know when
that is.
pc
Bob LeChevalier (lojbab)
>From: John Cowan <jco...@reutershealth.com>
>Py...@aol.com wrote:
> > That is the point of xu'a, to remind us that we are in some intensional
> > context like "Greek myths say." We do treat such sentences as true
> and ones
> > like "Pegasus is a unicorn" as false without the warning, so, in a logical
> > language, we need the warning, either contextually or explicitly.
>
>I grasp that now, but I think that we need full semantic world-setting,
>not just
>a syntactic marker. Ray Smullyan's skeptic, after all, believes that the
>mental
>states he is experiencing now (while awake) are the same in kind as those he
>experiences while dreaming, merely at a different level --- he would not be
>surprised to "wake up" from this current life.
If we need full semantic world-setting, we should have it in the form of
sei metalinguistic parentheticals. If there is an advantage to something
like xu'a (which is not clear) it would be for brevity. At which point we
might find that one of the evidentials or attitudinals will suffice. ka'u
or se'o, perhaps, may already be providing the role of xu'a.