SUO: Re: Sign Relations & Communication
¤~~~~~~~~~¤~~~~~~~~~¤~~~~~~~~~¤~~~~~~~~~¤~~~~~~~~~¤
Continuation at Mark: >>>--->>>
JA = Jon Awbrey
SR = Seth Russell
SR: I agree that the Interpretant column is problematic were we to
restrain our domain of discourse to humans. However, were we
to change our domain of discourse to computers, then perhaps
we can make some headway. Were we to talk primarily about
Human<->Computer and Computer<->Computer dialogue, would
you not agree that the Interpretant column for the
Computer's role can be found by what resides in
the computer's memory ?? <--- really I need
a direct answer to that question.
JA: The first thing to get clear about is that the distinctive content of
the three relational domains is a secondary issue. There is no sense
to look for distinctive essences that will tell you just from looking
at an item which column it belongs to. Indeed, it is perfectly sensible
to think about a sign relation L c OxSxI which has all three sets O, S, I
equal so far as sets go, that is, with O = S = I. That is slightly unusual,
but it is very common to contemplate sign relations for which S = I, for example,
with S = I = M c A*, where M is a formal language over an alphabet A. In such a
setting one could say that a word x in M, amounting to an element of both S and I,
is "virtually" in a computer memory, speaking in the usual loose-lisp sort of way,
to mean that x has been parsed by the duly appointed parser, has left a record of
its parsing in the form of a parse graph, or other style of data structure, and
has delivered the virtual acceptor that is virtually emulated by the computer
in question into a virtual state of the virtually accepting kind. Ultimately,
then, for the sake of maximal realism, if you like that sort of thing, one
would like to be able to replace the formally characterized sets O, S, I
with three domains that constitute the state spaces of real systems,
always, of course, at the appropriate level of abstraction -- nobody
ever gets all that real if you really wanna think about it in the
strictest possible terms.
JA: You asked the following question:
JA, quoting SR:
| would you not agree that the Interpretant column for
| the Computer's role can be found by what resides in
| the computer's memory ??
JA: Anyway, it seems that there are still a couple of things about
the basic concept of a sign relation that need to be clarified
before we can go any further.
JA: 1. The term "interpretant" in the pragmatic theory of signs is short
for "interpretant sign", so an interpretant has all of the other
properties that any sign might have, and is distinguished solely
by its role in the particular triples of the form <o, s, i> that
reside in the particular sign relation L c OxSxI in question.
JA: 2. Think of what I have been saying as a claim about the "minimal adequate database"
or the "minimal adequate symbol parsing table" that one needs either to model or
to specify a significant fragment of a real, not-excessively-trivial communication
process. Just by way of a catchy title, let's call it the Rosetta Stone Hypothesis.
JA: Applying this to the Computer question -- a section of my dissertation is on this,
and so I assure you that I have been and will continue to take it quite seriously:
JA: There will be a sign relation L that can be said to capture a significant aspect
of what the "computer", that is, a physical system under the control of a program,
is doing with formal syntactic domain S. Here, S will be a formal language over
some alphabet A, that is to say, S c A* = the set of all finite strings over A.
JA: Like I said once before, in these formal contexts we will almost always
have S = I, so the Sign and Interpretant columns will be filled with
elements from the very same formal language S = I c A*.
>>>--->>>
In so far as any agent is acting as an interpreter of a formal language M c A*,
according to the dictates of the sign relation L c OxSxI, where S = I = M, then
signs of M might be regarded as being "in the memory" of that interpretive agent.
But what that does really mean?
We have a habit of saying that signs are "in" the memory
and that the memory is "in" the agent, but really it would
make more sense to say that the agent is "in" the memory and
that the memory is "in" a certain sign. Let me explain this.
Agents are just systems that pass through various states over time.
The memory of the agent is really in its states, not in the partial
components that contribute to its being able to be in those states.
In effect, this means that we are always ultimately concerned with
the state space of a system of which the agent is a representative
point that traces over time one or another conceivable trajectory.
Accordingly, to say that the sign y is in the memory that is in the agent
really means that the agent is in a state that projects on the subspace Z
of features that are relevent to memory in a way that falls into a region
Y c Z that corresponds to the agent maintaining the memory of the sign y.
SR: Ok, if that be it, then fine. Perhaps, then, could we peruse your way
of thinking in another train? (... and I would, ok ok ?) But, Jon, in
this train, whether the Interpretant column can be exactly represented in
a computer's database memory *is* of paramount importance. What is important
to this train is not that an object, x, is somehow a member of each of the three
ideal sets {O, S, I} and hence that the sets can be somehow dealt with equivalently,
but rather what relationships *can be* reasonably depicted in the three different
domains of this ~new~ sign relationship which I am proposing. The ~newness~
coming exclusively from the idea that we are considering the Interpertant
column to be (and only be) that which we can be effectively put in
a computer's database and control its behavior.
I have read this paragraph 3 times and cannot tell
if I have yet answered the questions in it or not.
Please let me know.
SR: Now, would that be at least palatable to you way of thinking, you might then
reexamine my mentograph [1] in a different light and notice: not only the
fact that the three domains in question must needs look quite different,
but also that the relationships between the objects in the three domains
can (at a meta level outside of those domains) be quite mathematically
depicted. What is striking is that two of the domains (O and S of
which you and Tarski obsess ... quite understandably so because of
the one dimensional serializations to which semantics in the 20th
century were restrained) do look very much alike .... yet the
Interpertant domain (which you had initially suggested we ignore)
is where all the distinctive content actually does resides.
SR: [1] http://robustai.net/mentography/intensionExtension.gif
This picture still looks off to me.
1. What are the red boxes with x_1 written over i in them?
2. '"x_j" connotes x_j' seems wrong.
Signs connnote other signs, roughly.
3. 'i intension of y'? and 'y extension of i'?
Should be that i is a common intension of x_1, x_2, x_3,
also that x_1, x_2, x_3 are in the extension of "i", y,
where the later are taken as terms or as concepts.
4. 'x_j isa y'? Will need some concreteness:
x_j = cat_j, an object.
y = "Cat", a string of char.
'x_j isa "Cat"' >>>--->>> syntax error.
I think this is right -- could be tiredness, though.
I will have to "jog my memory" about that augmented sign relation
and what all we were talking about on the "Factorization" thread:
¤~~~~~~~~~¤~~~~~~~~~¤~~~~~~~~~¤~~~~~~~~~¤~~~~~~~~~¤
JA: Let me try to come up with a more concrete version
that has the same structure as the present example.
Then I'll go back and try to answer your questions.
JA glyphed:
o-----------------------------o
| Denotative Component of L'" |
o--------------o--------------o
| Objects | Signs |
o--------------o--------------o
| |
| i |
| /|\ * |
| / | \ * |
| / | \ * |
| o o o >>>>>>>>>>>> y |
| . . . ' |
| . . . ' |
| ... ' |
| . ' |
| "i" |
| |
o-----------------------------o
SR giffed:
http://robustai.net/mentography/intensionExtension.gif
JA: The initial problem had to do with "nominal" thinking versus "real" thinking.
<...>
JA: Let's try this:
x_1 = cat_1
x_2 = cat_2
x_3 = cat_3
JA: Options:
JA: 1. y = "Cat", interpreted as denoting each item of a category.
This is the nominal way of interpreting general terms,
namely, as applying to each separate member of a group,
but without having to posit the group as a whole or
any of its qualities as separately existing entities.
The nominal option is not to augment the sign relation,
but just keep trying to get by with multiple referents.
JA: 2. y = "Cat", interpreted as denoting a category of items.
Here, one is asserting that a category is an object
in its own right, over and above its items.
Here, object i is a new entity like a class or a set.
JA: 3. y = "Catitude", interpreted as denoting a quality that is
possessed in common or shared by cat_1, cat_2, cat_3.
Here, object i is a new entity like an intension or a property.
JA: So, in general, it can happen that a use of the string of char "Cat"
may denote a particular cat, a category of cats, or a catitudiosity.
¤~~~~~~~~~¤~~~~~~~~~¤~~~~~~~~~¤~~~~~~~~~¤~~~~~~~~~¤
SR: In other words, ~cats~ are just a figment of the computer's imagination,
we'll never find ~cats~ in reality. So if we want to talk about ~cats~,
we had better look into the computer's imagination.
Don't know what "~cats~" means.
SR: See I must admit that I have an agenda here too ... hopefully such
personal agendas are allowable when opening these sign inter-coms.
So, just as hopefully, you will recognize that were my agenda to be
swept away as unimportant, then I am not likely to hang in there on
my end of the line.
¤~~~~~~~~~¤~~~~~~~~~¤~~~~~~~~~¤~~~~~~~~~¤~~~~~~~~~¤