Jean-Luc Delatre
2005-08-23 08:01:40 UTC
"Linguistic symbols thus free human cognition from the immediate
perceptual situation not only by enabling reference to things outside
this situation ("displacement" Hockett, 1960), but rather by enabling
multiple simultaneous representations of each and every, indeed all
possible, perceptual situations."
From "The Cultural Origins of Human Cognition" by Michael Tomasello
http://email.eva.mpg.de/~tomas/
Cited in
http://mixingmemory.blogspot.com/2005/08/cogblog-tomasello-chapter-1.html
Thus, trying to "agree" on the "one-true-view" about intelligence,
cognition or anything else is just going against the very basis of our
intellectual capabilities.
Being fixated on logic is likely among the worst possible blind alleys
we can get trapped in.
*Of course* the implementation of any piece of software cannot escape
logic and determinism (even when making choices dependent on a random
noise source).
But this is an engineering constraint *not* in any way an usefull or
wise design guideline.
Apart from this engineering constraint another source of the obsession
with logic and ontology in Western tought comes obviously from the
Greek roots of our culture.
The Greeks tried to have their minds work as a computer before even
having one, consenquently we were bound to invent computers in the end,
not such a bad thing *if we can now go beyond their limitations*.
The strong appeal of logic and precise ontological definitions is just a
*technical artifact* of the use of derivation rules.
No matter which flavor of logic fits your taste, at some point you build
a "consequent sentence" from a bunch of antecedents which are deemed
*true*, by substitutions of symbols into rule patterns, which rules are
deemed "valid" and you expect the "consequent sentence" to also be
"true" or "valid" and usable further in the process.
So far so good...
But what does this entails about rules and symbols?
Rules *have* to be explicitely or implicitely under the scope of some
sort of universal quantification!
Rules must be valid for *any* syntacticaly legal substitutions, this
holds even if you think of any extra applicability/validity conditions
for the rules to be used.
The preconditions just get aggregated to the rule to from the real,
practical rule instead of the "shortened" version which comes more
readily to our mind and discourse.
Those preconditions *too* are under the scope of the "universal
quantification" of sorts.
(This is where many missteps of reasoning happen, forgetting about
implied preconditions)
Here now comes the Ontology, Being and Existence questions!
Because, what is an universal quantification if not the reverse of an
existential quantification?
"Every schoolboy knows" that (forall x R) is exactly (not (exists x (not
R)))
IN ORDER TO MAINTAIN CONSISTENCY OF OUR "DERIVATIONS" WE HAVE TO CHECK
FOR WHAT "EXISTS" OR NOT
But, but, this is only within our little games and plays with symbols
and discourse!
The "connection with reality" is *outside* this logic game, at the
stage where we choose to pretend that such or such word "refers to" or
"names" or (whatever you favorite wording is) a "thing in the world".
It is *deeply misleading* to confuse a technical requirement of symbolic
manipulations with constraints about "the world at large".
This leads to meaningless questions about non-problems which will not
contribute to a better "understanding of the world".
End of Rant... :-)
Now what can we do to improve our "understanding of the world"?
Language could help, in fact, language has to help because we have *no*
choice.
It is the only device we have to share knowledge among us and it is
hopefully "powerfull enough", witnessed by the very discussions on this
list which in spite of all disagreements and misundertandings still
convey some "meaning" and "knowledge" between participants, however
partial or distorted these meanings can be.
To help us the most usefull achievement would be to manage to get
computers to interact with us in manner closer to natural language than
to logic. Computers are already "perfectly logic", short of hardware
failure they do *exactly* what the binary code tells them :-D
To have any hope to do that we will have to somehow "understand" enough
about language to mimic at least part of it with logic (suitable to
computer implementation) NOT THE OTHER WAY AROUND (i.e. trying to
forcefully cast the richness of language into the straightjacket of logic).
It goes without saying that this DOES NOT MEAN that we have to have
"sentient computers" or "spiritual machines" or any delirious goals of
this kind, just being able to state things like "please use algorithm so
and so to deal with this statistical analysis" and get a sensible
outcome instead of scrambling for weeks patching the nitty-gritty
details for an unreliable result.
Trying to eschew this effort because "it's too difficult" or "logic is
known to be reliable" would be acting like the proverbial drunkard
looking for his lost keys under the lampost instead of where they
probably are because "it's too dark over there"...
Cheers,
JLD
========================
To post a message, send mail to cg-***@public.gmane.org
To unsubscribe, send mail to majordomo-***@public.gmane.org with the command 'unsubscribe cg' in the message body.
See http://news.gmane.org/gmane.comp.ai.conceptual-graphs for the mailing list archive.
See http://www.cs.uah.edu/~delugach/CG for the Conceptual Graph Home Page.
For help or administrative assistance, mail to owner-cg-***@public.gmane.org
perceptual situation not only by enabling reference to things outside
this situation ("displacement" Hockett, 1960), but rather by enabling
multiple simultaneous representations of each and every, indeed all
possible, perceptual situations."
From "The Cultural Origins of Human Cognition" by Michael Tomasello
http://email.eva.mpg.de/~tomas/
Cited in
http://mixingmemory.blogspot.com/2005/08/cogblog-tomasello-chapter-1.html
Thus, trying to "agree" on the "one-true-view" about intelligence,
cognition or anything else is just going against the very basis of our
intellectual capabilities.
Being fixated on logic is likely among the worst possible blind alleys
we can get trapped in.
*Of course* the implementation of any piece of software cannot escape
logic and determinism (even when making choices dependent on a random
noise source).
But this is an engineering constraint *not* in any way an usefull or
wise design guideline.
Apart from this engineering constraint another source of the obsession
with logic and ontology in Western tought comes obviously from the
Greek roots of our culture.
The Greeks tried to have their minds work as a computer before even
having one, consenquently we were bound to invent computers in the end,
not such a bad thing *if we can now go beyond their limitations*.
The strong appeal of logic and precise ontological definitions is just a
*technical artifact* of the use of derivation rules.
No matter which flavor of logic fits your taste, at some point you build
a "consequent sentence" from a bunch of antecedents which are deemed
*true*, by substitutions of symbols into rule patterns, which rules are
deemed "valid" and you expect the "consequent sentence" to also be
"true" or "valid" and usable further in the process.
So far so good...
But what does this entails about rules and symbols?
Rules *have* to be explicitely or implicitely under the scope of some
sort of universal quantification!
Rules must be valid for *any* syntacticaly legal substitutions, this
holds even if you think of any extra applicability/validity conditions
for the rules to be used.
The preconditions just get aggregated to the rule to from the real,
practical rule instead of the "shortened" version which comes more
readily to our mind and discourse.
Those preconditions *too* are under the scope of the "universal
quantification" of sorts.
(This is where many missteps of reasoning happen, forgetting about
implied preconditions)
Here now comes the Ontology, Being and Existence questions!
Because, what is an universal quantification if not the reverse of an
existential quantification?
"Every schoolboy knows" that (forall x R) is exactly (not (exists x (not
R)))
IN ORDER TO MAINTAIN CONSISTENCY OF OUR "DERIVATIONS" WE HAVE TO CHECK
FOR WHAT "EXISTS" OR NOT
But, but, this is only within our little games and plays with symbols
and discourse!
The "connection with reality" is *outside* this logic game, at the
stage where we choose to pretend that such or such word "refers to" or
"names" or (whatever you favorite wording is) a "thing in the world".
It is *deeply misleading* to confuse a technical requirement of symbolic
manipulations with constraints about "the world at large".
This leads to meaningless questions about non-problems which will not
contribute to a better "understanding of the world".
End of Rant... :-)
Now what can we do to improve our "understanding of the world"?
Language could help, in fact, language has to help because we have *no*
choice.
It is the only device we have to share knowledge among us and it is
hopefully "powerfull enough", witnessed by the very discussions on this
list which in spite of all disagreements and misundertandings still
convey some "meaning" and "knowledge" between participants, however
partial or distorted these meanings can be.
To help us the most usefull achievement would be to manage to get
computers to interact with us in manner closer to natural language than
to logic. Computers are already "perfectly logic", short of hardware
failure they do *exactly* what the binary code tells them :-D
To have any hope to do that we will have to somehow "understand" enough
about language to mimic at least part of it with logic (suitable to
computer implementation) NOT THE OTHER WAY AROUND (i.e. trying to
forcefully cast the richness of language into the straightjacket of logic).
It goes without saying that this DOES NOT MEAN that we have to have
"sentient computers" or "spiritual machines" or any delirious goals of
this kind, just being able to state things like "please use algorithm so
and so to deal with this statistical analysis" and get a sensible
outcome instead of scrambling for weeks patching the nitty-gritty
details for an unreliable result.
Trying to eschew this effort because "it's too difficult" or "logic is
known to be reliable" would be acting like the proverbial drunkard
looking for his lost keys under the lampost instead of where they
probably are because "it's too dark over there"...
Cheers,
JLD
========================
To post a message, send mail to cg-***@public.gmane.org
To unsubscribe, send mail to majordomo-***@public.gmane.org with the command 'unsubscribe cg' in the message body.
See http://news.gmane.org/gmane.comp.ai.conceptual-graphs for the mailing list archive.
See http://www.cs.uah.edu/~delugach/CG for the Conceptual Graph Home Page.
For help or administrative assistance, mail to owner-cg-***@public.gmane.org