IGS Discussion Forums: Learning GS Topics: Is there a logician in the house?
Author: Ralph E. Kenyon, Jr. (diogenes) Saturday, September 10, 2005 - 11:12 pm Link to this messageView profile or send e-mail

I recommend three books.
"On Intelligence" by Jeff Hawkins (2004)
"How the Brain Evolved Language" by Donald Loritz (1999)
"The Feelings of What Happens", by Antorio Damasio (1999)

They may shed much light on the foregoing "discussion".

I advise not to say that something is impossible or that it will never happen; I advise instead to say, we do not yet know how to do it, or that we have not yet done it. This is the difference between the higher level "judgement" and the lower level "description".


Author: Ralph E. Kenyon, Jr. (diogenes) Saturday, October 1, 2005 - 12:33 am Link to this messageView profile or send e-mail

quote{ ME: The question remains, what work has been done in GS as to its application?

I would very much like to hear if Dr. Kenyon has more to say on the subject.}

I am not aware of specific work attempting to apply fuzzy logic in a "general semantics application". I can speculate that it might be a useful challenge to a graduate student to build some specific models of abstracting from sets of instances to "generalizations" using various fuzzy logic techniques. I leave the design up to prospective student. The object would be simply to explore the notion of abstracting in some specific context area.

Regarding proclamations of "impossible" by "experts",


Disregard everything everybody has ever said; to start out from scratch as if nobody had ever had the sense to think about the problem before; to doubt most of all the opinion of the experts, for, obviously, if the experts were right then there would be no problem. - http://www.xenodochy.org/ex/abstract/eightkeys.html

I particularly liked this bit:


My main interest was in whether any GS experts had looked at the relationship between the 'multi-valued' logics of Korzybski's writings and those of 'fuzzy logic'.

I am certainly no expert in the field, but from what I read in K about 'multi-valued' logics and what I see in Zadeh look very similar. Both depend on probabilistic outcomes of evaluating functions.

A "multi-valued" logic system involves a finite state machine which takes input values from the set of allowed states and produces an output in the set. In binary logic there are two states, represented by "0" and "1" and three machines (which can be represented by "and", "or", and "not"). In tertiary logic there are three states, which can be represented by "0", "1", and "2". In Probability theory, there are "an infinity" of possibilities, the likelyhood of each is expressed by a number in the range of 0 to 1, where "0" and "1" represent "impossible" and "certain" respectively. These are the logics that Korzybski had some exposure to. Fuzzy logic uses characteristics of both of the former and the latter to create "measures" of set membership using an imprecise set of individual properties. Because the measures are not limited to "impossible" and "certain", the results of various operations that we might like to correlate with logical inference are themselves "uncertain".

Some recent work:
http://www-bisc.cs.berkeley.edu/BISCProgram/default.htm "Toward a Generalized Theory of Uncertainty (GTU)—An Outline, Information Sciences, 2005."


So you have here two levels of abstraction. The lowest based on two-valued logic;the highest, supposedly on more than two, say n-valued logic. Now software cannot transcend itself. This means the higher level is written in terms of the lower. If the higher level "works" for any given "n" greater than two. it means they are expressible in terms of the lower. That essentially says that 2-level suffices for all higher levels.

It should probably be pointed out that computers should not be seen as implementing pure "binary" logic, because many devices use three states, high, low, and disconnected - the high impedence state. The simplest devices (latch & driver) have an input lines, an output lines, and a control lines which enable when the input or output lines are active. The simplest one-bit memory cell has three possible outputs high, low, and disabled. My inclination is to think of computers as using tertiary logic to manage binary logic.