Topic: "Recent Contributions to the Mathematical Theory of Communication" by Warren Weaver
Impressions?
It is interesting that such early theories of Weaver are still important
today.
Weaver discusses three levels of communication:
A. Technical problem
B. Semantic problem
C. Effectiveness problem
He discusses in great detail level A, then claims that level B and C follow
from A: By understanding level A we can easily understand B and C. Has this
come true?
Hasn't come true: This idea is underdetermined, it does not take into account
the large amount of important factors of meaning.
The mathematical ideas of information do not necessarily apply to meaning.
Take for example, a completely random message.
Meaning requires looking at all possible messages, according to Weavers
theory. The information, and therefore meaning, of a single message cannot
be measured without looking at the entire set of possible messages.
Meaning makes a message redundant.
Information is only partially in the message; the rest is determined by what
the recipient does to it.
One person's noise may be another person's information.
There are an infinite number of meanings for a message.
There are a limited number of meanings for a message.
Seems to be two things happening:
1. Meaning is finite in that cognition is finite.
2. Meaning is infinite in a mathematical/computational since.
Weaver describes level B and C in terms of behavior rather than in terms of
representation.
Weaver defines feedback as the measure of how well meaning was communicated.
This is similar to the Turing Test.
Two important questions:
1. How much information is really in a message?
2. Redundancy in a message?
The amount of information is directly related to the amount of reduction in
uncertainty.
An answer to the question "Who has my book? Mary, Bob, Amy, Tim, or Rachel?"
has more information than the question "Who has my book? Mary, Bob, or Amy?"
because the initial amount of uncertainty that was reduced is greater for the
first.
But what if you know Mary very well, but not Bob or Amy, and the answer is
Mary? Wouldn't that have a higher level of information than if the answer were
Bob, who you don't know at all?
What if you have $20 in the book you are missing, then the question has a
different meaning.
(We continue to contemplate the many ways to interpret the meaning of the
question "Who as my book" with various modifications)
(Some important stuff linguistics and it's different levels came up in there
somewhere, but my mind was too boggled at the time to write it down, and I
don't know enough about linguistics to try to recap from memory.)
Layers have been shown to be very effective in computer science, such as the
TCP/IP network architecture: the same strategy can be effective in studying
human communication and cognition.
Yes, we understand "layers" in computer science but what about the human layer?
Information theory may not exactly explain or model human communication and
meaning, but it can be a very useful tool within the context of cognitive science.
