[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

real-time information



Al Bregman wrote on 03.03.2001:

...Using the word "information" to refer to any physical fact or relation
in the world broadens the idea of information too much. The thing that is
in the world, is "pattern", not "information".
...Pattern only becomes information when it is used in a communication
system to send a message from a sender to a receiver. Furthermore, it is
important to recognize that the kinds of pattern that count as information
will depend on the properties of the receiver.  If you send Morse code to a
TV receiver, the signal isn't information, but noise.
...Everything is caused by something, but if everything is information, the
word loses its unique meaning.
To be "information", the "pattern" must tell the brain (the receiver of the
"message") about something.  
The existence of "information" is a property of the whole system, not the
patterns taken alone. 
Since the properties of the brain (or *any* receiver) figure in the
definition of "information", then information can't exist, by itself, in
the environment. However, "pattern" can.

Shannon's information theory is widely used in order to optimize audio
systems. A paper by Borst in nature neuroscience tried to apply it to
neural signal processing in the auditory system, too. The above quoted
clarification nurtures my doubts to what extent this approach is justified.
One could ponder whether or not the receiver already includes the cochlea.
Anyway, isn't the audible pattern quite different from any stationary image
in that it continuously links the actual time with the past? Perhaps, just
a small part of what I wrote on 28.02.2001 was understandable. 

I reiterate: "More recent data, alias higher frequencies, are presumably
more important than the past." One could argue that the auditory nerve
receives only real-time data. This is not wrong but also not the whole
truth. Any piece of cochlear partition may be considered as contiouously
memorizing new and simultaneously forgetting old data. On average, the data
are more recent at more basal place on partition, that is, with higher
belonging CF. Mathematicians would reach emphasis on the more recent
aspects of the continuous signal by using the derivative of the signal. So
do those who calculate the instantaneous frequency dphi/dt. The result is
anything but sufficient. The derivative is more rough than the original
signal. Evolutionary selection has led to a much more clever solution.
However, this trick is so unique that there is no way to unite it with
traditional mathematics. I am not sure whether it is already a good example
of a contra-intuitive evolutionary solution. I tend to even regard the
spectral analysis contra-intuitive and phase deafness as well. 

A stingy engineer would not decide to hugely waste valuable bits for
opening an additional frequency-dimension as does cochlea. He would by no
means carelessly loose half of information. He would follow his intuition
within the horizon of his mathematical gospels and fail to come close to
the admirable natural solution. Actually, there was no reason for the
evolution to use Fourier transform. We only may realize that the solution
is similar to this mathematical tool to a certain extent. The phase
spectrum is not included. 

Those who try a lot of sometimes rather questionable effort in order to
violently force cochlear function below the rule of inappropriate
mathematics have to confess: "When various frequencies are simultaneously
present, we can hope that the instantaneous frequency gives a sensitive
average." 
Do not get me wrong. I do not deny the ability of mathematics to possibly
correctly grasp the essence. I just do not regard any solution towards this
objective already being on the right truck. Simply forget most of what I
consider dirty application of mathematics.

Typically, men are weaker than women in having problems to admit errors.
Let's show that we are exceptions. I made a mistake when I wrote actuality
and actualize instead of what is called Aktualität und Aktualisieren in
German, aktualnost in Russian, actualizar in Spain, etc. While actutum in
Latin means in the moment, actualis means efficient. In French actualité
means both reality and news. Topicality would not be a good translation
because the topic rather explicitly relates to the brain-based topic than
to the timing of the message itself. Relevance was perhaps a little bit
less misleading but still not yet adequate to audition in particular.
      

As a corollary, information theory seems to be equally perfect and distinct
from auditory function as is Fourier analysis, at least for those who
strive for correctness and for those who realized that practice suffers
from wrong theory. For that reason, I do not expect too much from
adaptation of both theories. Nonetheless, I am not sure, whether
information theory would allow an extension towards the aspect of gradual
partial neglect of the past. As a layman, I would be interested how the
similar problem of gradual limitation to the field of vision has been
addressed. I reiterate: "In vision, there is a similar gradual decline of
relevance with deviation from focus". Just a silly idea: Could it be
processed in a similar way like temporally delayed components of a signal?
Did not make Peter Cariani an allusion in that direction?

Eckard Blumschein