ANN & MEMORY - RKC

MOWRER
Yes, Bill, I remember visiting Urbana to meet with MOWRER and others, but
not much detail. We did discuss his book, I think, but I now have only
vague impressions of the book and the meeting.

SYMPATHY
Bruce, THANKS for the sympathy! I'm improving my efficiency, but the
quantity of incoming material is still overwhelming! (Is offering
sympathy "Positive Feedback?")

ARTIFICIAL NEURAL NETWORK and MEMORY
John Gardner (930121), refers to suggestions from Albus using a large
memory system relating arm states to values of muscle torques/forces.
Nothing is said about the controlling inputs ("Reference Signals) for this
system. Presumably they come from the operator of the system.

This closely resembles my view of the operation of Human Systems (Bob
Clark 921205).

When a decision is needed (a choice between/among alternatives), the DME
examines ("imagines") related experiences from memory. It considers
conditions (remembered) that may limit the selection(s). Anticipated
results (projected through imagination) are compared with the objectives
for acceptability. The DME may combine selected procedures sequentially,
or use an average (weighted, perhaps) of the imagined procedures.

These imagined procedures are used by the DME as structured inputs to the
corresponding levels of the hierarchy. Under ordinary conditions, this
may take only a small fraction of a second. But if the situation is
complex (and time permits), extensive investigation and study may be used
before finally selecting the procedure.

The whole process is so familiar and quick that it is easily over-looked.

This very general summary becomes more meaningful when applied to real
people in real situations. Saw a figure skating contest (pairs) last
night. VERY complex activities -- mainly muscle skills, but commentators
reported some of the personal interactions that can play a part. I was
struck by the situation when a DISTURBANCE occurred, a fall to the ice.
This is a very complex situation -- the planned sequence, with its timing
requirements, has been suddenly interrupted. This appears to require
extensive re-working of the many systems involved. However the response
-- compensating movements -- was within a fraction of a second! Clearly,
the skaters had available, almost instantly, an alternative procedure. It
was designed to both avoid injury and to continue the program.

Most of these skaters had ten or more years of practice. If you have ever
tried to ice skate, you know that much of early experience involved
learning how to fall without "bruises." Thus they have a large supply of
alternative memories that can be quickly applied when needed.

Notice, while this involves much "repetition," this is NOT
"reinforcement," rather it is acquiring a repertoire of alternative
variations of performance.

In terms of Orders of the Hierarchy, such a contest certainly involves
Interpersonal Relationships (my suggested Sixth Order) and, in various
degrees, all lower orders. In performance, the selected relationships are
played out. But in discussing the contest, communication skills are used.
Here words are used to represent perceived variables at several levels.
The ice, the skates, the arena are (more or less) Objects that can be
considered among the Second Modes of Sixth Order. The movements, with
their timing, would be Third and Fourth Modes. The combination into
skilled performance could be Fifth Order. Over-all there are the personal
interactions of the skaters in a framework of competition. Here we have
Sixth Mode of Sixth Order. This analysis can be carried further and
applied to other activities.

Regards,
Robert K. Clark

ยทยทยท

SUBJECT:RE: ANN & MEMORY - RKC