[From Rick Marken (941207.1745)]
Lars-Christian Smith (941207)--
What are some experiments demonstrating the existence and characteristics
of specific higher levels?
Great question.
First there are the "portable demonstrator" experiments on relative
levels. These experiments are described in "General Feedback Theory,
part II" in W. T. Powers "Living Control Systems". A somewhat more
quantitative approach to demonstrating the existance of two levels
of hierarchical control is described in the "Levels of Intention"
paper by Marken and Powers in Hershberger's "Volitional Action". I also
have a paper in the Fall, 1993 (v. 3, no, 4) issue of Closed Loop called
"The hierarchical behavior of perception" which presents circumstantial
evidence (from the psychological literature on perception and motor
control) of distinct configuration, transition, and sequence levels of
perceptual control. I have done some recent, mini experiments which
show that subjects can control sequences and programs only if they
can perceive these variales (they cannot perceive these variables if
they occur too rapidly); the limitation in the ability to control in
these experiments was unquestionably a perceptual-- not an output --
limitation, showing that, if you can't perceive it (even if you can
"see" it) you can't control it.
Another approach to looking at higher levels is somewhat more qualitative;
it is based on the "method of levels" which is described starting on
p. 41 of Powers' "Living Control Systems II".
There is surely other work on this topic that has not been published.
Tom Bourbon did some nice work on sequence control -- which is
published in the Proceedings of the European Conference of the CSG.
Are you interested in something in particular about higher levels ?
I think it would be great to discuss this because some very interesting
research ideas could emerge.
Best
Rick
Thanks for the references. Two aproaches are possible. The first would be
to discover actual levels. The second would be to design an organism and
find out what the necessary number of levels would have to be.
The first approach seems to have run into difficulties; are there 6 or 12
levels? If 12, why not 13? What is the test?
Are there any a priori reasons why a certain number of levels should be
present?
Has there been any discussion of Harry Klopf et al.'s paper: 'A
Hiearchical Network of Control Systems that Learn: Modeling Nervous System
Function During Classical and Instrumental Conditioning.' Adaptive
Behavior, Vol. 1, No. 3, 263-319, 1993.
Best regards,
Lars
···
Subject: Higher levels
Tom Bourbon [941209.0854]
Subject: Higher levels
Thanks for the references. Two aproaches are possible. The first would be
to discover actual levels. The second would be to design an organism and
find out what the necessary number of levels would have to be.
Lars, I believe the best place to for you to start reading is Powers's,
_Behavior: The control of perception_, published in 1973. The book is about
Bill's attempt to use what you called the first approach.
The first approach seems to have run into difficulties; are there 6 or 12
levels? If 12, why not 13? What is the test?
Are there any a priori reasons why a certain number of levels should be
present?
The book lays out some evidence that, when we talk about people (at least
when we talk about Bill Powers circa 1973), we need more than 6 levels, but
not as many as 13. Back then Bill "settled" tentatively on 10. Later that
went to 11 -- tentatively. The book lays out the rationale, and the test,
we use when we try to identify levels.
Has there been any discussion of Harry Klopf et al.'s paper: 'A
Hiearchical Network of Control Systems that Learn: Modeling Nervous System
Function During Classical and Instrumental Conditioning.' Adaptive
Behavior, Vol. 1, No. 3, 263-319, 1993.
No. The title sounds interesting. Can you give us a synopsis? I'm curious
about the kind(s) of control systems and hierarchies they describe. Are
they Powersian PCT systems, or something else? Also, do they use anything
that resembles PCTish reorganization?
Later,
Tom
Klopf et al.'s model is explicitly built on Power's concepts, which are
discussed in the paper (pp. 297-299).
I will give you Klopf et al.'s synopsis:
"A computational model of nervous system function during classical and
instrumental conditioning is proposed. The model assumes the form of a
hiearchical network of control systems. Each control system is capable of
learning and is referred to as an 'associative control process' (ACP).
Learning systems consisting of ACP networks, employing the direct-drive
reinforcement learning mechanism (Klopf, 1988) and engaging in real-time,
closed-loop, goal-seeking interactions with environments, are capable of
being classically and instrumentally conditioned, as demonstrated by means
of computer simulations. In multiple-T mazes, the systems learn to chain
responses that aviod punishment and that lead eventually to reward. The
temporal order in which the responses are learned and extinguished during
instrumental conditioning is consistent with that observed in animal
learning. Also consistent with animal learning experimental evidence, the
ACP network model accounts for a wide range of classical conditioning
phenomena. ACP networks, at their current stage of development, are
intended to motor semsorimotoro, limbic, and hypothalmic nervous
function, suggesting a relationship between classical and instrumental
conditioning that extends Mowrer's (1956, 1960a/1973) two factor theory of
learning. In conjunction with consideration of limbic system and
hypothalamic function, the role of emotion in natural intelligence is
mi\odeled and discussed. ACP networks constitute solutions to temporal and
structural credit assignment problems, suggesting a theoretical approach
for the synthesis of machine intelligence."
Regards,
Lars
···
Subject: Higher levels