[From Dick Robertson, 2007.11.19.1640CST]Hi Rick,I will tell the guys from Wiley
to contact you.Here is a bad copy of the prior article. The encyclopedia pages
havesuch narrow borders that I couldn't bend the book enough to get all ofthe
text into the scan. I have run out of time editing it, but you'llget enough to
see what the last one covered.Here it is"Perceptual control theory" is a name
adopted bya group of scientists interested in the feedback-system organization
of humanand animal behavior to distinguish their work from the control theory
field ofservo engineers. The members are engaged in the development and
applicationofthe thesis advanced in W. T. Powers's (1973) book, Behavior:
TheControl of Perception. While a majority of this group arepsychologists, it
also includes biologists, sociologists, systems engineers,mathematicians, and
members of other professions-all finding themselves able tocommunicate with the
common language of Powers's theory, a rather unusualexperience in this age of
high specialization. One member of thegroup, Richard Marken,
observed that Powers has not one but two accomplishmentsto his credit. First, he
discovered, or noted, that behavior is the control ofperception, contrary to
what psychologists have believed ever since Descartes.For Descartes, the
environment controlled behavior in the sense that he believedperception of
phenomena of the environment stimulates or triggers behavior byentering sensory
receptors to set offreflex responses in the organism. Second,Powers developed a
theory to explain how behavior does work, showing thatfeedback-control theory
can account for how perceptual variables are maintainedagainst external
influences! disturbances by control systems, whether inorganisms or robots. The
theory he developed finally provideda coherent mechanismfor the phenomenon of
homeostasis, which had been observed much earlier bybiologists but had remained
an inexplicable phenomenon. Although biologists hadgradually accepted
self-regulatory-homeostatic-mechanisms after Bernard asapplying to many bodily
functions, the idea that the same principles couldaccount for mental phenomena
gained ground more slowly. However, beginning inthe 1940s a number of
scientists, such as Norbert Wiener, began to suggest thatsuch principles could
explain certain aspects of behavior, if not all. Finally,in 1960 the team of W.
T. Powers, R. K. Clark, and R. L. McFarland published A General Feedback
Theoryof Human Behavior, which presented the first fully comprehensive view of
howall behavior could be accounted for by an integrated assembly of
hierarchically-orderedfeedback control systems. This work then led to Powers's
1973 book. The basic scheme is asfollows. Behavior occurs via a system
comprised of a closed feedback loop in which a variable perceptualsignal (PS)
is held to a specified value--reference signal (RS)-by the workingsof a
comparator (C ). The comparator subtracts the value of the referencesignal from
the value of the perceptual signal to obtain an error(ES) that is fed into an
output mechanismcapable of affecting the perceptual signal in such a way as to
counteract anydisturbance (D) comingfrom the environment. It does so by :'
driving the perceptual signal back toward (maintaining)itat the reference
value. The system worksto keep minimizing the error signal, and in so doing it
controls thepertinentperceived aspect of the environment as a by product.
Peceptualvariables in organisms derive from sensory signals.Reference signals
are previouslystored sensory signals. Output mechanisms are ultimately muscles
and glands. Powers pointedout a number of examples of feedback circuitry
inneuroanatomy in his book and has continued to find additional
anatomicalevidence in his further work.Figure 1 shows the feedback loop outlined
in itssimplest form. The same scheme is repeated in hierarchical fasw which
thesensory signal is relayed to higher order sys:; and the output ofthe
orderabove determines the RS ci order below. This simple schema may be
effectively ap;". to a host of everyday behaviors to explain what is hag ing. A
popularillustration is the action of driving a car:, driver keeps the car in its
laneagainst external dis bances such as wind, curves, bumps, and so forth by
toringthe relationship between the front of the car anc: edgeof the road, both
of which are perceptual variable:; ducible to a higher ordervariable: the
constancy ofrk lationship. The latter is thus the presumptiveperC-E~ signal of
interest. It is matched in the brain to the refe! signal;that is, the desired
condition of that relation the result being the continualflow of error signals
r!::.a::: minimized by the actions of the driver on thesteering. wheel. Powers
showed that this analysis could be quanri£ied in simultaneous equations
5:l1ultaneousequations as PS = ° + D andES =PS - RS, ,.-':iere D represents a
disturbing condition in the environI:'2nt,and 0,the output ofthe system, is
some function of ~ as determined by theproperties of the particular sys"L::D..
He applied these equations to manydifferent analy!E:oof behavioral phenomena,
adding constants as appro~ate to specificsystems, and created a number of
com;rc.:er simulations of various types ofbehavior showing that i:::man actions
can be imitated by programs using hisfunc-:lUllS, the implication being that if
a feedbackmodel, and .~ya feedback model, imitates a humanperformance, it
s=:gests that the behavior is feedback controlled. Theperceptive
reader might have noticed in the above that Powers has also solvedthe problem of
purpose, or intention, that has been a dilemma forstimulus-response
psychologists. It is identical with the reference signal. Thec:r:JITol system
realizes what one intends as the organism's rion brings whatis being perceived
to match the specifiman that is to be perceived. Powerswent on to propose
theoretical answers to many other questions that one might !""'- -eabout the
nature of behavior. He sketched out a hierJ!I"'.:nyof control systems to
account for the complexity of behavnir, in whichindividual control systems of
each level receive their reference signals fromthe output of systems of ttte
level above. Another proposal is that ofan organizing-reorganizing system
powered by an intrinsic system comprised ofgenetically determined reference
settings. He postulated that if any-readings" in the intrinsic system go into an
error state (ie indicatephysical malfunctioning), the reorganizing system would
be triggered to injectrandom signaling into ~ control-system hierarchy to bring
about changes inneural circuitry which, when successful in controlling some
JIFi' forthat organism) condition, constitutes what we regard. aslearning. His
insight was that only random action could afford the chance toproduce a new type
of action (in .~2n organism) because any disturbance toconditions al~iy under
control would immediately benullified by exiJK:.n:g systems. A moment's
reflection leads one toconceive of how thehuman being comes to have a
learnedhierarchy in the first place, startingfrom only an intrinsic system of
genetically given life-supporting systems atbirth, acting via the reorganizing
system upon a growing mass of uncommittedneurons forming and reforming
connections as human development proceeds. O)thermembers of this
group have gone on to apply Powers/s analysis to a widevariety of experiments
and applica;n:= based upon the idea thatliving organisms do not control Their
environments by controlling theiroutputs, but. by controlling their
inputs-theirperceptions. Much of this work constitutes a significant advance in
the testingof hyotheses by quantitative model building and computer simulation
rather thanby inferences of causality from correlations, as is commonly
practiced incontemporary psy::c: ~)gy. This work can be accessed through the
information :0-: CSGnet;the Control Systems Group homepage, http://
ed.uiuc.edu/csg/csg.html; W.T. Powers web site, and thos of manyothers.