Vous êtes sur la page 1sur 7

Turner, J.W., Jr. & Fine, T.H. (Eds.

) Proceedings of the Second


International Conference on REST. Toledo, OH: IRIS, 1987.
3
Zubek, J.P. (Ed.) Sensory Deprivation: Fifteen Years of Research. New A Cybernetic Model of Restricted
York, NY: Appleton-Century-Crofts, 1969. Environmental Stimulation Effects
Zubek, J.P. Behavioral and Physiological Effects of Prolonged Sensory and
Perceptual Deprivation: A Review. In: Rasmussen, J.E. (Ed.) Man in Michael Bross
Isolation and Confinement. Chicago, IL: Aldine, 1973, 9-83.
In the relatively short history of sensory restriction research the
field has progressed from the lament of "Many facts and few theories"
(Vernon, 1963) to a plethora of diverse theoretical models which attempt to
account for the variety of phenomena produced by restricted environmental
stimulation techniques (REST). One early attempt to impose a theoretical
framework was Schultz's (1965) neurophysiologically based concept of
sensoristasis, and the first major review of the area (Zubek, 1969) included
two important chapters on theoret i ca 1 developments, one emphas i zing the
neurophysiological perspective (Zuckerman, 1969), the other concentrating on
cognitively based explanations (Suedfeld, 1969) a dichotomy also retained in
Suedfeld's (1980) review of the field. A recent monograph which focusses on
IlEST phenomena as brought about by floatation (Hutchison, 1984) contains
thirteen chapters on different theoretical explanations; and while most of
these can also be classified into physiological and cognitive categories,
"ome also draw on evolutionary principles to explain certain REST effects.
(;iven this abundance of theoretical models, why look to general system
theory to propose an additional model? Firstly, and most importantly, it
would be desirable to have a theoretical orientation that is, at least in
principle, capable of not only explaining the great variety of REST effects
obtained, but also able to accommodate the broad range of the more specific
Ihl)oretical approaches which have been proposed. Secondly, it is my
'1lllLention that a cybernetic model ing approach as derived from general
.,y"tem theory is ideally suited to force investigators to carefully analyze
wh,iI the adjustment of an organi sm exposed to REST actually is in terms of
wh .• t it is designed to accomplish from the organism's point of view, an
.lppmach that looks beyond the empirical data and the identification of
'.1 till tural components involved in producing REST effects. Finally, there is
~ flliidamental philosophical reason for suggesting a cybernetic perspective
I" Ilr"tC'rstand REST effects, a reason that derives from the assumption that
"III'C'xIH·ripnce and understanding of the world is based on an internal model
"r "1'011 ily which is built lip and derived from the information we obtain from
30
31
the world via our sensory experience. given value, a value which is determined by the reference signal Xi. Thus,
While this is not the place to elaborate on this proposition, briefly, in an ideal situation where Xi = Xo (i.e., Xi - Xo = 0) the system produces
it derives from the philosophies of Berkeley, Hume, and Kant and has also no error signal and is said to be stable. The equation Xi - Xo is
been championed by Helmholtz who arrived at this position not from a constantly monitored by the comparator and whenever Xi - Xo = 0, e takes on
philosophical basis, but from the empirical consideration that we are never a positive or negative value and acts on the controlled quantity K in such
aware of objects or events per se but only know about them indirectly a way as to reduce or increase the feedback signal to approximate it again
through the activities of our nervous system. For both the philosophical to the value of the reference signal. By rearranging the layout of Wiener's
and the empirical arguments the conclusion is identical, namely that we basic model and substituting terms more congenial to living organisms, the
construct in our minds (from a philosophical - cognitive perspective) or in cybernetic control system in Figure 2 results.
our central nervous system (from an empirical - physiological perspective) Referenco
a model of the world which is utilized by us to orient, respond, and Signal, r
interact with the world. What makes REST relevant to this approach (and
vice versa), is that REST effectively reduces or alters the normal flow of
I ,Comporo-, I

sensory input which is used, in the first place, to construct our world lor <r...l
model and, also, serves the function of continually monitoring and/or
Perceptual Error
adjusting this model to achieve a close correspondence between the model and Signal, p S1gnol,8
the world.
The most basic type of cybernetic model was proposed by Wiener (194B)
in the form of a single loop control system that contained five variables Input Oulput
FUllction System Func I/on
(Figure 1).
fijl~'K~

SYS1U1'S

Proximal Envlronmen t ACTllXl CI~ II::


---,---r--r-

XL + e K
Physlcol
SlImull Proximal Resulls

Physlcol
Laws Physical
Laws

X 0, ~ 1
OIlflUllli lI1HlllY na

Idealised single-loop control system


riqure 2. Cybernetic Control System an~ ito interaction
with an environment. Adapted trom Powers. 197].
Where: Xi a reference or command signal
Xo a feedback or input signal
e an error or output signal The nature of the p and e signal s are readily ascertained in most
III a comparator (of Xi and Xo) cybernetic systems whereas the source and value of the reference signal (r
K a controlled quantity or Xi) is somewhat obscure. In mechanical systems r is normally set at a
The objective of such a control system is to keep the variable K at a given value, e.g., a thermostat in a climate controlled room. In liVing

32 33
systems we can, in general, distinguish three types of reference signals, system is in general too simplistic. Rather we deal with a hierarchy of
arranged hierarchically: horizontally as well as vertically interconnecting subcontrol systems where
a) Genetic:generated by DNA, RNA the e signals from one such subcontrol system become the p or r signals of
b) Intrinsic:generated by neuro-chemical reactions e.g., hormones other subsystems. Secondly, the most efficient type of control systems are
c) Acquired:generated by experience those which have a model of the controlled quantity K as part of their
All three types generate r values which are appropriate for specific system, an arrangement that leads to the optimal performance of such systems.
objectives at a given time and for a given state of the organism. The Figure 3 is an example of such a system.
genetic r signals must be regarded as more or less fixed, i.e., only subject Another important feature of incorporating a model of K into the system
to alteration under highly unusual (abnormal) conditions (e.g., X-ray is that the model can take on different parameters as a function of other
radiation), the intrinsic r signals are flexible within certain limits, and control systems, i.e., it can be a flexible component. Most control systems
acquired r-values have the greatest flexibility. From the perspective of operate on the principle of negative feedback where the objective is to
REST effects, the intrinsic and acquired r signal s have the greatest stabil ize the system at a given, constant value. Indeed, these types of
flexibility since the efforts of the organism in response to reduced levels systems are inherently unable to deal with positive feedback, i.e., where
of stimulation can readily be regarded as attempts to reinstate the equation the e signal produces (via K) greater and greater values of the p signal.
p = r by some means. Some types of control loops, however, are susceptible to positive feedback
in that this type of feedback leads to adjustments in the control loop and
here one can point to the model within the system as that component which
will respond to the positive feedback by altering its parameters.
The above considerations are drawn from systems engineering, and while
®~~®
c;cnU'AHAllHt similar or analogous control loops do exist in organisms with nervous
I' systems, an interesting case of the present discussion arises where the
.
(",lth aodcl of 10K")
model is not only a model of the controll ed quant ity, but al so a second
comparator that monitors, integrates and regulates some other important
I functions of the system itself. One specific example that would fall into
COtll'ARAl'oR this category is the regulation of sensory thresholds by an organism exposed
to sensory restriction. By assigning to the concept of "optimal level of
-~- ...__ .-_ .. ~ .... ------_..._....-_.._...- ..._.-_._... \timulation" a reference signal "ro " set at a certain value, the following
relationships can be diagramed:
K
The following interactions arise out of the relationships given
in Figure 4.
a) Sensory signals (P) go to both cortical and subcortical regions
in this case the reticular formation.
F I~,"~e 3. A Control system wJtt, an JIIlorllal
model of the controlled qu.'llilil Y K. b) The controlled quantity (K,) is the amount of afferent neural
excitation that is passed on to higher order neurons from lower
order neurons. Normally this amount is held below maximum
Before going into a specific example, two additional points must I", channel capacity by tonic inhibition (sensory gating) exerted by
considered. For one, the view of the organism as a single cybernetic contrlll efferent neural signals.

34 35
OrTlflAl. I.EVEI. or STlIlUl.ATIOff Assuming linear approximations, one can derive the foll owi ng set of
equations:
1. e = r - p
r (0)

2. p = Ke
3. p = K(r - p)
" (e)
4. p = KL
I + K

(ANAS)
For systems with more than one controlled quantity, equation (4) expands to
incorporate the different values of K.
"(NI)
In the case of 2 k variables of interest, equation (4) can be written
(in simplified form) as:
5, p = t'.,_r_
• (I)
_ _ _ _ •• _ .. _1 • , _••• _•• _ __ _. ._ . I + K2
If K2 represents a given unit of time, the value of p will represent the
k SUISOKY' Itf.S1M leTIUN loop gain of the system per unit time, and equation (5) will generate a
1
negatively accelerated curve that after a given time stabilizes at an
,,'jlju["e 4. A cybernetic JQodel of 8el~ory (jatinq. asymptotic level.
Now, this function fits quite well with empirical findings on the
I'rFect of sensory restriction on changes in sensory sensitivity. Taking some
c) The amount of afferent signals not subject to tonic inhibition d.lta from Bross and Zubek (1975) on the effect of 14 days of auditory
is regulated by the arousal states (needs) of relevant cortical r'('striction (silence) on the critical flicker frequency (CFF) of the eye and
processes (r o ) ' Il'lting K2 being equal to days of auditory restriction, K, set at optimum
d) Any inequality of p~ or r~ leads to a change in the reference PlFiciency for transduction (i.e., a value of 1.0), and r as an ad hoc
signal r AF for the reticular formation which results in: "pproximation of 2.0, the fit between the observed and predicted function is
I) a change in the value of the error signal e w normally (in ',hown in Figure 5.
the case of sensory restriction) a negative value which in In principle, one can move from specific subsystems which comprise the
turn would lower the amount of tonic inhibition in the lllllltitude of functions a living organism carries out to a perspective whir·
sensory gate, and, n''1ards the organism as a single cybernetic system by focussing on
2) a change in the val ue of e la" whose pathways can be regarded ,orllTol functions the organism carries out. For example. a ma,i"
as belonging to the ARAS, will lead to an increase in loop must be the organism's attempt to generate and maintain
arousal signal s to cortical areas to compensate for the 0' the world where the collective perceptual signals are r

input usually received via classical sensory pathways (p), 10 Ipst the correctness or val idity of that model, P
These relationships can be stated in a formal manner and cast in mathematical II,,· introduction, given the way we process in f
«'

form. ',1'0"'1 case can be made that we experien r ~ "I


"'-::> v
111,'"1"'1', For this casl' a cybernetic apr 0'" I'll! I'

Irlll'l'l'"t in'l. testable pred iet. ions. .'.11111 "

',1'111,,1-. is lTitical 10 Ihl' rp II I' 1111111'

36

orrltlAl. I.EVE•. OF STltlUl.ATION Assuming linear approximations, one can derive the following set of
equations:
1. e = r - p
r (0)
2. p Ke
=

3. p = K(r - p)
0' (e)
4. p = KL
1+ K
(ARAS)
For systems with more than one controlled quantity, equation (4) expands to
incorporate the different values of K.
I'(H)
In the case of 2 k variables of interest, equation (4) can be written
(in simplified form) as:
5. P = ]c;,_r_
• •..•.­ 1-,..-,-----1.•••­ ••••••­ •.­ .•..- . - - - - -.•.••...••••••.••••••• 1 + K,
If K, represents a given unit of time, the val ue of p will represent the
t:: StJISOMY Mt:SIMlt..:TlUN
loop gain of the system per unit time, and equation (5) will generate a
2
negatively accelerated curve that after a given time stabil izes at an
t"hJure 4. 1\ cylJernetJc Klodel of sCllJ50ry qat.inq.
asymptotic level.
Now, this function fits quite well with empirical findings on the
effect of sensory restriction on changes in sensory sensitivity. Taking some
c) The amount of afferent signals not subject to tonic inhibition data from Bross and Zubek (1975) on the effect of 14 days of auditory
is regulated by the arousal states (needs) of relevant cortical restriction (silence) on the critical flicker frequency (CFF) of the eye and
letting K, being equal to days of auditory restriction, K, set at optimum
processes (r o ) '
d) Any inequality of p~ or r M leads to a change in the reference efficiency for transduction (i.e., a value of 1.0), and r as an ad hoc
signal r~ for the reticular formation which results in: approximation of 2.0, the fit between the observed and predicted function is
1) a change in the value of the error signal e w normally (in shown in Figure 5.
the case of sensory restriction) a negative value which in In principle, one can move from specific subsystems which comprise the
turn would lower the amount of tonic inhibition in the owltitude of functions a living organism carries out to a perspective which
f(>gards the organism as a single cybernetic system by focussing on major
sensory gate, and,
I llntrol functions the organism carries out. For example, a major control
2) a change in the value of e la" whose pathways can be regarded
as belonging to the ARAS, will lead to an increase in IllllP must be the organism's attempt to generate and maintain a stable model
III the world where the collective perceptual signals are constantly utilized
arousal signals to cortical areas to compensate for the
input usually received via classical sensory pathways (p). I II test the correctness or val idity of that model. As briefly alluded to in
Ihp introduction, given the way we process information about the world a
These relationships can be stated in a formal manner and cast in mathematical
'.1 r'llng case can be made that we experience the world in this inductive
form.
OJ.lllller. For thi s case a cybernet i c approach to REST effects can make some
11I11'r'psting, testable predictions. Given that the role of sensory/perceptual
'.1I11,aI5 is critical to the maintenance of such a model, the controlled

36
37
44.5,­ 5
I

~
¥ of the variables involved such as length of SR, type of SR and so on, but it
could easily be applied to specific situations.
44.0 .­
~ A second instance where an interesting and also unique prediction can
o
be generated from the cybernetic model of REST is in the area of adaptation
z
a
N"l.S 3 to novel perceptual environments. Given that sensory reduction entails a
p significant decrease in the individual's ability to monitor the validity or
'"'
W
0..
2
~ "3.01_ ", '
correctness of his/her world model via sensory feedback, it should follow
'I~

that there is also a decrease in the probabil ity of the individual's


~ expectation that the model which is operational at the state of a sensory
b4U .­
.... _... __ ... _------------_.._-_.._-----._-...­ restriction period will have the same degree of validity after it ceases .
o
42.0 Assuming that the testing and adjustments of the world model has a high
priority in the hierarchy of control system, a reasonable prediction would
:( I I I I I I .1 f t I I I , ,., ,
0123456 7 B 9 10 11 12 13 14
be that any experimentally induced mismatch of sensory feedback and the model
DAY S following sensory restriction would lead to a more rapid adaptation (and
Changes in the CFF during 14 days of auditory deplivation adjustment of the model) by the individual than for someone who has not been
exposed to sensory restriction since the latter has not had the probabilities
Fi8"T" 5. EmpiricAl and theoretic"l curves of the loop
gain in the CFF. After Bross and Zubek, 1975. concerning the correctness of his/her world model reduced. A direct way to
evaluate this prediction would be to test it by means of an optical
rearrangement procedure (Held and Hein, 1958; Kohlers, 1964; Welch, 1978).
quantity r would not only consist of the total amount of p that can be By shifting the visual field through distorting prisms and assessing the
obtained by the individual, but also the type of information that will be speed by which an individual adapts his/her visu-motor behavior, it would be
preferred (utilized) to other types of information. Moreover, one should predicted that individuals exposed to sensory restriction adapt much more
expect that certain types of information normally disregarded or only qUickly than individuals who have not.
peripherally drawn upon will now be util ized to a greater extent. The In addition to generating predictions, a cybernetic approach to REST
following two predictions can be made on the basis of these considerations: as drawn from general systems theory is well suited to incorporate or
Given two types of information, both normally valuable to the complement a great number of the variety of theoretical explanations that
individual about the world, an individual exposed to sensory restriction wil I have been offered for REST effects. Thus the concepts of sensoristasis,
prefer to obtain that type of information which contributes towards thf' optimal level of stimulation and stimulus hunger can readily be incorporated
maintenance of a stable model about the world. into the approach outlined in this paper. Explanations such as Maclean's
For instance, an individual who is a devoted baseball fan and will 90 (1973) triune brain and Budzynski's (1983; see this volume) hemispheric
to great lengths to obtain information about the weekend results and lateralization model would represent the case where specific structural brain
statistics of baseball games should prefer other types of information i' ,lreas are assigned to the location of control systems with different
exposed to severe sensory restriction, e.g., verbal directions about till' functional objectives. Cognitive models of REST effects, too, can be
layout of his/her immediate physical environment, weather reports, generdl ,'ccommodated from a cybernetic perspective, for example, the concept of
and local newscasts. ullfreezing (Suedfeld, 1980) would represent an instance where the reference
This example is somewhat general and in need of greater specificatioo ',1'111<11 for a particul ar control system is significantly altered. One feature
III (0'lnitive models that is of particular interest is that they are more

3il
39
1ikely to represent "software" rather than "hardware" control systems, a Suedfeld, P. Restricted Environmental Stimulation: Research and Clinical
feature that makes them readily amendable to alterations and adjustments. Applications. New York, NY: John Wiley and Sons, 1980.
The considerable evidence on the facil itatory effect of REST on attitude Vernon, J. Inside the Black Room. New York, NY: Clarkson, 1963.
change, changes in self image etc., can be drawn upon in support of this Welch, R.B. Perceptual Modification: Adapting to Altered Sensory
contention. Environments. New York, NY: Academic Press, 1978.
The arguments presented in this paper do not, of course, present an Wiener, N. Cybernetics. New York, NY: Wiley & Sons, 1948.
exhaustive analysis of the viability of cybernetic theory to REST. The Zubek, J. P. (Ed.) Sensory Depri vat i on: Fi fteen Years of Research. New
specific implications of such an approach will have to be worked out in much York, NY: Appleton-Century-Crofts, 1969.
greater detail and subjected to empirical tests. The promise this approach Zuckerman, M. Theoretical formulations: 1. In: Zubek, J.P. (Ed.) Sensory
holds, in addition to presenting a more unified view of REST phenomena, is Deprivation: Fifteen Years of Research. New York, NY: Appleton­
that it will lead to a perspective that will enable investigators to identify Century-Crofts, 1969.
the reference signals, levels of control, and the controlled quantities by
which we maintain our everyday behavior and expectations about the world, as
well as our attempts to change these behaviors and expectations.

REFERENCES
Bross, M. & Zubek, J.P. Progressive increase in the CFF of the non-occluded
eye during prolonged monocul ar deprivation. Canadian Journal of
Psychology, 1972, 29, 340-347.
Budzynski, T.H. Biofeedback and the twilight states of consciousness. In:
Schwartz, G.E. & Shapiro, D. (Eds.) Consciousness and Self-Regulation:
Advances in Research, IV New York, NY: Plenum, 1976.
Held, R. & Hein, A. Adaptation of disarranged hand-eye coordination
contingent upon reafferent stimulation. Perceptual and Motor Skills,
1958, .6., 87-90.
Hutchison, M. The Book of Floating. New York, NY: Morrow &Co., Inc., 1984.
Kohler, 1. The formation and transformation of the perceptual world.
Psychological Issues, 1964, 1.
MacLean, P.O. The triune brain, emotion, and scientific bias. In: Schmitt.
F.O. (Ed.) The Neurosciences. New York, NY: Rockefeller University.
1970, 336-348.
Powers, W.T. Behavior: The Control of Perception. Chicago, IL: Aldint',
1973,
Schultz, D.P. Sensory Restriction. New York, NY: Academic Press, 1965.
Suedfeld, P. Theoretical formulations: II. In: Zubek, J.P. (Ed.) Sensory
Deprivation: Fifteen Years of Research. New York, NY: Applel.oll
Century-Crofts, 1969, 433-448.

40 41

Vous aimerez peut-être aussi