JPRS ID: 9560 USSR REPORT LIFE SCIENCES BIOMEDICAL AND BEHAVIORAL SCIENCES

• Document Type: CREST
• Collection: JPRS
• Document Number (FOIA) /ESDN (CREST): CIA-RDP82-00850R000300080043-3
• Release Decision: RIF
• Original Classification: U
• Document Page Count: 114
• Document Creation Date: November 1, 2016
• Sequence Number: 43
• Content Type: REPORTS

APPROVED FOR RELEASE: 2047102/08: CIA-RDP82-00850R000300080043-3
HOR OFFICIAL USE ONLY
JPRS L/9560
20 February 1981
_ USSR Re ort
~
LiFE SCIENCES
BIC)MEDICAL AND BEHAViORAI SCIENCES
(FOUO 4/81 ~
Fg~$ FOREIGN BROADCAST INFORMATION SERVICE
- FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/08: CIA-RDP82-00850R000300080043-3

APPROVED FOR RELEASE: 2007/02148: CIA-RDP82-44850R000300084443-3
NOTE
.TPRS publications contain in�ormation primarily trom foreign
newspapers, periodicals and books, but also from news agency
transmissions and broadcasts. Materials from foreign-language
sources are translated; those from English-language sources
are transcribed or reprinted, with the original phrasing and
other characteristics retained.
Headlines, editorial reports, and material enclosed in brackets
are suppli~d by JPRS. Processing indicators such as [Text]
or [Excerpt] in the first line of each item, or fo'lowing the
last line of a brief, indicate how the original information was
processed. w'here no processing inciicator is given, the infor-
mation was summarized or extracted.
Unfamiliar names rendered phoneticzlly or transliterated are
enclosed in parentheses. Words or names preceded by a ques-
tion mark and enclosed in parentheses were not clear in the
original but have been supplied as appropriate in context.
Other unattributed parenthetical notes within the body of an
item originate with the source. Times within items are as
given by source.
The conkents o.f this publication in no way represent the poli-
cies, views or attitudes of the U.S. Government.
~OPYRIGHT I.AWS AI~1D REGULATIONS GOVERNING OWNERSHIP OF
MATERIALS REPRODL'CED HEREIN REQUIRE THAT DISSEMINATION
OF THIS PUBLICATI~V BE RESTRICTED FOR OFFIC~L USE ONI.Y.
APPROVED FOR RELEASE: 2007/02/08: CIA-RDP82-00850R000300080043-3

APPROVED FOR RELEASE: 2007/02/48: CIA-RDP82-00850R000300084443-3
FOR OFFICIAL USE ONLY
JPRS L/9560
20 February 1981
USS~ REPORT
- LIFE $CIENCES
BIOMEDICAL AND BEHAVIORAL SCIENCES
(FOUO 4/8~.)
CONTENTS
HUMAN FACTORS
Engineering Psychological Ilesign of Semiautomatic Systems That
Ma,ke Use of the Tracking Principle 1
Collaboration of Socialitst Nations in the Field of Engineering
Psychology 15
PHYSIOI,OGY
Patterns of Perception of Visual Signals 25
Automatic Analysis of Color Vision 36
Optical Methods of Transforming Visual Feedback .o 63
PSYCHOLOGY
Somes Aspects of the History of Developmen t of Soviet Military
Psychology 74
Scientific Conference of the Gor'kiy Department of the Society ~
of Psychologists 88
Titles of Candidatorial ?issertations Defended in 1979 90 _
Topics of Scientific Research Dealing With Psychology 92
Psychology and Robot Technolog?y 94
Fifth All-Union Conference on ~gineering Psychology 105
, - a_ [IlI - USSR - 21a S&T FOUO]
APPROVED FOR RELEASE: 2007/02/08: CIA-RDP82-00850R000300080043-3

APPR~VED FOR RELEASE: 2007/02/08: CIA-RDP82-00850R000300080043-3
FOR OFFICIAL USE ONLY '
HUMAN FACT~fi
ENGINEERING PSYCHOLOGICAL DESIGN OF SEMIAUTOMATIC SYSTEMS THAT MAKE USE OF THE ~
TRACKING PRINCIPLE
. Moscow PSIKHOLOGICHESKIY ZHURNAL in Russian No 5, 1980 pp 105-116
[Article by A. P. Chernyshev, submitted 30 Nov 79]
[Text] The growth of semiautomatic systems in the world in the last few years indi-
cates that such systems have b~come predominant at the present stage of technologi-
cal progress [6]. In our opinion, the causes are as follows: in the first place,
the limited technical capabilitiea for creating systems with complete automation
of control proc~sses and, in the second place, the drastic increase in their cost
- and complexity with increase in level of automation.
All semiautomatic devices are referable to the man-machine system class, and for
- this reason engineering psychological studies of interaction between them and man
are gaining decisive importance to their deaign and operation. _
Suhstantial changes have taken place in the requiremenCs refersble to man's mental
activity in connection with scientific and technological progress; situa~iona have
begun to arise more and more often where the human operator is at the limit of his -
capabilities. In such casea, the human factor limits the function of the entire
system.
Yet, there is still a rift between the approaches to describing the psychophysio-
logical characteristics of man and machine parameters, due to the specifics of
rese~rch methods in psychology and engineering. However, to develop an integral
' ["single"] man-~machine system there must be an integral approach to it as a whole
- and ~ singZe language to describe it. Nzvertheless, only one subsystem has been
- scbmitted to estimation and design thus far, tite ob~ect of control, and the reason
for this is the lack of well-su~bstantiated principles of modeling the performance
of an operator.
; In order to take into consideration the human (pri;r1arily ps4chological) factor, it
is partf cularly important to develop approachea to describing human performance that
would pex~tt ready retrieval of inf ormation to solve technical problems and, at the
same time, to see the psycholQgic~l meaning of manifestations of operator per-
formance by means of criteria used in engineering practice.
_ At present, a general conception of engineering psychology has been formed as a
science of information-related interaction between man and machines, as well as
psychological regulation of controlling actiona. The problem of designing the
work of an operator has taken a central place in this conceptian.
1
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/08: CIA-RDP82-00850R000300080043-3

APPROVED FOR RELEASE: 2007/02/48: CIA-RDP82-00850R000300084443-3
FOR OFFICIAL USE ONLY
, This problem was formulated for the first Cim~ by B. F. I,omov, in ].963, a~td in 1967
it was formulated as one of the most important and vromising problems of engineering -
psychology [5]. The past years have shown that this problem is becoming more and -
fiore pressing. The works of V. F. Venda, B. A. Dushkov, N. D. Zavalova, G. M~
Zarakovskiy, V. P. Zinchenko, A. A. Krylov, V. I. Nikolayev, V. A. Ponomarenko, V.
F. Rubakhin, A. 7. F3.lippav.and V. D. Shacirikov have dealt with various aspects of
this problem. 'Phe work of K. K. Platonov, V. V. Chebysheva and others in the field
of psychological analysis of work performance are very important to it. Academicians
A. I. Berg, B. N. Petrov and V. A. Trapeznikov [1, 7, 12] have indi~ated the import-
ance of work in this direction.
However, while there have been considerable advances in working on the bases of
engineering psychological design of systems and performance, we must note that ~
there has been extremely little work on deveiopment of specific methads for design-
ing man-machine systems, in the development of which several problems arise. We
shall list here the mast important ones.
l. The problem of creating a common approach to the description of operation of the -
technical part of the system and operator performance.
2. The problem of cons~deration of individual psychophysiological characteristics
of performance (the~e differences are of a random nature in a selected group of
people).
3. The problem of consideration. of dynamics of performance characteristics in the
training process.
4. The problem or screening operators with Lhe necessary characteristics for work-
ing with a specific control objecC (problems of training and occupational screening
emerge as phases of the systems approach to the design of human work).
_ In aur opinion, any technical part of the system, no mat~er how complex it is, is
vie~wed as a tool of labor. On the one hand, the innat.^. 13mitations of human capa-
bilities require that work tools be created that broaden these capabilities; the
other hand, the work tools developed impose certain demands on human performance.
As we know, efforts tu describe the operation of the machine part of a syatem and
performance of an operator were made in engineering psychology in analyzing operator
performance in ttie tracking mode. Use was made of mathematical models taken from
automatic regulation theory (chiefly transfer functions). The first transfer
functions dzscribin~ operator performance in the tracking mode appeared in the
United States in the 1950's (the models of McRuer, Krendel, Elkind and others
[15, 16J). Subsequently, the system of nonlinear theory of autumatic control
(the model of Diamanrides [17]) was used to create performance models; nonstationary
and time variable models of operator performance were cre~ted [14, 18]. Such models
were also developed in our country (A. V. Drozdov, I. Ye. Tsibulevskiy and others
[9, 10]).
_ In the vast majori~y of cases, the developers of Chese models proceeded from the
premise that the human operator is the funczional element of a technical system, and
the methods developed for analysis of such systems were used to describe his per-
formance. The main description task was to define the "output" reactions of man to
certain "input" signals. -
L
FOR OFFICTAL USE ONLY
APPROVED FOR RELEASE: 2007/02/08: CIA-RDP82-00850R000300080043-3

APPR~VED FOR RELEASE: 2007/02/08: CIA-RDP82-00850R000300080043-3
FOR OFFICIAL USE ONLY
The process of constructing msthematical models of performance is done mainly by -
t~ao methods.
The standard model method consists of having the model, which is formed as an analog
circuit [scheme], connected in parallel in the operating circuit with the operatoY.
Man's action is compared to the model's output signal with regard to some criterion
chosen in advance (most often the criteriun of inean-square error). The parameters
of the model are so ad~ usted as to minimize the selected criterion of comparison.
This method makes it possible to identify the actions of the operator, but does not
disclose the psychological characteristics of the actions proper. Such models can
. be useful and applicable only when the man in the system must be replaced by an
- automaton.
~ The method of spectral analysis permits description of the composition of response
actions of the operator. The appearance of the link that performs such a conver-
sion is determined from the appearance of the spectral composition of output and in-
put signals as an approximation. The models created by this method are referable
to the quasilinear class. They too do not permit the study of psychological distinc-
~ tions of human actions (their structure and mechanisms of psychological regulation).
Whatever the means used to develop them, most mathematical modpls did not find
practical applications. And this is not because the very idea of developing -
mathematical models of performance is not fruitful. On the contrary, as observed
by B. F. Lomov, modeling is "a powerful means of constructing psychological theory"
[4J. A. A. Krylov, V. N. Nikolayev, V. F. Rubakhin, V. D. Shadrikov and others also
rate highly the role of modeling in analysis of operator performance and its forma-
lization. The fact of the matter is that the above~mentioned mathematical models
did not adequately describe the psychological distinctions of operator performance.
_ First of all, they did not reflect important features thereof, such as dependence of
reaction on form and significance of input signal, adaptive capabilities of the ope-
rator, his fatigability; dependence of the reaction on conditions of performance,
level of operator motivation, his functional state, etc. They overlooked the fact
- that operator actions are mediated by mEntal reflection, by the conceptual models,
o~erational images of the control ob~ect. As a result, the operator could not be
distinguished f rom the machine, and his specific traits were overlooked. The
basic principle of Soviet engineering psychology, which states that the relation-
ship between man and machine in control systams is the relationship between the
"subject of labor and tool of labor" [4], is not applied in the construction of these
models. For this reason, only the external, resultant manifestation of distinctions
of the work under study was reflected in the models in question.
Most systems, with the exception of the simplest ones, can be described only approx-
imately by mathematical equations. This is attributable to the fact that we either
do not know all of the factors that affect their behavior, or that we obtain
excessively cumbersome equations that are difficult to solve by modern methods.
Applied problems usually deal with a small number of aspects of system behavior.
For this reason, the mathematical model has to be made as simpie as possible, in
order to concentrate on the study of the parameters that have the strongest in-
fluence on the behavior of the system. Consequently, as the first step in our
study, we must become convinced of the fact that the mathematical model is adequate
for the system modeled.
' 3
FOR OFFT(:TAL USE ONLY
APPROVED FOR RELEASE: 2007/02/08: CIA-RDP82-00850R000300080043-3

APPROVED FOR RELEASE: 2007/02/48: CIA-RDP82-00850R000300084443-3
FOR OFFICIAL USE ONLY
When working with a simplified modeZ, one must be caut~ous and constantly check
whether the assumptions made are still valtd wRen changing to the study of a new
system. It is desirable for the model to be general enough, so that a wide range
of processes and vari at~ons of the work under study can be simulated.
Mathematical models help us comprehend the behavior of complex systems, and to ~
predict it; they are useful in training. They can be used to analyze various
control situations, to demonstrate or formulate new problems, develop new principles
for the design of technical systems and algorithms of operator performance.
As we have already stated, the use of transfer function is one of the popular methods
of modeling human performance. Indeed, this function is the most convenient form
of describing the dynamics of system function in the practice of designing automatic '
- control systems. This rather simple idea drew the attention af many researchers
concerned with modeling operator performance [3, 9, 10, 14], although the term
"human transfer function" is not used quite correctly in the psychological litera-
ture. It seems obvious that there urlll be different correlations for open and
closed systems including an operator. Man performs a different conversion of
input information in an open and closed system, F~hile the usual element of
the automatic control system does not change its properties, whatever circuit it _
is connected to. Moreover, having identified the mathematical model of operator
perforniance in one situation, it is virtually never possible to extrapolate it to
another situation. For this reason, the question arises as to the validity of
- using the concept of transfer function to describe the per�ormance of a human
operator.
What is a transfer function? To what cl~.s ~~f systems does this concept apply7
Can one extrapolate this function to the ~c.~cription of man's performance in the
system? If so, what is the range of application of this concept?
In automatic control theory, the concept of transfer function was initially intro-
duced for the class of linear systems, i.e., systems to whi~h the superposition prin-
ciple applies. It consists of the following: if severul perturbing factors are
- applied to a linear system simultaneously, their ~oint effect equals the sum of
effects induced by each factor separately.
The principle of superposition enables ug to describe ~he reac:ion of a linear
system to an arbitrary perturbation in the form of the sum of reactions of thi~
system to elementary perturbations, for which purpose it is sufficient to make an
expansior.. Then, knowing the reaction o~ a linear system to elementary perturba-
tions of this type, we can define its reaction to arbitrary perturbation by apply-
ing the superposition principle. As a result, the dynamic properties of the linear _
, system are entirely characterized by its reaction any standard type of perturbation. _
- In principle, one can always choase any type of elementary gerturbation and, accord-
ing to it, define the characteristics of a linear system.
In so doing, one must be governed by the following consideration~: ~he class of the
function, which can be expanded according to ~ype of elementary perturbation of
the type we selected, must be as broad as possible; such expanslon should not pre-
sent any difficulties.
We obtain various characteristics of the linear system, depending on the choice of ,
type of perturbation. But each of these characteristics will be exhaustive, since -
knowledge thereof is sufficient to find the reaction of the linear system to any
_ _
FOR OFFIC?AL USE ONLY
APPROVED FOR RELEASE: 2007/02/08: CIA-RDP82-00850R000300080043-3

APPR~VED FOR RELEASE: 2007/02/08: CIA-RDP82-00850R000300080043-3
FOR OFFICIAL USE ONLY
perturbation. There can be the following types of elementary perturbation: ha~monics,
delta [Dirac] function, or atep function. Accordingly, a distinction is made of the
following: frequency characteristics in the case of harmonic perturbation; pulsed
transient function with perturbation in the form of a Dirac function; transient
characteristics in the case of stepwise perturbation fed to the system's input.
They are all ir.terrelated and can be derived from one another. If one of these
characteristi~s of the system is known, one can define the output signal with
arbitrary perturbation. Thus, the frequency, pulse and transient functions of the
system, or any other characteristic, completely define the properties of a linear
dynamic system. They depend only on the system's own dynamic properties, and they
_ are unrelat~d to the parameters of factors influencing it. The frequency charac-
teristics of the system ~(~jiu) are related to the pulsed transient function k(t) as
follows:
;
- ~U i j~,~l ~ k i t) e-~Wr~lt.
0
The pulsed transient ftinction of the aystem ie related to the frequency characteris-
tic by inverse r^ourier transform:
c+1ao
k(() = 2n f Q~ (j~) c~`"1dw.
N~~
_ The transient charact~ri$tic h(t) and pulsed k(t) are related as follows:
k ~t~ _ ~(4 -
Transfer function ~(s) of the linear system is expressed through a pulsed transient
function by means of direct Laplace transform:
~
~ (s~ _ ~ ~ fl . e-,rdt.
J
0
The transfer function is defined as the ratio of the system~s output, transformed
accordin~ to Lap~ace, to the sya~tem's input transformed according to Laplace at
zero initial conditions.
This definition is often used in the psychological literature without consideration
of the system~s class, whicl-~ leads to erroneous i*~terpretation of the concept of
transfer function. For exsm.ple, in j8] there are 20 mathematical models of operatoi
performance only in the tracking mode~ which were obtained by different suthors
between 1951 and 1968. Since the transfer. function of a linear system is unrelated
5
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/08: CIA-RDP82-00850R000300080043-3

APPROVED FOR RELEASE: 2007/02148: CIA-RDP82-00850R040340080043-3
FOR OFFICIAL USE ONLY
to the type of input factor (it is determined solely by the properties o~ the system
itself), the profusion of mathematical models, even for the same type of performance,
obtained by different authors indicates that the principle of superposition for this
class of systems has not been adhered to. This means that the concept of "operator ~
transfer function" is invalia, even for the same type of work.
In spite of the fact that the concept of transfer function is valid only for linear
systems, efforts were made to apply it to analysis of nonlinear systems [2]. The
analog of transfer function for nonlinear systems is known under the name of image
[representative] function. This concept is based on. the fact that the response of
a nonlinear system to harmonic perturbation is occasionally very similar to the
harmonic signal (thus, if we take into consideration only the first harmonic of the
output signal and disregard the presence of higher harmonics, we can find at each
frequency the ratio of amplitudes of the system's output to its input. In the
general case, this ratio depends on the amplitude and frequency of the input signal,
and in a linear system only on frequency). The procedure of substitution of a non- ~
linear system with a linear one is called linearization of a system, i.e., the
actual performance of the operator is described by a linear model that offers the
best approximation to real performance.
Let us consider the possibility of constructing an image function of operator per-
formance, limiting ourselves to modeling simple sensorimotor activity, w~ithout _
dealing with more comp�licated cases. Compensatory tracking in one plane of a har-
monic signal may be a forra of such activity.
The proposed experiment consisted essentially of the following. A 1ow-frequency
generator of periodic oscillations was LS~:.' deliver a harmonic signal; the ope-
rator tracked it by means of a potentiomete: control handle. The tracking error
that was formed in this process was fed to a display (cathode-ray oscilloscope -
with screen 15 =m in diameter and 8 cm range of beam oscillations). The distance
between the display screen and the subject's eyes was 70 cm. The experiment was
- conducted in daylight. The control had a 1-m lever (distance from point of appli-
cation of controlling action to axis of potentiometer s~nsor). An amplification
unit served as the ob~ect of control.
The operator can predict well the harmonic signal, and he adapts easily to changes
in its characteristics. Thus, a harmonic signal is fed to the input of the man-
machine system (Figure 1), and there must be the same harmonic signal at the output
of this system, without phase lag and, consequently, in the ideal case, the transfer
function of the man-machine system should equal one.
For a formalized description of performance, we must Consicier the exogenous eFfect
on the operator and his reaction. Here, we can distinguish two phases. The first
phase is when the operator perceives a mistake in the system, which is delivered from
the display to the visual analyzer in the form of a harmonic s3gnal. There is no
human reaction. This is the phase of perception of the harmonic signal. The second
phase is when the motor reaction, even of an experienced operator does not conform
with the delivered sigcial. The operator perceives the error on the display in the
form of a stationary random function of time. This is the phase of statiot~ary
' activity. The error is a stimulus that causes regulation of sensorimotor trans-
_ formation. However, we cannot rule out the possibility that termination of the
~
FOR OFFII'iAL USE ONLY
APPROVED FOR RELEASE: 2007/02/08: CIA-RDP82-00850R000300080043-3

APPR~VED FOR RELEASE: 2007/02/08: CIA-RDP82-00850R000300080043-3
FOR OFFICIAL USE ONLY
tirst phase does not mean that there is traceless disappearance of the sensory
graphic image of the harmonic signal. It may be that further tracking occurs with
involvement of representation mechanisms. To rule out or accept this hypothesis,
the followiiig experiment is conducted: the operator is asked to track a harmon~Lc _
signal while the mean of tracking error disappears from the display screen at a
time that is not known in advance to the operator. The operator continues with the
tracking, reproducing the previously presented signal by representation (the
frequency of which was fixed in each.specific experiment). The experi.ment was
continued in the range of 0.05 Hz to disruption of operator tracking.* Since track-
ing error is a variable, determination was made of statistical characteristics of
error in tracking when the display beam is on. The boundary of the statistical
"tube" of error was plotted. An error is a stationary random process with a poly-
harmonic component, and it is a stable feature for each operator. Then the segment '
of oscillogram obtained with the display beam off was submitted to statistical pro-
- cessing (Figure 2). The time during ~rhich the statistical operator error was
in the range of the "statistical tube" depends on the frequency of presented har-
monics and is little-related to individual distinctions of trained operators.
Figure 3 illustrates such a function for two operators:
t us
T~ ct) = oe'! c~~ ' ,
where Tc~f9 is the relative time of operator work in the "statistical tube" of _
error with disp lay beam off; t~ Z(f) isabsolute time of operator work in the
statistical ran ge of error with display beam off; cT~ Z(,~ is mean-square deviation
of error signal with display beam on; f is frequency of presented signal.**
~~t~ E~t~ E~t~~ Ob ~ Ct- � ~~t~ -
- Display Gperai:o ~
con rol
Tracking error
' Out~ut sxgnal o~ system
~ Input signal -
Figure 1. q(t)--input signal of system e(t)--tracking error
x(t)--output signal of system r(t)--operator's control signal
As a result of processing, it was found that the maximum time that an operator can
work within the statistical tube of error does not exceed 4 s at a frequency of
0.05 Hz and 2 s at 0.1 Hz. The time diminishes drastically witti increase in
*Disruption of tracking refers to the magnitude of.~ dispersion of error, which equals
dispersion of input signal, i.e., the ~ase that is tantamount to operator inaction.
**Subscripts "c" and "e" refer to "correct" and "error."
7
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/08: CIA-RDP82-00850R000300080043-3

APPROVED FOR RELEASE: 2007/02/48: CIA-RDP82-00850R000300084443-3
FOR OFFICIAL USE ONLY
f.requency of delivered signal. Consequently, the action of the mechanism of repre-
sentations is limitPd to a maximum of 4 s. Evidently, the hypothesis of operator
function in excess of 4 s according to represent ation [imagination?] should be
set aside because it was not confirmed experimentally.
e(t) ~(t)--tracking error
- "Tube" of statistical. Time display beam t--current time, s
tracking errors is turned off ~.eP--time of operator work by
~ representation
~ ~ mZ ~mathematical expectation of
t, L error signal in partition
�S t~ S interval
' Figure 2.
The experiments revealed that the error displayed to the operator on the oscillo-
seope screen, in spite of its seeming randomness, contains a harmonic component,
in addition to random noise, which corresponds to the harmonic of the signal deli-
vered to the input of the man-machine system. Altheugh the mafn harmonics con-
- tained in the tracking error are much smaller in amplitude than the harmonics of
the delivered signal, this information is quite sufficient for dominance of ampli-
- tude at the frequency of the input signal in the ~perator's reaction. The appear-
ance of additional harmonics in the oper^.ror's response is indicative of the effect
of nonlinear transformation of input signu by the operator. Nonlinearity of
transformation can be evaluated as a functi~.i of coherence.
r~, s ~
4
lst oper
-2d oper T~--redu.ed tracking time according
3 i to iepresentation, s
~ ~ f--frequency nf delivered harmonic
Z signal, Hz
~ .
1 ` ` -
0 D, 2 D, 4 0, 6
f, xz
Figure 3.
First of all, we shall demonstrate that the coherence function equals one with
linear transformation of the signal:
" Is x(~)I' = 1.
a
,t Spg Sxx ,
~
FOR OFFICTAL USE ONLY
APPROVED FOR RELEASE: 2007/02/08: CIA-RDP82-00850R000300080043-3

APPROVED FOR RELEASE: 2007/02/48: CIA-RDP82-00850R000300084443-3
, FOR OFFICIAL USE ONLY
The spectrum of the output signal in the case of linear transformation is expressed
by the function: _
_ S~ ; I m(1~) IY Saa .
where ~(,jw) is the system's frequency characteristic,.
The reciprocal spectrum Sgx(w) can be readily found if we write down the output
of linear transformation of the system in the form of Duhamel's integral:
~
X(t) = f k(T) g(t - T) dT, �
0
where k('[) is the pulsed characteristic of the system in question. By multiplying
the left and right parts of this equation by the signal g(t), averaging a~d taking
the Fourier transform, we shall obtain the reciprocal spectrum Sgx(w) in the
following form:
. S~(~) =~(~1~)S~a(~)�
Substituting the found values in the equation for coherence function, we find that
for a system that performs linear transformation the coherence function equals one. -
In order to find the zone where ordinary linearization of performance is possible
in the mode of compensatory tracking, we plotted the function of coherence between
the error presented to an operator on the display and his motor reaction.
In our experiment, we found that the coherence function is close to one only when
_ the frequency spectrura of the input signal does not contain frequencies above 0.5 Hz.
The presence of higher frequencies lowers the coherence function and, consequently,
the operator p erforms nonlinear transformation of the input signal. Thus, for
input signals containing no frequencies above 0.5 Hz, the operator controlling the _
object of control described by an amplifying unit can be formally identified as an
image function.
Figure 4 illustrates the spectra of operator error in the course of training in
tracking. Figure 4 shows that new elements appear in the spectrum of operator
actions with increase in number of exercises. A change in contr~l situation,
parameters of input signals, parameters or structure of the control object leads to
a change in these components.
Analysis of the spectrum of reactions of a trained operator shows that, in addi-
tion to the required ha~onics, it contains several other harmonic components and
a random process, which are not provided by the objective of a given activity and
are an error for the man-~machine system. Such movements, which are an error from _
the standpoint of efficient operation of the system, are at the same time necessary
9
FOR OFFIC'TAi. USE ONLY
APPROVED FOR RELEASE: 2007/02/08: CIA-RDP82-00850R000300080043-3

APPROVED FOR RELEASE: 2007/02148: CIA-RDP82-00850R040340080043-3
_ FOR OFFICIAL USE ONLY
to formation of an image of the controlled ob~ect, and they give the operator in-
- formation about the status of the ob~ect of control. In order to obtain the
~ecessary information to f orni adequate controlling actions, the operator must make
additional motions that diminish the accuracy of operation of the man-machine sys-
tem but permit perforatance of the tracking process.
Sw � Sw ~
1 1
' ~ -o~ Z r~-o,~ Z :
- ~i w` -
~ . � ' .
. S~ ' ~ �S~
> ~ .
- ' 2 Hz q2Hz . ~ ,
, .
~i wi
_ S~ S~
1
0,3 z 3 Z
~i ~i
0 D, Z 0, ~ 0,6 0, B! 1, 2~ 0 0, Z 0, 4 0, 6 0, 9! 2
Start of training End of training
- Figure 4. Sw--standardized spectral density of delta remnant
Wi--current frequency, Hz
f~--frequency of delivered signal, Hz _
This spectrum of additional movements in relation to ideal control is needed by the
operator to "learn" and control the control process was named the delta remnant, but
only ir_ the case where the transfer function of the ob~ect of control equals one
and the input signal is a harmonic. This definition is convenient in that a change
in any factor affecting the quality of tracking, for example, parameters of the
- object of control, characteristics of exogenous factors, changes in psychophysiolo-
gical state of the operator, etc., leads to appearance of new qu~..~itative and
qualitative changes in the spectrum of the tracking process, and these factors can
be evaluated as characteristics of the spectrum of additional movements.
The informative nature of additional tracking movements was determined experimen~ally.
Dynamic inerti~l and oscillatory elements were connected into the control circuit,
and they constituted filters of specific frequencies. Their filtering properties
lU
FOR OFFICTAL USE ONLY
APPROVED FOR RELEASE: 2007/02/08: CIA-RDP82-00850R000300080043-3

APPROVED FOR RELEASE: 2047/02/08: CIA-RDP82-00850R000300080043-3
- FOR OFFICIAL JSE ONLY
- ~,re determined by the values of the param~ters. Tnclu~ion of these elements in
th~e circuit, which simulate the propertiea oP the ob~ect of control, rendered opera-
tor work aubatantially more difficult; there was a drastic increase in tracking
error and disruption occurred sooner. It was faund that if such a filter element
- "cut off" part of the spectrum of reactions that contained information for the
operator there was a drastic deterioration of quality of parformance. On the other
hand, addition to the circuit of control of an elemant generating additional har-
= monics (nonlinearity) also increased the number of additional operator actions and
thereby worsened tracking quality. Thus, according to our data, the operator uses
information that he provokes hiraself in order to form controlling actions in track-
ing. A decrease in such information due to filtration or "enric~ment" thereof due
to generation of new harmonics worse.ns operator performance.
The results of experimental analysis of operator performance referable to compen-
_ s3tory tracking disclose the sort of dual nature of additional movements. On the
one hand, they constitute a tracking error, i.e., they are a negative factor; on the
other hand, a certain part of these movements, which is inf ormative in nature and
a manifestation of specific operator activity, is needed for successful tracking.
It is important to recall that, from the standpoint of regulation theory, minor
additional movements contained in the error do not play a role in the tracking pro--
cess, since they are not controlling. For expressly this reason, they were not
considered in creating mathematical models.
Determination of the informative nature of errors made it possible to consider the
different aspects of operator performance of the tracking type from a d3fferent
angle.
In the study of the process of acquiring tracking skill, dete~ination was made of
the coherence function, which characterizes the type of sensorimotor transformation
_ of information delivered to the operator's visual analyzer. With increase in
number of practice exercises the coherence function diminished, and it became
constant by the end of the training period.
This shows that formation of tracking skill is related to formation of informatienal
movements that distort the linearity of sensorimotor transformation. It was demon-
strated that the more training an individual has had, the less~l~~.near h3s trans-
_ formation of input signal into an output signal. This finding, which appears para-
doxical at first glance, indicates that there is reorganization of processes of
receiving and processing information by an operator in the course of training.
Thus, we consider minor additional movements as necessary elements of a mathematical
model of a man-machine system, which are determined by the distinctions of inental
regulation of human actions.
Development of a linear model requires that these movements be disregarded because
of their smallness; however, if they are overlooked there is loss of information
about psychological regulation of activity. Slight movements are one of the
indicators of psychological distinctions of operator work in the tracking mode.
The lack of formalized description of the properties of these movements in mathema-
tical models of performance is the cause of their inadequacy. Tnclusion thereof in
_ mathematical models makes it possible to take into consideration the psychological
distinctions of human performance. This proves the usefulness and applicability of
11
FOR OFFICIAL USE ONLY ~
APPROVED FOR RELEASE: 2007/02/08: CIA-RDP82-00850R000300080043-3

APPR~VED FOR RELEASE: 2007/02/08: CIA-RDP82-00850R000300080043-3
FOR OFFICIAL USE ONLY
the proposed method of analyzing small additional movements to gain deeper understand-
ing of psychological distinctions of operator performance, and also offers new
quantitative criteria for evaluating this performance.
Since any system is developed for a certain set of people, of course the character-
istics of such movements are of a random nature from person to person, F~hile the
characteristics of a set of people that have to operate the system can be described .
by a certain law of distribution. The range of scatter in this set is set by the
tactical and technical specifications, and it determines the accuracy of a man's
work in this system.
The elements of the technical part of the system are always made with certain allow-
ances due to the inevitable technological f laws in production, which are of a random
nature. For this reason, the characteristics of the system as a whole are random,
and any man-machine system is stochastic.
Analysis of accuracy of the most frequently encountered block diagrams (structural
diagrams) describing a semiautomatic system, in the presence of scatter of para-
mete rs, makes it possible to solve the problem of defining requirements in common
for the scatter of machine parameters and functional characteristics of an operator.
By solving this problem, we can define the permissible scatter of parameters of the
mach ine part of the system and permissible scatter of functional characteristics of
man in accordance with the tactical and technical specifications made of the system
being developed. The permissible scatter of characteristics of operator performance
defines the ranges of parameters, in the presence of which the set of operators is
suitable fo r working in a given system. Otherwise,~the operator does not meet the
requirements for a given performance in a:vSL~m, which are imposed by the work
tools on the functional characteristics of n.~.n. ,
At the presenC time, these specifications are usually formed after the system has
already been developed. However, even at the design stages, it is imperative to
take into consideration the functional characteristics of a human operator, and
more over of an entire set of operators, rather than a single one, as they will be
oper ating the system in question. The early stage of system design is characterized
by the fact that there is neither an object of control nor set of operators whose
performance must be designed. The mechanocentric approach made it necessary to
first design the object of control, then to screen operators for this object that
have the required psychophysiological characteristics. The proposed systems
appr oach means that the object and human performance are designed from the same
~ vant age point and with the same specifications at the very earliest stages of
develc~pment oi mar~-rachine systems.
As a result of studying slight movements, it became possible to assess analytically
the share of error contributed to output signal er:or of the system attributable
to both the performance of the operator and scatter of parameters of any element
of the machine part of the system. This, in turn, made it possible to synthesize
a sy stem in accordance with specified tactical and technical reqL_'.-ements, i.e., to -
determine the permissible scatter of machine parameters and scatter of functional
- characteristics of operators accarding to known output characteristics of the system
determined by tactical and technical specifications.
The capabilities of the method developed,are to specify an allowance for scatter
of p arameters of any element of the system according to specified accuracy-related
12
FOR OFFICTAL USE ON'LY
APPROVED FOR RELEASE: 2007/02/08: CIA-RDP82-00850R000300080043-3

APPROVED FOR RELEASE: 2007102/08: CIA-RDP82-00850R000300080043-3
FOR OFFICIAL USE ONLY ~
characteristics of the system's output signal and, in addition to 3ndustrial'.,
technological considera~tions,they should also be based on economic considerations.
on the basis of analysis of statistical data on system production and data per-
taining to the training of highly skilled operators, analytical functions were
obtained of cost of production of system elements (and the entire system) as re-
lated to precision of production and expenses for operator training as a function
of qualification requirements. These functions were named cost functions.
- Cost functions define the restrictions imposed by economic considerations on the
production of semiautomatic systems. The range of permissible parameters estab-
- lished as a result of system synthesis makes it possible to vary the garameters
of the nachine part of the system and functional characteristics of operators in
accordance with the nature of changes in cost functions referable to production of
_ its elements and operator training. Use of cost functions in the practice of
design makes it possible to create economical systems.
Thus, consideration of design of performance using the systems approach makes it
possible to analyze from the same [common] positions the exogenous manifestations
of human performance and operation of the *_echnical part of the system, as well as
to study the mechanisms of formation of subjective reflection of the states of the
object of control. _
BIBLIOGRAPHY
1. Berg, A. I. "Cyberne~ics---a Science Dealing With Optimum Control," Moscow,
_ 1964.
2. Blak'yer, 0. "Analysis of Nonlinear Systems," Moscow, 1969.
3. Biquet, J. "The Human Operator in Control Systems," in "Sovr.emennaya teoriya
sistem upravleniya" [Modern Theory of Control Systems], edited by K. M. Leondes,
1970, pp 454-485.
4. Lomov, B. F. "Man and Machine," Moscow, 1966.
5. "Mater. V Vsesoyuz. kongressa po fiziologii truda" [Proceedings of 5th A11-
Union Congress on Industrial Physiology], Moscow, 1967.
_ 6. Lomov, B. F..(editor) "Fundamentals of Engineering Psychology," Moscow, 1978.
7. Petrov, B. N.; Tubinskiy, A. I.; and Ryl'skiy, G. I. "Problems of Eff iciency
and�Reliability of Man-Machine Systems in General Control Theory," in
"Materialy IV Vsesoyuznogo simpoziuma po effektivnosti i nadezhnosti sistem
chelovek-tekhnika" [Proceedings of 4th All-Union Symposium on Efficiency and
Reliability of Man-Machine Systemsj, Scientific council for the complex problem
of "Cybernetics," Moscow--Leningrad, 1975, pp 2-7.
8. "Methodologiya issledovaniy po inzhenernoy psikhologii i psikhologii truda"
[Methodology of Research in Engineering Psychology and Industrial Psychology],
Moscow, Pt 1, No 4, 1974.
13
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/08: CIA-RDP82-00850R000300080043-3

APPR~VED FOR RELEASE: 2007/02/08: CIA-RDP82-00850R000300080043-3
FOR OFFICIAL USE ONLY
- 9. Sergeyev, G. A., and Romanenko, A. F. "Statistical Methods of Evaluating
Efficiency of Human Operator's Tranefer Function," VOPR. PSIKHOL. [Problems
of Psychology], No 4, 1965.
10. Taran, V. A., and Kofanov, Yu. N. "On the Question of Detertnini~g Operator
Transfer Function by Means of an Analog Camuuter," Ibid, No 3, 1969.
11. Trapeznikov, V. A. "Man in Control Systems," AVTOMATIKA I TELEMEKHANIKA
[Automati~n and Telemechanics], No 2, 1972.
12. Chernyshev, A. P. "On the Question of Operator Transfer Function," in "Mater.
V. Vsesoyuz. s"yezda psikhologov SSSR. Psikhologicheskiye problemy povysheniya
effektivnosti i kachestva truda'" [Proceedings of Sth All-Union Congress of
USSR Psychologists: Psychological Problems of Improving Efficiency and Quality
of Labor], Moscow, 1977, pp 63-65.
13. Sheridan, T. B. "Experimental Study of Changes in Time of Transfer Functions
of Human Operators," "Tr. I kongressa TFAK AN SSSR" [Proceedings of First
Congress of International Federation of Automatic Control, USSR Academy of
Sciences], Moscow, 1961.
14. McRuer, and Krendel E. S. WADS TECHNICAL REP., 56-524.
- 15. Elkind, J. I. "Technical Rep.,F'��No TII, Lincoln Lab., Massachusetts Inst.
Technol., 1956.
16. Diamantides, N. D. "Man as a Link i:l 3.:ontrol Loop," ELECTRO-TECHNOLOGY,
Vol 69, No 1, 1962, pp 361-371. ~ ~
17. Sheridan, T., and Ferreff, W. "The Man-Machine Systems Information Control
and Decision Models of Human Performance," Camt ridge (Mass)--London, 1974.
COPYRIGHT: Izdatel'stvo "Nauka", "Psikhologicheskiy zhurnal", 1980
- [92-10,657] -
,
10,657
C50; 1840
~l~
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/08: CIA-RDP82-00850R000300080043-3

APPR~VED FOR RELEASE: 2007/02/08: CIA-RDP82-00850R000300080043-3
FOR OFFICIAL USE ONLY .
, COLLABORATION OF SOCIALIST NATIONS IN THE FIELD OF ENGINEERING PSYCHOLOGY
Moscow PSIKHOLOGICHESKIY ZHURNAL in Russian No 5, 1980 pp 146-152
- [Article by V. F. Venda]
_ [Text] Spec:Calists of social countries have ceployed in recent years research in
the field of engineering psychology and have made considerable strides in a number
of new directions, in implementation of the decision of national congresses of
communist and worker parties pertaining to development of automated control systems, _
improving the efficiency, quality and safety of labor.
Attention was concentrated mainly on refining methods of psychological analysis of �
the structure of work activity, evaluation of its intensity, development of theory _
and pract3cal methods of improving the quality and efficiency of controlling machines,
units, industrial, transportation, energy, aviation and space equipment, by optimiz-
_ ing interaction between man and machine. The recommendations to take into consi-
deration the psychological distinctions of performance are used extensively in pro-
cesses of designing, operating and rationalizing machines, systems and control
equipment. At the present time, work on many promising systems is starting with
preparation of an engineering psychological plan for operator performance, definition
of principles and methods for operator screening and training.
It is becoming possible to solve such a complex problem as mutual adaptation of
a human operator and control equipment thanks to the intensive development of the '
systems approach in psychological research. In this regard, the work of B. F.
Lomov [1], V. P. Kuz'min [2], A. A. Krylov [3], D. Kovac [9], J. Daniel [10],
F. Klix [12], W. Hacker [13], N. D. Naplatanov [14], Yu. P. Marinov [15] and other
scientists from socialist countries played a fundamental part.
~ The systems approach made it possible to refine engineering psychological design
and rationalization of modern control equipment, as well as to undertake developmen~t
of forecasts of promising structures of man~nachine systems.
At the same time, the practice of building of communism and socialism poses more
and more complex problems to specialists in the field of engineering psychology.
They can be solved, provided the efforts of special3.sts of socialist natir,ns, '
CEMA members, are integrated.
One of the most pressing theoretical and practical directions is development of
engineering psychological requiremente for equ3pment that displays information to
15
FOR OFF7~TAT, USE ONLY
APPROVED FOR RELEASE: 2007/02/08: CIA-RDP82-00850R000300080043-3

APPROVED FOR RELEASE: 2007/02108: CIA-RDP82-00850R000300080043-3
~
FOR OFFICIAL USE ONLY
operators, including equipment for output of data from computers. This direction _
was approved as CEMA pro~ect 1-37.IV, work on which is being pursued in accordance
with the 1976-1980 program for multilateral collaboration.* Specialists from the
People's Republic of Bulgaria, GDR, Poland, Czechoslovakia and the Soviet Union are
working on this project. The Institute of Psychology, USSR Academy of Sciences,
was assigned to implement scientific coordination of work on this proiect. _
The research program for pro~ect 1-37.IV specifies three tasks. The first is
"Development of theoretical bases of engineering psychological design of equipment
for displaying [reflecting] information."
This task involvec3 development of a classification of types of information display
equipment (IDE), which was comuleted by the Industrial and Scientific Research
Laboratory of the Higher Mechanics and Energy Institute (Sofia, People's Republic
of Bulgaria) headed by Yu. P. Marinov; the results of these studies have been pub-
lished [15]. Within the context of the same assignment, the Institute of Psychology,
USSR Academy of Sciences, is studying determinatian of processes of solving opera-
tional problems by the type and structure of IDE, as a result of which there was
formulation of the structural psychological conception of analysis and synthesis of
IDE j4], which was discussed at a scientific coordinating meeting dealing with
project 1-37.IV in Moscow, in 1978, and which was approved as the main theoretical
conception.
This conception is a special offspring of the systems approach to engineering psy-
chological studies, which was developed by B. F. Lomov [1]; here, the structure of
the information display system is relate: *o the values of psychological factors of
difficulty of solving operational problems.
On thia bases, the ob~ective of engineering psychological design is to aid in opti- .
mizing psychological factors of difficulty jcomplexity] (PFD). In the course of
the scientific coordi_nating meeting, it was noted that this approach to engineering
psychological design of IDE has some advantages over otl-~r conceptions in engineering
psychological design. In particular, it takes into coi~sideration the many variants
of processes of solving operational [immediate, ongoing] problems, and it interprets
overtly the principles of "assembly," "organization," "structuring" of information
display systems as means of materialization in their structures of a priori solution
strategies; it permits addition and analysis of a'~quantitative gage of determination
of real strategies by a priori ones.
The systems approach makes it possible to view problems of engineering psychological
design not only as preliminary adaptation of a machine to man, but on a nuch broader
scale, as multistage, multilevel adaptation, both reciprocal and related: man to
machine and machine to man. As a reault, broader self-organization properties are
imparted to the system.
Use of inethods of analysis and organization of interaction between a priori and
actual strategies is noC limited solely to instances of engineering psychol~gical
design of opera~or performance. This approach is also used to solve other ~roblems ~
*This pro~ect is part of the work dealing with CIIMA problem No 1-37,~"Development of
scientific bases of ergonomic standards and requirements." A coordinating council
for this problem was established at the All-Union Scientific Research Institute of
- Esthetics in Engineering, and this council is headed by V. M. Munipov, cardidate
of psychological sciences.
16
FOR OFFIC'_'AL USE ONLY
APPROVED FOR RELEASE: 2007/02/08: CIA-RDP82-00850R000300080043-3

APPR~VED FOR RELEASE: 2007/02/08: CIA-RDP82-00850R000300080043-3
FOR OFFICIAL USE ONLY
of design of man-~machine and man-computer systems, for example, to determine the
desirable level of automation of experimental scientific research. Tn the case
where a priori strategy is taken as the basis of some paradigm of research and it
must be executed many times to accumulate homogeneous experimental data, i.e., when
the a priori strategy can be repeated without any creative intervention on the part
of experimenters in the course of the experiments, there can be complete automation
of research. But when the experimenter actively influences refinement of inethods
(i.e., a priori strategies) of research, as is usually the case in modern applied -
psychology, the experimental device must be designed on the basis of adaptive
informational interaction between man and machine.
Organization of adaptive interaction of a priori and real strategies is the road
toward more efficient use of the capabiliti.es of modern information and computer
technology, as well as definition of the prospects of its development to create
more sophisticated man-machine systems.
In the future, the basis may be formed for synthesis of "hybrid intelligence" systems
[16, 17] on the basis of the principles of organization of adaptive informational
interaction between complexes of a priori and real strategies offered by information
and computer technology and active participants in the solutions [decisions].
The Leningrad State University, All-Union Scientific Research Institute of Esthetics
in Engineering (Leningrad branch), Leningrad Electrical Engineering Institute,
Scientific Research Institute of Peripheral Equipment (Kiev) and the Central Scienti-
fic Research Institute of Complex Automation are working with the Institute of
Psychology, USSR Academy of Sciences, on development of theoretical and methodological
- bases of engineering psychological desj_gn of IDE. The work of these organizations
is related to the study and refinement of languages for dialogs between man and com-
puters, studies of ~ension and fatigue of individuals working at a console with a -
screen (display), and development of inethods for engineering psychological design
of automated control systems.
The intensive work being done on methodological bases of engineering psychological
analysis and design of man-machine systems, with the participation of many scientists
and specialists, made it possible to create an hierarchy of approaches and methods
that permit analysis of such systems on different levels to varying degrees of
detail.
The highest level invol.qes the structural systems approach [1]. This is followed by
analysis of psychological factors of difficulty of reaching goals, problem solving
by man. Such analysis is based on the structural psychological conception [4]. If
the problem-solving process and performance .of functions by man can be presented
in the form of a stable block diagram, analysis of efficiency and reliability of
the system is made on the basis of the generalized structural method developed by
A. I. Gubinskiy [S].
When it is possible to provide not only a block description, but one of each
operation, one after the other, in the processes of operator work, one uses algo-
rithm methods, which have been studied comprehensively in application to man-
machine systems by G. M. Zarakovskiy and A. I. Galaktionov [6]. In the general
case, block and operation analysis of performance are used to define certain PFD,
such as the number of operations in the algorithm o� ~iecision making and its -
implementation.
17
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/08: CIA-RDP82-00850R000300080043-3

APPR~VED FOR RELEASE: 2007/02/08: CIA-RDP82-00850R000300080043-3
FOR OFFICIAL USE ONLY
The use of simulation models 3s succe~sful to some degxee when the operations per--
formed by humans, both individually and as a group, in the course of ~oint control
of the system are known.
When analyzing the dynamics of deterministic man~machine systems, a systemat3c
refinement of description thereof is achievPd by turning to discrete, pulsed and,
finally, analogous (continuous) mathematical models. This level corresponds to the
most comprehensive and strict description of work processes, which is applicable
to a relatively limited range oF well-known functions of an operator, such as compen-
satory tracking, for example [6]. Studies must be made on each level of analysis~of
the possibility of optimizir.g the system c:a the basis of reciprocal adaptation of
- man and machine.
The results of studies dealing with the theoretical aspects of engineering psycholo-
gical design of IDE are summarized in the following books: "Engineering Psychology.
Theory, Methodology and Practical Applications" ("Nauka," Moscow, 1977) and "Psycho-
logical Problems of Mutual Adaptation of Man and Machine in Control Systems." The
second book is ready for publication by "Nauka" Publishing House, and its authors
include scientists f rom the People's Republic of Bulgaria, GDR, USSR and CSSR (the
editors in chief are B. F. Lomov, V. F. Venda and Yu. M. Zabrodin).
The second assignment in pro~ect 1-37.IV is "Engineering Psychological and Ergonomic
Studies of Information Receiving and Processing as Related to Differ~nt Features of
IDE." These studies are coordinated by the Institute of Labor Safety (Prague, CSSR), -
and the following are participants: Institute of Experimental Psychology (Bratislava,
CSSR), Institute of Engineering Cybernetics, Bulgarian Academy of Sciences (So�ia),
Center for Esthetics in Industry and Artis:~c ~Psign (Sofia), Leningrad State Uni-
- versity (USSR) and Institute of Esthetics il: Engineering (Warsaw, Polish People's
Republic).
Research on preparation of information for decision nakiiig when working with various
code symbols, studies of the principles of multidimensional coding of visual and
auditory information are being pursued at Leningrad State University, under the
guidance of Prof A. A. Krylov; together with this university, the Tnstitute of
Psychology, USSR Academy of Sciences is studying information Yeceiving and processing
in the case of simultaneous delivery thereof via visual and auditory channels. K.
Lapa:hevska and D. Senk, at the Institute of Esthetics in Engineering (Warsaw) are
stud;!ing the structure of the visual field as related to intensity of flow of
infc,rmation.
Extensive studies are being pursued by J. Daniel and his coworkers at the Institute
of Experimental Psychology (Bratislava, CSSR) in the area of information receiving
and processing as related to change in intensity of delivery thereof to IDE; special
emphasis is laid in this work on studies of the influence of information load on
cognitive processes when introducing automated control systems and computers.
At the first stage of this research, J. Daniel concentrated mainl; on determination
of the main parameters of physical and mental tension and study of perceptual pro-
cesses in the presence of a load. Catecholamine levels were chosen as an indicator
of the load. It was shown that individuals with good xesistance to loads (both
physical and mental) secrete more epinephrine in experiments. At the second stage,
adaptive mechanisms associated with long-term exposure to loads were the main topic
of studies. It was established that, after an anticipation phase characterized by
18
FOR OFFIC'AL USE ONLY
APPROVED FOR RELEASE: 2007/02/08: CIA-RDP82-00850R000300080043-3

APPR~VED FOR RELEASE: 2007/02/08: CIA-RDP82-00850R000300080043-3
. FOR OFFICIAL USE ONLY
.
intensive output of epinephrine, these is a decline of catecholamine level with
further exposure to stressors, labor efficiency improves while the f eel3ng of
fatigue diminishes. A consistent link was demonstrated between the effect of
an information load and such personality trzits as level of anxiety, neuroticism,
integration of Fersonality and strength of nervous system (according to the teets
of Ayzenko, Strelau and Mikshik queationnaires).
The interdisciplinary approach to analyais of work processes, which is being deve-
loped at the Institute of Experimental Psychology (Bratislava) under the guidance
of D. Kovac, made it possible to combine previously diff erent levels of struc-
tural-functional, psychological, psychophysiological and physiological analysis. '
Use of this systems methodology yielded a number of new findings and results in
studies and rationalization of various industrial occupations. Special mention
must be made of the significant contribution of M. Strizenec [11] to the complex
psychological work on the problem of man's interaction with computers.
Extensive studies of cognitive processes that regulate work performance are being
conducted by Prof W. Hacker at Dresden Polytechnical University (GDR). He nses
a combination of inethods: analysis of production process, observation of work pro-
cesses on different levels of differentiation and field (natural) exp eriments.
This author demonstrated that the separate, unrelated use of any of these proce-
dures alone is ineffective.
Modern ana].ysis of work processes is based on principles of quantitative analysis
of psychological regulation and psychodiagnostic principles of demons tration of
stable prerequisites (abilities) for a given type of work. ':T. Hacker demonstrated
that psychological analysis of work processes must include consideration of their
processual and structural elements in order to demonstrate and upgrad e the
control and effector elements of performance. There must also be analysis of
~oal formation, specifics of recognition and differentiation of state s. Special
attention must be given to diagnostics [identif ication], i.e., classification and
evaluation of states, transformation of graphic and conceptual (logical) elements
of regulation, including the decoding of information, recognition, mathematical and
logical operations, conversion of reference systems, etc. The study of "managerial
knowledge," system of immediate [operational] patterns, actualization thereof when
searching for solutions, processes of formation of new work proc.edures, prediction
of possible consequences, performance of monitoring and effector sensorimotor
- operations constitute an independent group of problems.
N. D. Naplatanov, Yu. P. Marinov, P. Khadzhiyev, K. Tropolov and other scientists
of the People's Republic of Bulgaria are elaborating principles of IDE synthesis
intended for the control of stationary and moving objects, on the bas is of the
methods of psychological and psychophysiological analysis of work activity, which
are being developed by the scientists of CSSR and GDR, within the f ramework of the
second assignment contained in pro~ect 1-37.IV.
_ N. D. Naplatanov et al. [14] demonstrated that man~machine systems have many circuits
and many channels; they contain internal links that are extensive in nature and
goals, elements of a rather high hierarchic level of organization and high-density
flows of information circulating over the external circuits of the sy stems. For
this reason, studies, deaign and construction of such systems should take place as
a functional entity by meana of inethods used for the study of cybernetic systems,
regardless of the fact that the main units are governed by their own specif ic
19
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/08: CIA-RDP82-00850R000300080043-3

APPR~VED FOR RELEASE: 2007/02/08: CIA-RDP82-00850R000300080043-3
FOR OFFICIAL USE ON'i,Y
principles and patterns in thetr function. These studies are based on the systems
approach. It consists of complex, interrelated and proportionate consideration of
all factors, ways and means of solving a complex, multifactorial and multivariant
problem, a typical example of which are problems related to the study of man~nachine
control systems. To solve these problems, there must be close correlation of a wide
range of scientific and practical data, as well as large material-technical and
labor resources, flexible enough and suitable for the relevant level of adaptation
of forms of organization. The systems approach consiclers control systems as an
integral whole, with all its aspects, proportionately to their weight and signifi-
cance.
Uniike classical engineering psychological design, with the use of the systems
approach all factors of the system being designed are taken into consideration
ti~nal, psychological, social and esthetic), which leads to appearance of r.ew phases
in the designing process and, first of all, in engineering psychological design.
N. D. Naplatanov lists the following among the main principles involved in design-
ing man-machine systems.
1. Proportionate and successive adherence to optimum factor correlations
between the main elements of the system in a dynamic mode.
_ 2. Redistribution of volume and nature of functions among the main elements
of the system.
3. Provision for mutual coordinat 1 of main parameters and elements of the ~
- system in the information, ps~~cno_hy~iological, and economic-technical
aspects.
' 4. Dynamic regulation of level of centralized control.
The possibility of taking into consideration contradictc+ry factors and requirements
referable to different elements and subsystems is an i*nportant distinction of the
systems approach. One proceeds from the goal-or3ented function of the system in
solving this problem (which is, in principle, mu~ticriterial and multivariant).
Sin e computers were designed to help man, development of dialog between man and
com~~~lters results in more natural forms of comm~unication in man-machine systems,
use of standardized, free language forms (algorithm languagues) tliat are inputted
visually, verbally or by touch. `
The systems approach to psychological analysis of interaction between man and
machine, methods of studying the psychological structure of operator performance
and the structural-psychological conception of analysis and synthesis of IDE con-
stitute the theoretical foundation for preparing practical recommendations on TDE
design. Applied work is being pursued in accordance with the third assignment of
CEMA pro~ect 1-37.IV.
This includes development of inethods for engineering psychological design of IDE -
for technological automated control systems of different levels and sectors of
industry. T~is work was assigned to the Central Scientific Research Institute of
~U
FOR OFFICTAL USE ONLY
APPROVED FOR RELEASE: 2007/02/08: CIA-RDP82-00850R000300080043-3

APPR~VED FOR RELEASE: 2007/02/08: CIA-RDP82-00850R000300080043-3
FOR OFFICIAL USE ONLY
_ Complex Automation (Moscow). Extensfve work is bei,ng done at ~ne Leningrad Electrical
Engineering Institute on refinement of symbol systems used on displays and methods
of training operational personnel to work with displays. Studies of procedural and
rhythmic-time structure of dialog interaction between man and computers occr,py a
significant place; these studies are being conducted by the State Construction
Office for the Design of Calculators (Leningrad).
Engineerir.g psychological requirements pertaining to means of output of data from
YeS computers were formulated as a result of the work done by the Leningrad branch
of thp All-Union Scientific Research Institute of Esthetics in Engineering. In the
future, these findings, along with the results of studies pursued at the Higher
Mechanics and Energy Institute (Sofia, People's Republic of Bulgaria) and Institute
of Psychology, USSR Academy of Sciences, could be used as material for the CEMA
- s tandard. _
Some of the results of studies pursued on project 1-37.IV have already found prac-
tical applications.
The Institute of Psychology, USSR Academy of Sciences and All--Union Scientific Re-
search Instit,~'te-of Esthetics in Engineering have developed and introduced recom-
mendations dealing with organization of the labor of dispatch~xs of the Ura1
Unified Power System. The technical and economic effect from adoption of the
f~rst phase of the engineering psychological design constituted about 150,000 rubles
per year. The expected effect of making ~ull use of the results of research will
be over 500,000 rubles per year. The Central Scientific Research Institute of
Complex Automation, Institute of Psychology of the USSR Academy of Sciences and
Institute of Automation (Kiev) have developed and introduced several engineering
psychological designs of operator stations at electric power plants: TETs-21 of
Mosenergo [Moscow Regional Administration of Power System ManagementJ, Staro- ~
Beshevskiy GRES, Chigirinskiy GRES. The overall economic effect of this work
constituted more than 1 million rubles per year.
The All-Union Scientific Research Institute of Esthetics in Engineering and Insti-
tute of Psychology, USSR Academy of Sciences, developed an experimental mockup of
a console for the traff ic control system in Moscow, which is being introduced in
our country for the first time. The expected effect of fo_llowing engineering psy-
chological recommendations in this system will be about 300,000 rubles per year.
The high efficacy of making practical use of engineering psychological recommenda-
tions dealing with organization of labor in wider scale occupations, such as
instrument control workers referable to chemical units, lathe operators, assembly
_ workers, has been convincingly demonstrated on the example of the work of Yaroslavl'
StaCe University at the Yaroslavl' Tire Plant, that of the Al1-Union Scientific
Research Institute of Esthetics in Engineering at the Shchekinskiy Chemical Combine,
that of Moscow State University at the Moscow Transformer Plant, and others. The
time within which the engineering psychological designs were reimbursed averaged
- about 6 months.
_ The scientists of the People's Ftepublic of Bulgaria, GDR, Polish People~s Republic -
and CSSR have made considerable strides with respect to introducing the results of
engineering psychological research int~ the national economy of their countries. Tn
- particular, Polish psychologists and ergonomists have made a large contribution to -
improvement of production by the instrument-~making industry; Bulgarian scientists
_ 21
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/08: CIA-RDP82-00850R000300080043-3

APPROVED FOR RELEASE: 2007/02/48: CIA-RDP82-44850R000300084443-3
FOR OFFICIAL USE ONLY
developed several engineering psychological designs for man-~machine systems in the
field of transportation. The scientists a:~d specialists of GDR and CSSR did much
in the area of psychological analysis and rationalization of labor in enterprises
of the chemical, textile, wood working industries, as well as in machine building -
and energy.
Practical use of the results of engineering psycho logical research, made in the
course of work on project 1-37.IV, shows that the economic effect from introd~iction
thereof constitutes an avera~e of 80,000-150,000 rubles per year per technological _
automated contro:I system. If we consider that there will be several thdusand such
systems by 1985, the overall annual effect of using the results of psychological
research could amount to hundreds of millions of rubles.
Engineering psychology is called upon to play an important role, not only in
refining control of individual technological units, complexes and enterprises, but
sectors of industry; in other words, in extending the knowhow accumulated on the
example of technological automated control sysCems to large-scale organizational
automated control systems.
In a modern organizational automated control system the daily intensity of the flow
of processed information constitutes several million symbols. For a number of
such systems, engineering psychological recommendations have been developed and
introduced to organize and refine the processing of these enorcnous flows of in-
formation.
Work dealing with preparations for changin~ `rom mechanized to automated processing
of documents without loss of information e~ :ential to making managerial decisions
is of special importance [8].
Thus, engineering psychology is involved in solving one of the most important prob-
lems of building of communism and socialism, that of improving the immediacy of
control of the national economy, growth of efficiency a~3 quality of labor.
The experience of working on CEMA project 1-37.IV, "Development of Engineering
Psychological Requirements Referable to Equipment for Displaying Information to
Ope'ators," has demonstrated the great effectiveness of integrating the efforts
_ of :~cientists and specialists of socialist nations in the area of engineering
psy.�~iology.
It is imperative to provide for an even wider scale of such successful collaboration
in this field in the next five-year plan.
In our opinion, in planning future joint work, one should concentrate, first of
all, on continued development of Marxist-Leninist systems methodology, strengthen-
ing the theoretical bases af engineering psychology, preparing a long-term forecast
of development of this branch of science under socialism. It is important to refine
methods of psychological analysi s of work processes, to work on pioblems of multi-
level adaptation of man and machines in control systems, psychological principles of
transmitting information to operators and effective use of computers in cognitive
processes and control. Optimization of informatio n-related interaction between
people who make joint decisions for the purpose of intensification of solutions
of complex intellectual problems in the area of control [management), science and
technology constitutes a special group of theoretical and applied problems. -
22
FOR OFFICTAL USE ONLY
APPROVED FOR RELEASE: 2007/02/08: CIA-RDP82-00850R000300080043-3

APPROVED FOR RELEASE: 2007/02108: CIA-RDP82-00850R000300080043-3
FOR OFFICIAL USE ONLY
It would be desirable to integrate work in the area of training specialists in
engineering psychology, including scientist researchers of the highest qualif ication
an~ administrators of engineering psychological designs.
Joint development of designs of psychological services in socialist countries will
make it possible, in the future, to coordinate more efficiently the recommendations,
- norms and standards pertaining to psychological aspects of labor safety, design of ~
machines, components, instruments, computers, control systems, as we11 as to
improve within a very short period of time the supply of engin.eering psychalogy and
ergonomics laboratories with the latest, most refined equipment.
In addition to solving new problems, even broader introduction of the results of
joint studies pursued in the years of the current 5-year plan into the national
economy of socialist countries may constitute an important objective of multilateral
collaboration in 1981-1986.
BIBLIOGRAPHY
1. Lomov, $.~F._--'"The Systems Approach in Psychology," VOPR. PSIKHOL. [Problems
of Psychology), No 3, 1975.
2. Kuz'min, V. P. "The Systems Principle in Theory and Methodology of K. Marx,"
Moscow, 1976.
3. Krylov, A. A. "Man in Automated Control Systems," Leningrad, 1975.
4. Venda, V. F. "Multivariant Nature of Decision Processes and the Conception
of Engineering Psychological Design," in "Inzhenernaya psikhologiya. Teoriya,
metodologiya, prakticheskoye primeneniye" [Engineering Psychology. Theory,
Methodology and Practical Applications], Mosrow, 1977.
5. Gubinskiy, A. I., and Yevgrafov, V. G. "Ergonomic Design of Mari~irne Control
Systems," Leningrad, 1977.
6. Galaktionov, A. I. "Engineering Psychological Design of TP* Automated Control
Systems," Moscow, 1977.
7. Bodrov, V. A.; Zazykin, V. G.; and Chernyshev, A. P. Compensatory Tracking
of Harwonic Si nal " in "Inzhenerna a sikholo i a. Teoxi a metodolo i a
g ~ Y P g~y y~ g Y~
prakticheskoye primeneniye," Moscow, 1977.
8. Nikolayev, V. I. "Engineering Psychological Aspects of Constructing Man-
Machine Complexes," Ibid.
9. Kovac, D. "K integracii v psychologii. Psychodiagnosticke a didakticke testy,"
Bratislava, 1975.
10. Daniel, J., et al. "Psychologicka analyza cinnosti operatora," Bratislava, 1975.
1975.
*Translator's note: TP could refer to Freight Train or Commercial Port.
23
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/08: CIA-RDP82-00850R000300080043-3

APPROVED FOR RELEASE: 2007/02108: CIA-RDP82-00850R000300080043-3
FOR OFFICIAL USE ONLY
11. Strizenec, M. "Clovek a pocitac," Bratislava, 1978.
12. Klix, F. "Infozination und Verhalten," Berlin, 1972.
13. Hacker, W. "Allgemeine Arbeits- und Ingenieur-Psychologie," Berlin, 1978.
- 14. Naplatanov, N.; Marinov, Yu.; and Khadzhiyev, P. "Nov pc~dkhod za izgrazhdane
- na ergonomichni du-ergatichni sistemi za iz ledovane, klasifikaksiya i
prognozirovane s"stoyanito na choveka," in "Kibernetichen aspekt na ergonomiyata"
[Cybernetic Aspect of Ergonomics (in Bulgarian) Sofia, 197r�.
15. Marinov, Yu. "Information Characteristics of Human Operator" ~in Bulgarian],
Sofia, 1978.
16. Lomov, B. F., and Venda, V. F. "Human Factors: Problems of Adapting Systems
for the Interaction of Information to the Individual," in "Proc. of the HFS
21st Ann. Meeting," San Francis co, 1977, pp 1-9.
17. Idem, "Methodological Principles of Synthesis of Hybrid Intelligence Systems,"
in "Proc. Intern. Conf. on Cybernetics and Society," Tokyoq 1978.
COPYRIGHT: rzdatel'stvo "Nauka", "Psikhologicheskiy zhurnal", 1980 '
- [92-10,657]
10, 657
CSO : 1840
z4
FOR OFFIC.TAL USE ONLY
APPROVED FOR RELEASE: 2007/02/08: CIA-RDP82-00850R000300080043-3

APPROVED FOR RELEASE: 2007102/08: CIA-RDP82-00850R000300080043-3
FOR OFFICIAL USE ONLY
PHYSIOLOGY
~ ,
PATTERNS OF PERCEPTION OF VISUAL SIGNALS
Moscow PSIIQiOLOGICIiESKIY ZHURNAL in Russian No 5, 1980 pp 60-74
[Article by A. N. Lebedev, aubmitted 11 Oct 79]
[Text] Formulation of the Problem
The perception process is inconceivable without operations of comparing presented
cionals to those stored in memo~y. The hypothetical mechanisms of compariso~ a?-e
~o--
being studied in many laboratories. Schneider and Shiffrin [7] published the
most recent special survey of such studies. Ultimately, these authors proposed a
system of ec~uations that best explain the broad range of accumulated experimental
data, including their own, but yet did not cover all of them.
In the first place, they did not include in their system the link between duration
of perception and short-term memory span. As a result of analysis of numerous data,
Cavanagh [5] discovered a certain link of this sort.
In ~the second place, the above system of equations does not include the parameters -
of neurophysiological processes upon which perception is based. ~
Our objective was to fill this gap, compose a system of simple equations and find
a general~zation of the laiown patterns of perception.
First, we must mention the inherent distinctions of the experiments in question [4,
7]. A certain number of visual symbols is presented to the subject, for example,
digital ones, after instructing him to retain them for a short time. The sub~ect
himself limits the time of exposure of the symbols to be retained. Then, after a
warning signal, several frame [still, image?] or test symbols are presented
simul.taneously.
According t~ the instructions, the subject depresses one button, as fa.st as
possible, if the frame symbols do not i~clude a single labeled one, i.e., one
that was retained before, otherwise he depresses another button. The unlabeled
symbols are also called distractors. The situation is called negative if all of
the frame symbols are distractors. Otherwise the situgtdon is positive. Details
about these tests were described elsewhere [4, 7].
It is assumed that before pressing on a button the sut~ject successively compares
each labeled signal to each fraxne signal until he is sure that a11 of the frame
25
FOR OFFICIAL USE ~Ni,Y
APPROVED FOR RELEASE: 2007/02/08: CIA-RDP82-00850R000300080043-3

APPROVED FOR RELEASE: 2047/02108: CIA-RDP82-00850R000300080043-3
FOR OFFICIAL USE ONLY
symbols are diatractors (in a negative situation), or unti7. he detects a labeled ~
signal among the distractors (in a positive situation).
Comparison time as a function of number of frame and labeled symbols must be deter-
mined.
For th is purpose, we measure the time between presentation of frame symbols and
th2 subject's depression of the button, i.e., the discrete or latency period of the
reaction. Part of the latency period is unrelated to the number of frame and
labeled symbols, and it is called the constant, or nonreducible lag.
The rest of the time is a function of number af labeled and frame symbols. This is
the comparison time. The problem is to find a theoretical definition of the sought
fu.iction.
The simplest hypothesis is that comparison time is proportionate to the number N
of a11 necessary comparisons of frame and labeled symbols. Its lowest value equals
the product of number m of labeled symbols multipled by number K of frame symbols,
if each, as.it is generally believed, is successively compared to another, one by
one.
Taking these circumstances into consideration, we find that comparison time in a
negative situation is determined by the formula:
t = Nt ~1)
neg
and in a positive situation, by: '
t = tJ+l ~2~ i
pos 2
where t is the duration of one operation of comparing a frame and labeled symbol.
The hy~otheses concerning functions of types (1) and (2) were first expounded in
(6],
Schr~eider and Shiffrin [7] tested the above hypotheses in their experiments and,
st~.ict ly speaking, re~ected them, because the experimental values for comparison
time were above the minimum level predicted by formulas (1) and (2). For this
reason, they proposed their own system of equations. The following turned out to
be the most accurate:
;neg aniK -E- b (mK - K -f- l) -I- ~ (m!( - m 1) tn, (3a) (3a)
tpo~ a(m 2~- 11 + b rmK -2 21 + ~ i mK -2 -f- 21 + t P (3G) (3b)
. ) l 1 ~ 1
where a, b, c, tn and tP are constants found in the experiment.
Table 1, which is taken from [7], lists the values for the constants according to
the data of two teams of researchers. ,
'Lb
. ,
APPROVED FOR RELEASE: 2007/02/08: CIA-RDP82-00850R000300080043-3

APPR~VED FOR RELEASE: 2007/02/08: CIA-RDP82-00850R000300080043-3
~
FOR OFFICIAL USE ONLY
Taf~le 1. Constants in the equations of Schneider and Shiffr3.n
Source I t JConstaitsQ msl b I c ID~v~na .
Schneider, Shitfrin (?J 6?3 552 13 45 0 2l,3
Briggs, Johnsen (4] l 569 I 500 I 38 I 29 ` U 23,7
Judging by the low standard deviation of actual values from the predicted ones,
the equations are rather accurate, but they are cumbersome, contain many constants
and do not have clearcut neurophysiological. bases.
Expour.ded Hypothesis
Our solution wae based on known data [1, 2] concerning trQ relation of coherent,
narrow-band oscillations of neuronal activity to mechanisms of reflex activity and
memory. We assumed that, at some stages of storage, perceived signals are coded by
- a diversity,..of phases of narrow-band coherent oscillations of neuronal activity. 'The
phase combinations, which fo~ a system, store a certain portion of the informatior..
The beginning of formation of such a system coincides with the time of primary
positivity of cortical evaked potentials, i.e., with the very first reactive dis-
charge of many central neurons localized in different parts of the brain. The
first wave of coordinated neuronal activity appears. It is followed by several more
waves, i.e., coherent neuronal discharges. They all depend on the first discharge
- and form a chain of successive discharges, or a wave packet, since the coherent
discharges of numerous [or a set] of neuron~ are related to the slow, wave-like
oscillations of potentials.
Not all neurons are fired at the time of primary positivity. This is known. Some
of the neurons are first fired at a later time, for example, at the mament of
secondary positivity. These prim3ry discharges evoked by afferent signals form the -
beginning of a second wave packet, the second chain of successive discharges.
There are neurons that first show discharges at an even later time. They form the
beginning of the third wave packet, etc. ~
. There is an interval of at least 0.01 s between appearance of different packe~s. This
corresponds to the time of relative refractoriness following each neuronal impulse,
as well as time of summation of postsynaptic potentials causing neuronal discharge.
The minimum intervals between similar phases of different waves within the same wave
packet are of the same duration.
The diversity of phase combinationa storing information about signals impressed in
memory depends not only on the absolute value of the above~mentioned relative re-
fractoriness, but the absolute period of oscillations forming the system of in-
- formation storage.
It is known that the waves of bioelectrical activity have different periods, and
that, according to the autocorrelograms, attenuation also proceeds at a different
27
- FOR OFFI~TAT. USE ONLY
APPROVED FOR RELEASE: 2007/02/08: CIA-RDP82-00850R000300080043-3

APPROVED FOR RELEASE: 2007/02148: CIA-RDP82-44850R000300084443-3
FOR OFFICIAL USE ONLY
rate. A1pha waves persist for the longest time. Probably, expressly they imple-
- ment storage of information in memory. The periods.of alpha oscillations constitute
a mean of 0.1 s, differing slightly from one another. M. N. Livanov was the first
to attribute the phenomenon of alpha spindles, ~.e., pulsation of frequencies xn
the alpha range, to such differences.
Each system that stores information has its own frequency within the alpha range,
- and all wave systems with different frequencies that are formed under the influence
of the same perceived signal store information about this signal. As a result of
frequency interference, the phase combinations are repeated with pulsation [beat]
periods. The closest frequencies form pulsations with the longest periods, up to
1 s. Calculation of this value is described below. More detailed information about
_ the mechanisms of stability of alpha waves and their mathematical model is given
i~~ [1] ,
Such is the hypothetical mechanism of storage of information by packets of alpha
waves. Let us now turn to the mechanism of comparing perceived signals to those
stored in memory. Both are coded by specific phase combinations, i.e., specific
wave patterns. The activity of neuronal systems that store information changes in
a wave-like fashion, and it is only at specific points in time that reaetive wave
patterns corresponding to those stored in memory are put together. The time of
uniting similar wave systems corresponds to the time of successful comparison of
presented information to that stored in memory, i.~., the time of recognition of
delivered signals.
If there are several standard signals, i.e., those stored in memory, all of them
are compared to each of the presented sig~al~ successively, one by one, within one
pulsation period. The successiveness of c~~parisons is attributable to the inde-
pendent oscillations of different systems. '
At the same time. we cannot rule out the possibility of parallel comparisons. If
there are severai signals perceived and each of them is compared to the standard
signals independently, the number of such parallel procedures of comparison does
not esceed the number of all signals perceived simultaneously. Each procedure
includes the operations of successive comparison of signals stored in memory to
each of the presented signals, as noted above.
Th~se hypotheses can be submitted to experimental testing.
Experimental Testing of Hypothesis
Span of short-term memory: Signals stored in short-term memory are coded by
packets of coherent alpha waves. The tighter the waves are packed, the shorter the
intervals between ad~acent phases of waves with the same period and the more signals
can be retained in short-term memory.
If M is the ntimber of signals in the alphabet, the span of shor*-term memory can
reach value H, which is defined by the equation:
h - (aP -11 log~y I aP - 11 , _ ~ 4 ~ .
\ / ~ /
2$
FOR OFFICT_AL USE ONLY
APPROVED FOR RELEASE: 2007/02/08: CIA-RDP82-00850R000300080043-3

APPROVED FOR RELEASE: 2007/02108: CIA-RDP82-00850R000300080043-3
�
. FOR OFFICIAL USE ONLY
where a is the mean frequency of apatially coordinated oacillations of alpha rhythm
storing information about retained eignals, a~ 10 oscillations/s; p is the time
- between ad~acent phases of oscillatioas, which equals the relative refractoriness
following each neuronal impulse, p= 0.01 s.
This formula predicts the maximum diversity of packets of waves storing information
about perceived signals.
We checked the conclusion that the span of short-term memory is limited together with
I. A. Komarova; we used an epidiaecope to exhibit to sub~ects, for a short time,
lines consisting of random, equally prabably digital or alphabetic symbols, as
well ae random syllables of the consonant-vowel-consonant type. The sub~ect recalled
each line after it was presented for 2 s. Symbols called with indication of their
position and meaning were considered to be correctly recalled. A total of 10 people
= participated in these experiments, and each of them recalled a total of 30 lines of
symbols. The mean data on correctly recalled signals, syllables and symbols, are
listed in Table 2. The etandard deviations from means constituted 12-18x of the
mean value for each subject.
Table 2. Short-term memory span as a function of signal alphabet
Alphahet of N~ber of s3~ als .M~ory
~ignal St~a~ reaented p~ecalled SpSn' ~
~ ba.ts
Syllable 4000 3 2,3 28 .
Letters 32 7 5,6 23
Dig3.ts i0 9 6,3 2i
The experimentally found values for span of short-term memory do not exceed the
theoretical range indicated in equation (4). Spans calculated for artificial
syllables are closest to this range.
Consequently, artificial syllables similar to the structure of the sub~ects' native _
language are the most convenient and most economical carriers of information that
provide for error-free recall. It is more economical and reliable to use syllables,
- rather than digital or letter codes, wherever it is necessary to name events or -
ob~ecte, for example, in numbering transportation means.
Experience has shown that symbols that represent a short alphabet are recalled
more poorly than prediced by theory. Nevertheless, the order of distribution of
symbols according to recall is consistent with the forecast defined in equatibn
(4): the larger the alphabet of symbols stored in memory, the fewer symbols are
correctly recalled.
In the next series of experiments, we compared the number of all signals recalled
without exception, including incorrectly recalled ones, and the number of correctly
recalled symbols. The instructions were the same as before: recall the exhibited
lines of symbols as accurately and completely as possible. A total of 83 people
ranging in age from 18 to 25 years participated in these tests, and each of them
recalled 20 different lines each consisting of 10 digital symbols. The line of -
symbols was delivered either visually for about 3 e or orally (they were read
29
FOR OFFICIAL USE ONLY -
APPROVED FOR RELEASE: 2007/02/08: CIA-RDP82-00850R000300080043-3

APPROVED FOR RELEASE: 2007/02148: CIA-RDP82-44850R000300084443-3
FOR OFFICIAL USE ONLY
out by the experimenter one after the other within 3-5 s). As in the preced3ng
tests, the subj ects started to recall a line immediately after presentation, and
not any sooner.
- Table 3. Short-term memory span
N
Type of overall correct onl
memory ~e~ standard ~ean standard
deviation deviation
Visual 8,44 I 4,04 I 6,85 i,04
Auditory 8~~ ` i,i9 6,74 i,27
It was found that, in this case too, the experimentally determined short-term memory
- span did not exceed the theoretical limit of 29 bits, or 8.6 decimal symbols. The
number of all symbols recalled, including incorrect ones, coincided almost exactly
with the theoretical limit (Table 3).
Duration of comparison operation: Frame and labeled signals, i.e., those stored in
memory, are compared one by one because of the undulant fluctuations of neuronal
activity. Probability p of instantaneous (with no lag) comparison o~ a labeled
signal to a frame one is expressed as follows:
- �o
_ P--,,K. ~ (5)
where the numerator indicates the probability of excitable state of neurons forming
an alpha wave w-ith their activity and the denominator sh~ws the number of all
equally possible comparison operations at a given momezt.
It is l~nown that alpha waves have slightly different periods,because of which there
is pulsation, or "spindles" of alpha rhythm.
The vulsation period is determined by the follok~ing equation, with accuracy down to
the smallest interval of time by whlch alpha waves differ, i.e., with accuracy to
the relative refractoriness:
T aZP . ~6)
All o� the frame and labeled symbols in short-term memory (in the sysLem of wave
packets) will have time to be compared to one another within one ~~ich period. The
intervals between adjacent comparisons are determined by the following equation:
z~~
f(K, m) = T(1 - px) P J~ 1- x)x dx, (7 )
. ~-v
30
FOR OFFICT_AL USE ONLY
APPROVED FOR RELEASE: 2007/02/08: CIA-RDP82-00850R000300080043-3

APPROVED FOR RELEASE: 2007102/48: CIA-RDP82-00850R000300084443-3
~ FOR OFFICIAL USE ONLY
and the solution of this equation leads us to the sought formula that determines the
duration of one comparison operation:
- _ T - OK) - P)x . ~ ~8~
K -f- 1
This formula predicts quite accurately, as it was previously detexmined jlJ,
choice reaction time as a function o~ number m of equally possible alternative
signals with K = 1.
If the comparison operations occur in parallel and the number of parallel processes
equals the maximum possible number of syu~oYs in short-term memory, the mean
time for one comparison operation is reduced to:
- ~ = H � ~ (9)
In the general case, taking into consideration sub~ective probability pm of
expectation of labeled signals m in alphabet M of all signals, the time of the
comparison operation is set by a function that ensues from the original equation
(7) :
, t_ pTt~K~ m) + (1-p~,)t (K, M).
In this case, it is assinned that the sub~ective alphabets of expected signals may =
change in the course of the test from a value of m tc a value of M.
Duration of matching [comparing] procedure: In order to match a presented, i.e.,
frame, symbol with a specific labeled symbol corresponding to it that is stored
~ in memory, one to a number K operations of comparison are required, where K is .
the number of all frame symbols. Everything depends on the position of the
test frame symbol in the series of other symbols similar to ite Ultimately, the
maximum possible number of comparison operations reaches a value that is determined
by the equation:
K mK(K-}-1)
N = ~ mi = , (10)
r-i 2
- where is the position number of the test frame symb4l.
Now we can find the duration of the entire comparison procedure. For this purpose
we shall subatitute in equations (1) and (2) the values from equations (9) and
(10), i.e., the duration of one comparison operation and total number of such
operations.
We used the data in Table 1 to check the theoretical calculations.
31
FOR OFFI~TAT. USE ONLY
APPROVED FOR RELEASE: 2007/02/08: CIA-RDP82-00850R000300080043-3

APPROVED FOR RELEASE: 2007102/08: CIA-RDP82-00850R000300080043-3
FOR OFFICIAL USE ONLY
In our calculations, we took into consideration the typical mean frequency of human
' alpha rhythm and relative refractoriness indicated above, as well as the mean short-
term memory span expressed in digital symbols (which were those mostly used in the
experiments), which equals A~ 7.7 symbols, according to the data of many suthors
summarized in [5].
_ All three parameters are the only constants in our equations; they have clear phy-
siological and psychological meaning and, what is the main point, they can be sub-
mitted to indEpendent experimental verification in neurophysiological and psycholo-
gical experiments.
The results of analyzing the system of equations are listed in Table 4.
Ta~le 4. Theoretical and experimental estimates of comparison time (ms) for
labeled and frame signals in positive and negative experimental situations
Number of frame siqnalG ~
~ I Z I ~
- 3ource nu~Se~ of labeled signals ' �
I I 2 I 4 ( 1 I 2 I 4 I l I c'. I 4
Negative situation
[7] 0 58 174 !3 129 361 39 '!71 735
[4] 0 67 201 38 172 460 4~~ 382 918
' Mean of [4, 7] 0 62 t87 36 i5n ~it0 3�6 8~~6
Theory 0 39 156 31 148 ~i04 ~?3 :i"�_ 82~i
~os~tive situation
( 7) 0 29 8~ 6 fr'+ 180 '3~) 13G 368
[4] (1 34 10~1 I 19 86 ?20 47 145 355 '
Mean of [4, 7] 0 3'l 92 1'l 75 200 33 4~i0 3S1
Theory 0 30 97 '~U 88 2t8 50 165 412
The data listed in the tabie are not in contradiction with th~ hypothesis under dis-
cussion. The predicted comparison time deviates by a mean of on].y 24.7 ms from
Che experimental estimates.
Di~::ussion
Cavanagh's constant: By solving the system of equations (2) and (6)-(10) with
parameters K= 1 and m= H, we can find the time of comparison of signals in short-
. term memory to presented frame signals in a positive situation (frame signal~-
labeled):
1 ! I-xp'
t _ ~ 4 +4~ " ~ 1 H ~ � (11)
a2p
When the span of short-term memory exceeds three units, ~his time is about 0.25 s,
i.e., it equals the constant demonstrated by Cavanagh [5] in his analysis of numerous
experimental data.
~2
~
APPROVED FOR RELEASE: 2007/02/08: CIA-RDP82-00850R000300080043-3

APPROVED FOR RELEASE: 2007102/08: CIA-RDP82-00850R000300080043-3
FOR OFFICIAL USE ONLY
I.ink between rate of retrieval of signals from memory and memory span: The incre-
ment of comparison time with successive increase in number of mark~d signals per
symbol is generally considered signal retrieval time, or signal search time in -
short-term me~ry. If ~e divide the above-determined constant (11) by the number
of increments, we shall find the sought time:
~ - . . t~ _ r (H~ C12,
H-t ~
The theore tical and experimental data [5] are summarized in Table 5.
Table 5. Link between short-term memory span and time of search of signals in it
Span, Search time, ms/symbol
. Natiire of signal svmbols experim. theory
Syllables 3,y 73 73
Random shapes 3,8 68 66
Geometric figur 5,3 50 : 48
~ Words 5.5 47 46
Letters 6.35 40 40
COlOrS 7+f 38 36
Diqits 7,7 33 33 .
Table 6. Co~parison time !ms) as a function of number of labeled symbols and
memory span ~n a positive situation
Theor ' Theory
Number Experi umber Exper.
of inent span of span
symbols ~ e.~ I~.~ I~.~ symbols a.~ I~.~ I
, 2 35 34 32 29 5 i50 151 f42 i3!
3 ?5 ?3 69 64 6 19G !89 178 , 164~
4 115 112 !06 98
Sternberg~s function: The data in Table 5 convince us of the val~;d~ty a~ the ay~tenn
of equat3ons, w~.th which we can predict the experimental data obtained i~n lo].
After subst~tuting K= 1 in the system of equations, we shall obtain (in the
same terms):
' t m+~~l-~-apl' (13)
~s 4// m f
The v~t~ues of the determined function are 13sted in Tab1e 6, al,ong with experimental
est~mates, and hence the shor~-term memory span af Sternberg4s sub~ects zaost
prob$b1y cona~~tuted 6.7 symbols. This is in the known range of values, although
~,t differs f~om the general mean calculated in [5].
33
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/08: CIA-RDP82-00850R000300080043-3

APPROVED FOR RELEASE: 2007102/48: CIA-RDP82-00850R000300084443-3
~
FOR OFFICIAL USE ONLY
Table 7. Comparison time (ms) as a function of number of repeated presentations
of each combination of frame and labeled signal
Repetitions t C~~ RePetit~ons i
x~SOns+++ I ~ I 2 I 5 I l0 I ~5 13oriS'" ~ I 2 I 5 I 10 I 15
~
1 U 0 0 0 0 4 1i2 81 3i 6 i
2 34 24 9 2 0 5 !5! l09 4i 8 2
3 i3 53 20 ~4 6 189 136 5i l0 2
Sternberg discovered that the function in question was about the same for both a -
positive and negative situation in~his tests. This phenomenon could be attributed
to erroneous actions of sub~ects, to which attention is called in [7]. However,
there is more to the problem.
Comparison dynamics: We know from [7] that comparison time is drastically reduced
if the marked symbols remain constant, the same, in the course of the tests. Shiffrin
and Schneider called this decline a transition to automatie mode of comparison,
but they did not offer any hypotheses concerning the dynamics of thts process.
We know of several learning equationa. We took one of them [1]. It is an equation
of the following type;
(14)
where pi is the probability of occurrence of a learned event after the ith repeti- '
tion of the comparison operation and pl is probability determined with equation (5).
Let us indicate that the value s=1 corresponds to probability of retaining in long-
term memory all of the contents of short-term memory after single recegtion of
information. According to the rate of retention of ~igital symbols [1], probability
s= 0.15. With s= 0.15 and B a 6.7, we have the findings listed in Table 7. -
Conclusions
A ne~ system of equations of perception, which ensues from neurophysiological pre-
misc~ contained in [1, 2], explains the main patterns of compar~son of perceived
signals to thoae impressed in memory. For the first time, it was possible to
generalize diverse data [4-7] concerning the rate and volume of perception with
two constants that have overt physiological meaning: period ~f alpha rhythm which
averages 100 ms and relative refractoriness of 10 ms.
The following perception patterns were demonstrated: -
1. The volume of perception (short-tex~ memory span) is determinc' by the ratio of
- period of coordinated oscillations of neuronal activity to minimum difference in -
their phases, which equals relative refractoriness and does not exceed the values
spec3fied in equation (4), i.e., about 30 bits.
2. The duration of the variable perception lag depends on the period of frequency
pulsation in the alpha range and does not exceed the value specified in equation (6), -
i.e., 1 second. -
j4 _
APPROVED FOR RELEASE: 2007/02/08: CIA-RDP82-00850R000300080043-3

APPROVED FOR RELEASE: 2007/02108: CIA-RDP82-00850R000300080043-3
FOR OFFICIAL USE ONT~Y
3. Duration of operations of comparing signals that are perceived and stored in
memory is functionally related to their number and parameters of oscillations of
neuronal activity in accordance with equation (8).
4. The time of signal search in short-term memory and its span are functionally -
interrelated in accordance with ~he solution of equations (9)-(13).
The direction of our aearch was prompted, to some extent, by the problems of en-
gineering psycholog~ defined in [3]. We believe that this system of patterns of
visual perception will be useful in solving a number of problems of engineering
psychology. _
. BIBLIOGRAPIiY
' 1. Zabrodin, Yu. M., and Lebedev, A. N. "Psychophysiology and Psychophysics,"
Moscow, 1977.
2. Livanov, M. N. "Spatial Organization of Cerebral Processes," Moscow, 1972.
3. Lomov, B. F. "Man and Machine," Moscow, 1966.
4. Briggs, G. E., and Johnsen, A. M. "On the Nature of Central Processes in
Choice Reactions," MEMURY AND COGNITION, Vol 1, 1973, pp 91-100.
5. Cavanagh, J. P. "Relation Between:th,: Immediate Memory Span and the Memory
Search Rate," PSYCHOL. REV., Vol 79, 1972, pp 525-530.
6. Sternberg, S. "Memory Scanning: Mental Processes Revealed by Reaction Time
Experiments," AMERICAN SCIENTIST, Vol 57, 1969, pp 421-457. ~
7. Schneider, W., and Shiffrin, R. M. "Controlled and Automatic Human Information
Processing: Detection, Search and Attention," PSYCHOL. REV., Vol 84, 1977,
pp 1-66.
COPYRIGHT: Izdatel'~tvo "Nauka", "Psikhologicheskiy zhurnal", 1980
[92-10,657]
10,657 =
CSO: 1840
35
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/08: CIA-RDP82-00850R000300080043-3

APPROVED FOR RELEASE: 2007/02148: CIA-RDP82-44850R000300084443-3
FOR OFFICIAL USE ONLY
AUTOMATIC ANALYSIS OF COLOR VISION
M~acow PSIKHOLOGICHESKIY ZHURNAL in Russian No 3, 1980 pp 58-84
[Article by Ye. N. Sokolov, M. M. Zimachev, Ch. A. Izmaylov, N. P. Brusentsov,
S. P. Maslov and H. Ramil Alvares]
[Text] Evaluation ["diagnostics of human color vision is used in occupational
screening in the most varied branches of industry: chemical industry, polygraphy,
transportation and others. In developing methods for such evaluation, one must
take into consideration two asp~cts: simplicity of practical use of a method
and achieved accuracy of color vision characteristics.
Accuracy of analy~is is determineci by the resolution capacity of a method. Since
diagnostics is a classific ation process, the more classes there are for the
distribution of subjects, the more accuza.e `he diagnostic method. In this sense,
_ maxi~mum accuracy of a method consists of t~.~ possibility of differentiating between
any two sub~ects.
At present, there are developed and tested methods, which can be compared to one
another on the basis of how they combine simplicity and accuracy.
The simplest and most widely used method is evaluation with the use of tables. It
consists of classifying a sub~ect as belonging to one of four types, according to
his color vision distinctions: 1) normal trichromats, 2) anomalous trichromats,
3) dichromats and 4) monochromats. The second and third types are each subdivided
into three forms: deuteranomalop es, protanomalopes and tritanomalopes for the
fo~ mer, deuteranopes, protanopes and tritanopes for the latter.
This classification takes into consideration only very drastic variations in color
vision. In the Soviet Union, polychromatic tables were developed in the laboratory
of Ye. B. Rabkin, snd they permit finer differentiation between two forms of ano- -
malous color vision, deuteranomalopia and protanomalopia [4]. With these tables, .
one can make a distinction between three degrees of an~unaly for each of these forms:
_ C--rmild, B-~moderate and A--severe. .
The polychromatic tables are very simple to use; 10 min are sufficient foi an ex-
perienced examiner to make a diagnosis. This is a significant advantage c,f the
method. Its flaw is, primarily, that the accuracy is low. Even the most refined
tables of Ye. B. Rabkin permit classification in only 11 classes. This is a
rather rough evaluation, as compared to the information about individual distinc-
tions of color vision that are known in the science of color and which practice
. 36
APPROVED FOR RELEASE: 2007/02/08: CIA-RDP82-00850R000300080043-3

APPROVED FOR RELEASE: 2007102/48: CIA-RDP82-00850R000300084443-3
FOR OFFICIAL USE ONLY
requires. Moreover, the method of polychromatic tables requires individual testing,
airl there is no guarantee against simulation.
Anomaloscopy is another classification method used in practice. It involves the use
of special instruments, ammaloscopes. The sub~ects must compose Rayleigh's color
equation (535+6~0 = 589 nm and 470+517 =~+90 rnn) [14] o n two halves of a test field. -
The subjects are classified according to the proportion of addends in the left part
of the equation. The anomaloscope is so designed that this ratio is equal to about
one for a normal subject. This ratio maj~ be greater or lesser than one, depending
on the form of anomaly. The so-called anomaly coefficient is obtained by standardiz-
ing this ratio for a given subj ect according to the mean statistical ratio for the
norm. One can plot a partially ordered scale from the value of anomaly coefficient,
having selected arbitrarily certain intervals of values as indicators of types of
anomaly. For example, in the manual of Wyszecki and Stiles (1967), it is suggested
that sub~ects with a ratio of more than 1.5 be considered as deuteranomalopes and
with less than 0.75 as protan4malopes, etc.
Anomaloscopy is a much more refined method of evaluation than polychromatic tables;
however,'the coefficient of anomalousness [or anomaly] alo ne is mt enough to
subdivide subjects within each class; for such subdivisio n, one has to take addi-
- tional measurements of spectral sensitivity, or color discrimination function, etc.
In addition, like any threshold measurement, anomaloscopy involves some rather
labor-consuming procedures, which must be performed in order to obtain statisti-
cally reliable results. For this reason, in practice one generally makes one or
two tests, from which one can determine the form of color vision rapidly (although
with considerably less accuracy). _
Anomaloscopy is performed i nlividually, and this is inconvenient for mass-scale
studies. The advantage of anomaloscopy is that there can be no simulation on the
part of the subject; for this reason, it is recommended that an anomaloscope be used
in practice along with the polychromatic tables.
In ad~ition to anomaloscopy and polychromatic tables, color vision is evaluated by
means of color mixing functions [2], on the basis of which one can determine the
different aspects of human color vision with greater accuracy. The main element
of this m~thod is that the observer composes equations of color cor~fusion from which -
a spatial model of confusing equally bright colors, the so-called color diagram,
is constructed, which permits quantitative evaluation of the sensitivity of the
eye to barely noticeable differences in wave length, estimation of color comple-
, mentarity function, color opposition, etc. It must be noted that the color -
[chromatic] diagram plotted from equations of color confusion does not permit un-
equivocal determination of color difierences. The relationship between the metrics _
" of color confusion and metrics of minor and ma~or color differences has not yet been
sufficiently studied to interpret the results of different tests, ann efforts to
solve this problem have not yet gained wide recognition [12, 16]. However, even the
information that is contained in the chromatic diagram makes it possible to make a
very fine classification uf different forms (and within forms) of color vision.
But, in spite of the adequate accuracy of this method, it is not in wide use because
of the difficulty involved in composing the equations of color confusion, which
serve as the base data. They can be obtained under special laboratory conditions,
after a lengthy and complicated observation procedure, and testing only one ob-
server each time.
37
FOR OFFi('1A~ i(CF, nNi.Y
APPROVED FOR RELEASE: 2007/02/08: CIA-RDP82-00850R000300080043-3

APPR~VED FOR RELEASE: 2007/02/08: CIA-RDP82-00850R000300080043-3
FOR OFFICIAL i~~E ONLY
Thus, we find that simple methods of evaluation yield little information about the
colo r vision of an observer, and for this reason do not permit making a fine classi- ,
fication, while the informative methods are based on inconvenient and complicated
procedures of threshold measurements.
Consequently, the task is to develop a testing method that would be simple to use _
and yield sufticient information about the observer's color vision. We propose here
~ a new approach to resolving this tasky which is based on a spherical model of color
discrimination, to construct which the method of multidimensional scaling is used.
Evaluation of Color Vision on the Basis of a Spherical Mc~del of Color Discrimination
TY~e new approach to evaluation of color vision of man is related to analysis of
' ra~ings of aupraliminal color differences by the method of multidimensional scaling
' [7, 8]. Analysis consists of representing the set of colors to be evaluated in the
fo~-m of configuration of points in geometric space. The di~tances between points are
set as a certain monotone function of in~tial differ ences between stimuli. This -
means that very mild restrictions are imposed on the initial evaluations, and in
particular it is sufficient for them to represent at least a sequential scale
(for example, simple ranking of interstimulus differences). For this reason, the
actual testing grocedure can be very simple.
Having obtained from rhe testing ratings of all paired differences between stimuli,
special mathe~atical proc2dures are used to calculate the positions of the points in
Euclidean space so *_hat the distances between points conformed with the initial
evaluations of differences. Pro3ecting the obtained configuration in a smaller
; space [space with less dimensionality], e nbtains the minimum possible (from the
standpoint of retention of conformity betw.en interpoi.nt distances and initial
evaluations) space, the axes of which are explicitly interpreted as the main sub-
jective characteristics determining the initial set of evaluations. The remarkable
' advantage of multidimensional scaling is that "two conditions that are nonmetric in
content--monotony and minimal dimensionality--make it possible to gain complete
metric information from ordinal data" [17, 18).
A numter of studies of color discrimination using methods of multidimensional scal- ~
ing [3, 5, 6] demonstrated that the matrix of wide supraliminal differences between
co:or stimuli contains virtually all of the information about human color vision.
As ~ result of these studies, a geometric model of color discrimination was con-
st;~ucted, in the terms of which one can describe quantitatively the most varied
phenomena of color perception.
The model consists of a sphere in three-dimensiorial Euclid~an space (Figure 1),
each point of which characterizes a specific color. The three mutualiy orthogonal
axes in space are intergreted as three opponent systems (red-green, blue-yellow and
white-black). The chromatic opponent axes form the equatorial plane of the sphere. -
Monochromatic colors form a curvilinear tra~ectory on the sphere, which ends with ~
purples. The pole of the sphere represents white. The horizontal angle character-
izes the color tone and vertical angle the color saturation, whiie the sub~ective
difference between two colors is determined by the central angle of the srrall arch
of the large circle that connects the two points an the sphere correspond~:ng to
these colors.
It is mor~=~ convenient to depict the spherical model graphically in the form of ~
projection on the equatorial plane Xl~, as illustrated in Figure 2a, where the thin _
38
FOR OFFICIA i T 1SE ONLY
APPROVED FOR RELEASE: 2007/02/08: CIA-RDP82-00850R000300080043-3

APPR~VED FOR RELEASE: 2007/02/08: CIA-RDP82-00850R000300080043-3
FOR OFF[CIAL USE ONLY
solid line connecting the small dots shows the tra~ectory of spectral colors from
440 to 660 nm, which was obtained as the mean for three subjects with normal color
vision. The base data for these subjects were taken from the work of Boynton and
Gordor. [10], according to which the spherical model was constructed by the method
of multidimensional scaling, as described in [3]. The tra~ectory of spectral colors
in Figure 2a arbitrarily ends at the small arc of the large circle (dash line), which
charactErizes Che position of maximally saturated purple-crimson colors. The part
of the chromatic sphere limited in this way (which we shall call the spherical ~
chromatic diagram) constitutes the entire set of equally bright colors that the sub-
- ject sees under standard viewing conditions. Just like the traditional chromatic
diagram in coordinates XY~ the spherical diagram describes the general structure of
subjective chromatic space. -
As shown in (3], one can calculate all
of the main color functions characteriz-
X~ ~ ing the sensicivity of the observer's
I color analyzer, in particular, the
I function of saturation, opponent func-
tions, functions of first threshold of
~ ~ color discrimination, on the basis of _
� ~ 3~ the spherical model of color discrimina-
,a--~- t---.~ tion. A comparison of chromatic func-
g2'eeh ~ 9, tions estimated within the framework
_ ~~%~c\ i/,~
e ~ of the spherical model of color dis-
, ~ ' crimination to analogous ones measured
j o~t~~\.r.e by threshold methods indicates that
/~e1~' ~d these estimates are just as accurate
as threshold data.
X X.
z
Moreover, it was found possible. to -
Figure 1. obtain some chromatic characteristics
Appearance of spherical model of color on the basis of the spherical model
discri~ination. Points i and ~ represent that would have been quite difficult
pairs of colors on the surface of the to measure by threshold methods. We
color sphere; oc2~, Oz and ~i are the refer, for example, to the function
angular characteristics of points on the that describes in terms of wavelengths
sphere, which are interpreted as color sensitivity to a barely noticeabie
discrimination, color saturation and color change in color tone only or saturation
hue [toneJ, respectively alone for monochromatic colors.
Thus, the spherical model makes it possible to describe an observer's color vision
both in general and with regard to some aspect of color vision. In this sense, -
the spherical model of color discrimination is analogous to both the international
_ XY system and other popular models of color discrimination, for example the system
proposed in [13, 14].
However, as compared to all models proposed thus far, the spherical model of color
discrimination has a suhstantial advantage, which is the extreme simplicity of the
procedure of ranking differences, which is used to obtain base information. This
advantage is of decisive significance to rlassification of subjects according to
form of color vision, with which we are concerned here, as well as to a number of
other applied problems.
39 _
FOR nF'FIC'iAT i 1.cF, ONI.Y
~ ~
APPROVED FOR RELEASE: 2007/02/08: CIA-RDP82-00850R000300080043-3

APPROVED FOR RELEASE: 2007/02/48: CIA-RDP82-44850R000300084443-3
FOR OFFICIAL USE ONI.Y
a) I,o X As shown by experience, it is a rather
easy tiask to actually construct a sphe-
yy~ y~~ B rical modEl of color discrimination,
� i.e., to assess differences between
P O,S y90 colors in ranks. Unlike the trad3.tional
~ pR ~ ~R~ approaches to construction of a system
. w s~ of color specification, where the main
X~ ob~ective is to obtain as accurate as
- -/.o -0,5 ~s ~p ~ possible base data, the proposed ap-
� 660 s~p ~ proach impuses very mild restrictions
R 0 yG G .
on base data. All of the difficulty
6Z0 600 -p,s 570 SSO and labor are transferred to the experi-
mental section while processing the
base data. Analys~s of the matrix of
differenc~s, construction of the spheri-
cal model and calculation of chromatic
b~ functions--all this involves, of course,
a large amount of complicated mathema-
, tical procedures that can be performed
~ only on modern computers.
. I The proposed method for evaluation of
~ color vision has another advantage: it
~ � can be used to test stimuli with a com-
S00 plex spectral composition, rather than
~ monochromatic. This permits more com-
y50 SSO 600 6S0 ~ rehensiv~e description of the general '
~
. structure of chromatic space, using t'
� both colors close to monochromatic and ~
_~s -o- X~(R-G) thos~ with little saturation, incl~,iding i~
~-a- XZ(B-Y) white. Technically, this means that one
- X~(rq_g~ - can use luminophore instruments, the
control of which can be readily auto-
-1,0 g~ G GY YG Y 0 R mat~d, as sources of stimulation.
Figure 2. Use of this approach on a modern tech-
Sph~~rical model of color discrimination nical basis, with the use of digital
for aormal trichromatic vision computers and a color television has
a) projection of chromatic sphere on ~de it possible to automate, to a
equatorial plane X1X2 (spherical large extent, the proEedure for indivi-
diagram of color; described in dual testing, concurrently with testing -
text) of a group of sub~ects.
- b) opponent color functions calculated
within the framework of the apherical -
_ model for monochromatic colors (the
dots connected by thin lines) and non- Key:
spectral colors (circles--red-green P) purple w) white
function, squares--blue-yellow and PB) purple-blue R) red ~
triangles--white-black). Under the PR) purple-red 0) orange :
graph on the x-axis are the posi- B) blue [dark] Y) yellow
tions of nonspectral colors in the LB) light blue YG) yellow-green
~ spectrum according to dominant RG) red-green GY) green-yellow
wavelengths G) green B1) black
-
FOR OFFICIAL L1SE ONLY
APPROVED FOR RELEASE: 2007/02/08: CIA-RDP82-00850R000300080043-3

APPR~VED FOR RELEASE: 2007/02/08: CIA-RDP82-00850R000300080043-3
- FOR OFFIC[AL USE ONLY
An automated system of group analysis of color vision was designed on the basis of
equipment developed by the computer laboratory of the Scientific Research Computer
Center at Moscow State University for the computerized teaching system, Nastavnik
[1], to which a digital-computer-controlled Raduga-701 color ~elevision was
added. The equipment of the Nastavnik system contains 27 miniterminals which
operate in a time-sharing mode with a small Setun'-70 digital computer. A special-
~ ized Setun'-72 computer is used tor direct contorl of the miniterminals; it per-
forms commutation and buffering of communications, as a result of which the system's -
reactions to signals delivered from the miniterminals are quite rapid (each ter-
minal is interrogated 46 times per second).
The miniterminals make it possible for interaction with the Setun'-70 computer by
means of transmission of symbols, for which purpose each has a keyboard with 11
keys, and to obtain symbols outputted by the computer on a fluorescent digital
display, which is the receiving organ of the terminals.
The computer controls the Raduga-701 color television by~ means of digital converters,
which permit display on the video screen of any of 27 p~reviously selected colors at
27 selected levels of brightness. The block diagr2~ of the unit is given in
Figure 3.
I I I
Digital Monitor~ng & Color
converter measu~ing television
device
. , , ~ ' ~ ! .
Setun'-70 Setun'-72 I I I I
computer computer I i I I 2 I I 3 I 4 I _
' Observer's consoles
Figure 3. Block diagram of unit for automated evaluation of color vision
The program e~ecuted by the computer delivers color stimuli to sub~ects for the
specified time and at set interval~, as well as the audio signal "attention,"prior
- to each presentation. It generates a pseudo-random sequence of stimuli of a
specified length. The same program records the evaluations of subjects, forming
information for each ~ubject about the results of the experiment in the form of
a matrix of subjective diff erences (Tables 4a-6a).
In order to clarify the practical aspect of the proposed method of evaluating color
vision, we shall describe in detail the diagnostic tests conducted simultaneously
on several sub~ects.
_ Method
~ l. Sub~ects: As sub~ects we used school children and students on the psychology
faculty, who had no experience in direct appraisal of color differences. In view
of the limited amount of publications, we shall furnish here only ~art of our
results: averaged data ref erable to 15 subjects with normal color vision and aver- .
aged data for subjects with protanomalopic and deuteranomalopic color vision, as
well as individual data on one trichromat (subject M. V.), one protanomalope ~
111.
F(lA (1FFf!"'i A T t TCF (1NT .Y
APPROVED FOR RELEASE: 2007/02/08: CIA-RDP82-00850R000300080043-3

APPROVED FOR RELEASE: 2007/02/48: CIA-RDP82-44850R000300084443-3
FOR OFFICIAL USE ONLY
(sub~ect S. A.) and one deuteranomalope (sub~ect P. R.). At first, the evaluation
of each subject was made by two methods, with the use of Rabkin's polychromat-~.c
tables [4] and the R::utian AN-69 anomaloscope. Table 1 lists the results of using -
traditional diagnostic methods.
TBble 1. Results of examining subjects with the use of anomaloscope and poly-
chromatic tables
Sub~ect Rabkin polychromatic tables Rautian AN-59 anomaloscope,
~nomalousness coefficient
S. A. (left eye) Protanomalopia, grade A Protanomalopia, 0.22
S. A. (right eye) Same Protanomalopia, 0.11
S. A. (binocular) Same
M. V. (left eye) Trichromatism Trichromatism,0.99
M. V. (right eye) Same Trichromatism, 1.03
M. V. (binocular) Same
F. R. (left eye) Deuteranomalopia, grade B Deuteranomalopia, 4.2b
P. R. (right eye) Same Deuteranomalopia, 4.70
P. R. (binocular) Same
2. Stimulation~ Color flashes with complex spectral composition served as stimuli,
and they were exhibited on the color t~levision screen. The brightness of the sti- -
muli constituted 25(�2) nit. The characteristics of stimuli in the MOK-31 system _
~ urere measured with a standard calorimet Y(Table 2). The stimuli were delivered
successively in pairs, 10 timea each. Ir~ ,11 there were n(n-1)/2'10 deliveries,
where n is the number of stimuli. Duratiou of each stimulus was 0.3 s, the inter-
val between stimuli in a pair was 0.5 s and between pairs it was 1.5 s. The stimuli -
were round, with angular size of 2�. The tesr was conducted aftei dark adaptation
(5 min). In all, we used 13 stimuli that were so s.:lected as to have the basic
- colors (blue [dark], light blue, green, yellow, orange. red and purple) and differ-
ent degrees of saturation (from close to monochromatic to white) represented in the
test. In this case, white color was close to the color of a standard source C
[or S?].
TaL1e 2. Coordinates of color stimuli in the MOK-31 system
M I Color � I X Y I x~q I~ I Color x I Y(~.eq
i B~ue ~ O,i2 0,15 480 I R Orange 0,56 0,32 645
2 Light blu;e i),14 0,32 484 9 Red 0,59 0,31 620
3$lue-gr�en u,20 0,34 49~ t0 Purple-red 0,29 O,i7 558
4 Green U,2i 0,59 520 if Purple-blue 0,2~, 0,14 567
' 5 Green-yell 0,35 0,56 535 12 Purple 0,25 0,16 565
6 Yellow-gre 0,35 0,52 560 f3 White 0,3t 0,32 - ~
7 Yellow 0,70 0,40 588
_ 3. Instructions: The sub~ects must evaluate the difference between stimuli in a
pair on a scale of 0 to 9. Zero corresponds to two identical stimuli in the pair.
Maximum differences (9th rank) are determined by the sub~ects individually in the '
I}L
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/08: CIA-RDP82-00850R000300080043-3

APPR~VED FOR RELEASE: 2007/02/08: CIA-RDP82-00850R000300080043-3
EOR OFFICIAL USIE ONLY
courae of preliminary training, when all pairs of stimuli are presented once in -
random order. Before each pair, an audio signal, "attention" (600-Hz buzzer) was
_ given. The x'eaction consists of depresaing one of the 10 keys (0, l, 9) on
the sub~ect's console.
4. Base data. The results were processed in two separate stages. Primary pro-
cessing uras performed in the course of the test, for all participants at once,
on the Setun'-70 compufer. After such processing, all of the ratings of each
subject were summarized on an individual triangular matrix, the element cf which =
was the rating averaged for the number of presentations and multiplied by 10. Thus,
the matrix describes for us all of the differences between stimuli, pair by pair.
For each sub~ect we obtained three separate matrices of interstimulus differences
in relation to one of three observation conditions: binocular and monocular for
the lef t eye, and monocular for th~ right eye. Subsequent data processing, which
consisted of analysis of base differences by the method of multidimensional scaling,
was performed on an M-6000 computer using a program based on the algorithm of Young
and Torgerson [19].
Construction of Spherical Model of Color Discrimination
. 1. Dimensionality of subjective space: As a result of analysis by the method of
multidimensional scaling for each matrix of differences one calculates the coordi-
nates of points in an n-dimensional Euclidean space (n is the number of stimuli
used) and characteristic roots, from which the significance of each spa.tial axis
is assessed. Theoretically, the characteristic roots may be either positive or
equal to zero. The number of positive roots indicates the dimensionality needed
foi each matrix of the real Euclidean space, in which one can describe without
error the base data as distances between points. However, because of ~udgment
errors made by the subjects, zero root values may change in both the positive and
negative direction. For this reason, in practice, one determines the true dimen-
sionality according to the largest positive roots. -
Table 3 lists the characteristic roots for aYl matrices that are analyzed in this
article. These data show that there is a change in decimal order of value between
the third and fifth roots. Up to the third root, all of the values have four ~
digits ["decimal" digits] or more (with the exception of the mean matrix for
th~ d~uteranomalope), whereas from the fifth root on all of the values have only
three or less digits. The fourth root serves as a sort of boundary between signi-
ficant and definitely ingignificant axes. This shows that the subjective space of
color discrimination must have at least three dimensions [measurements].
It must be noted that the value of a characteristic root per se cannot, of course,
serve as an adequate criterion of minimal dimensionality, particularly in the
case of noise-covered data, which usually occurs in experiments. However, one can
estimate the necessary boundaries of true dimensionality from the values of the
roots. Here, as in other formal analyaes, the final solution depends on intrinsic
[proper] interpretation of obtained data.
. As we have already stated, for su~~ects with normal trichromatic vision, the three
dimensions of space are related to three oppone~it characteristics of color discri-
_ mination. Of greatest significance is the characteristic root of the red-green .
spatial axis, followed by the one of the blue-yellow axis, while the characteristic
43 -
FnR (1FFIC'fAT i16F, nNT.Y
APPROVED FOR RELEASE: 2007/02/08: CIA-RDP82-00850R000300080043-3

APPR~VED FOR RELEASE: 2007/02/08: CIA-RDP82-00850R000300080043-3
FOR OFFICIAL USE ONLY
root of the white-black axis has the lowest value; the difference between the third
root and the first two is usually cons3derably greater than the difference between
the latter (see, for example, Table 2 in [3]). The same findings are observed in
Table 3. In the columns with the headings of "mean for 15 sub~ects" and "sub~ect
M. V.," where data are listed f4r normal sub~ects, the order of the characteristic
roots is the same, from the red-green to the white-black axis.
This three-dimensional interpretation is generally found to als~ apply in the case
of color anomaly; however, we then observe same changes that mi~st be mentioned.
In the deuteranomalopic sub~ect (P. R.), the order of axes ac~�ording to value of
characteristic roots correpsonds to the order of axes for a t.richromat, i.e., the
root for the red-green axis has the highest value, then comes the one for the
blue-green a~cis and the lowest for the whf.te-black axis, but there is a greater
~ifference between the roots of the first and second spatial axes, and particularly
between the second and third. In the deuteranomalope P. R., the third (white-
black) axis is negligibly greater ttg n the fourth spatial axis, which we inter-
pret as the result of experimental interference ["noise"]. This means that the color
color discrimination space of a deuteranomalope is "flatter," i.e., it is close to
an Euclidean plane. The findings are quite different for the protanomalope (sub-
~ect S. A.). In this case, the blue-yellow axis is the most significant; the character-
istic root of this axis is highest in value and differs drastically from the other
roots. Next is the root of the white-black axis, while the characteristic root of
- the third, red-green axis is lowest in value. In subject S. A., the increased signi-
ficance of the white-black axis signifies in spatial terms an even greater devia-
tion from Euclidean space than in the case of the normal sub~ect. In the following,
we shall discuss *_he interpretation of ~hese differences in terms of a spherical
model of color discrimination, where we .za'yl discuss the obtained data.
2. Spherical nature of subjective space: The spherical nature of the con�iguration
of points obtained by means of multidimensional scaling is conf irmed by the fact
that one can always find a geometric center for a specified configuration of
points, i.e., a point that is equidistant from all existing points. There must
remain a high degree of correlation between the base estimates of differences and
interpoint distances. The distance from the center of the sphere to each point _
may fluctuate due to judgment errors made by the sub~ect. I'or this reason, in
practice one must find a point as the sphere center for which the scatter of
tlese distances (radii) is minimal. An interative procedure ~s used for the search,
ar4 it minimizes the standard deviation of radii from the central radius which are
c~lculated at each step. One takes the center of gravity of the initial configura-
tion of points as the starting point. After finding the optimum center in this
sense, the entire configuration o� points is shifted linearly so that the center
of the sphere coincides with the start of the coordinate axes. The scatter of
radii is measured by variability, as a percentage, of ratio of standard deviation
to mean radius.
The results of the calculations are listed in the table of Euclidean coordinates
obtained for each matrix of differences (Tables 4b-6b). At the ~nd of each table,
the values are given for the coefficients of correlation of interpoint distances
and base differences with the variability of sphere radius obtained for a given
configuration of points.
~
FOR QFFICTA:. USE ONLY
APPROVED FOR RELEASE: 2007/02/08: CIA-RDP82-00850R000300080043-3

APPROVED FOR RELEASE: 2007/02/08: CIA-RDP82-00850R000300080043-3
FOR OFFICIAL USE ONLY
~ ~ $ a~~i ~ ~ � ~ ~ ~ ~ o
v N ~ ~ ( ~
'7 " eMr~ G? N CO m~~~~ ~ i i ~
~ ai l~ V' N
v
� J
~ c~ v c~ ~ n~ o v oMO aQi cNi ~
~~~~~~N ~ I I I I
~ t~ O G~ V~ t0 ~ p ~ N~~'
~ Q ~~~.h-~ m i
i eD O N c~D 00 .r ,
~ � w M~.� ~ I I I'"
~ " ~ ~
w
~ � C ~ ~pp c~ ~n
w a^ u~a ~ N cq ~ aN0 ~ O i c+~ ~ O N .
N ~ ~ m ~ ~ ~ ~ ~ ~
( ~
'd
W ~
~ p Q~ ~i !y~~ e~ ~ ~ ~
~ , ~ eD ~N ~ C~ c~D N ~ ~ i i i i
~ ~ -vl r. (
~ '
a~
_ ~d ' a~
~
. p o~`.-~� c~ ~ c o o i e
i o~
~ f~. ~ a~o u~ �o ~ ~ I ~
N aA
~
~
~
- ~ o ~ c~ m co .r w ao ao
~ a ~ ~ ~ ~ N o ~ ~ ~ � ~ w'" ~ ~
~ ~ I
~ ~
0
w �
a w ~ $ ~ ~ ~ c~ i o ~ ~ ~ c ~ 1
~ a~ ~ I ~ ~ ~ I
cti ~
a~
.o
O
Vl ~ O o N ~i O c~*~ c.~~ .~i~
o o~~,~N~ I I I I ~
o ~ ~
u
fA ~ ~ N t~'J' ~~l7 N N O ~ M ti M
O~ c~ c, ~ e~ ~ m.-~ ~ i~ i i
~ ro o~ ~N
~ �1 ~
~ ~ ~ ,
~ U ~p
~ ~ ~ ~ ~ m W ~T N ~ ~ O ~ N N ~
I I I ~
~ N
~ -
ri
~ ~ r. e~ c� v~ ~m m ~ ao a~
~ a
45 -
FOR OFFICIAI r icF ONi,Y
APPROVED FOR RELEASE: 2007/02/08: CIA-RDP82-00850R000300080043-3

APPR~VED FOR RELEASE: 2007/02/08: CIA-RDP82-00850R000300080043-3
,
FOR OFFICIAL USE ONLY
Table 4a. Matrix of suh3ective d~fferences (mean for 15 sub~ects)
COZOY' ' I~~ 1 I 2 I 3 4 I 5 I 6 I 7~ 8 I 9 I l0 l 1l I iZ l 13
~
Blue 1 1i 42 60 60 61 88 73 76 57 47 82 62
Light blue 2 3~ 53 60 57 6i 72 75 56 50 50 55
Blue-green 3 ' 38 4i 42 50~ 60 66 58 47 49 33
Green 4 i8 29 55 88 68 67 63 68 46 -
Green-yellow 5 i9 54 63 69 67 60 6L 43
Yellow-green 8 37 52 56 5~, 5i 58 29
Yellow 7 24 32 43 44 44 36 ~
Orange 8 g 45 45 48 53
- Red 9 43 49 44 53
Purple-red 10 ~ i2 48
- Purple-blue !t 7 33
Whi~e i3 ~ ~
Table 4b. Conrdinates of color points in Eucl~dean space (mean fo~ 15 sub3ects)
CQ10z I M I x~ I X~ I X~ , I R
I
Blue 1 .~,6 37,5 10,9 41,7
Light hlue 2 i5,? 34,0 i2,2 39,5
Hlue^gxeen 3 i9,6 ii,6 32,7 39,9
GXeen 4 32,7 -i4,5 29,t . 46,1
Green~Xellow 5 29,i -:4,5 33,8 46,9
xello~t-gzeen 8 55,2 -i6,7 35,6 42,2
~ ~
Yellow 7 -i8,i -i4,~ 32,6 38,9
.Orange ~ 8 -30,8 -t5,6 23,4 41,7
Red 9 -35,i -i6,8 2:,6 44,5
Purpl'e-red SO -l6,! i6,8 32,2 44,8
Purple-blue ii -17,1 20,2 35,3 44,i
Purple i2 -22,7 2i,7 32,9 45,5
White i3 '!,8 1,3 44,4 44,5
~Mean radius 43,i
S tarrlard deviation 2~6
Variance 6, i
Coefficient of correlation U,9R3
A comparison of these figures to one another indicates that the i*~dices of ~pheri-
city are very similar to one another in all of the sub3ects. Variability of the
radius ranges from 4 to 11X, with a coefficient of correlation of 0.96-0.98. This
is quite consistent with the data obtained in [3, 5], and it indicates that
one can compare not only sub~ecta with normal coloL vision to one another, but
_ also subjects with anomalous color vision in terms of a spherical model of color
discrimination.
FOR OFFICIAL rSE ONLY
APPROVED FOR RELEASE: 2007/02/08: CIA-RDP82-00850R000300080043-3

APPROVED FOR RELEASE: 2007/02/08: CIA-RDP82-00850R000300080043-3
FOR OFFICIAL USE ONLY
Table Sa. Matrix of subjective differences (sub~ect P.R., mean for monocular and
binocular presentations)
COIOY' I Nv I: i 2 I 3 I 4( 5( 6 I 7 I 8 I e I 10 ( 11 I 12 ( IS
Biue 1 0 i9 64 64 58 77 84 87 4i 32 37 36
Light blue 2 !7 65 57 53 76 83 82 42 35 36 30
Blue-green 3 3i 34 34 57 67 83 38 33 3i 1y
- Green 4 i ii 71 84 85 67 55 62 40
Green-yellow 5 6 61 77 79 64 53 58 36
Yellow-green 6 35 5i 58 44 43 47 28
Yellow 7 ! 9 40 46 47 44
Orange 8 i 33 54 50 63 �
Red 9 4! 57 48 59
Purple-red i0 6 2 25
Purple-blue ii 0 !9
Whi~e i3 � ~
Table 5b. Coordinates of color points in Euclidean space (subject P. R., mean for
monocular and binocular presentations)
Color w X~. x, x. I R '
I I
~lu~ ! -ii,8 -4i,8 64~1 ~ 77,5 ~
~,j~c~ht blue 3 -i9,8 -i4,7 72,9 80,0 .
Blue-green .
Gzeen 4 -41,7 i7,5 6i,7 68,7
Green-yellow 5 -33,0 i6,8 55,3 66,6 ' -
Yellow-green 6 -l2,3 !9,! 57,i 6i,4
Yello~n , 7 28,3 24,i 56,3 67,7 .
Re~nqe 8 43,2 20,3 48,1 67,7
9 46,i 20,4 43,9 66,9
Purple-red f0 ~ 22,3 -i3,9 65,7 . 70,7
Purple-blue !i 7,3 -i8,8 59,7 , 62,4
Pur le i2 i3,6 -i7,9 59,5 63,6
Whi~e 13 -5,3 -8,i 8i,4 8i,8
,
� 7,~,2
~~andardl~eviation ~ 6,9
Variabi? it , $ ~ 9'8
Coefficien~ of correlatio ~~982 .
~
3. Rotation of subjective space; The system of Cartesian coordinates in three-
dimensional Euclidean space, which we obtained by the method of multidimensional
scaling and in which we pursued our analysis to this point, is in no way related -
to the obtianed configuration of points, since all calculations up to now were
~ade in terms of distances, which are unrelated to the selected system of coordi-
nates in Euclidean space. In the spherical model of color discrimination, each
Cartesian axis is interpreted as one of the same opponent Gharacteristics of '
color, and it must be related to the color points by means of a certain form of
~.7
FOR OFFICIAi ricF (1Ni.v
APPROVED FOR RELEASE: 2007/02/08: CIA-RDP82-00850R000300080043-3

APPR~VED FOR RELEASE: 2007/02/08: CIA-RDP82-00850R000300080043-3
F'OR OFF[CIAL USE ONLY
function; for this reason, of the many possible syatems of coordinates we must
choose the one that conforma with thia interpretation.
_ Table 6a. Matrix of subjective differences (sub~ect S. A., mean for monocular and
, binocular presentations)
COZOY' I Ni ( 1 I 2 I 3( 4 I 5 I 6 I 7 I 8 I 9 I 10 I 11 ( 12 13
Blue ! I I ii I 6t 79 8t 87 85 84 R7 3? 27 30 65
Light blue 2 58 72 79 80 83 80 82 45 29 27 60
Blue-green g 60 59 57 63 65 66 38 47 42 10
Green 4 i1 !6 20 29 33 70 8i 78 56
Green-yellow 5 9 9 23 25 65 77 68 54
Yellow-green g 4 9 i8 66 76 68 48
Yellow 7 7 i3 68 68 70 53 -
Orange 8 9 66 72 77 61
Red 9 6 i 73 74 58
- Red-purple ~ l0 . 27 25 40
Blue-purple it ti 51
Whi~ee i3 53
Table 6b. Coordiizates of color points in clidean space (subject S. A. , mean for
_ monocular and binocular presentations)
~ Co1or I x, I x, I x, I x. ` I R
Blue i i,2 4h,1 3,3 44,3
Light blue 2 8,1 39,i 2,4 40,0
Blue-green 3 l0,7 - ii,4 40,8 43,7
- Green 4 t7,5 -35,7 8,g 40,7
Green-yellow 5 2,2 -35,4 t7,5 39,6
Yellow-green s -1,7 -40,4 23,9 45,1
_ ~Xellow 7 -11,3 -37,3 21,0 44,3
- Orange 8 -i2,6 -37,5 i6,0 42,7
Red 9 -17,5 -37,4 21,1 46,3
Purple-red !0 -13,2 24,0 28,0 39,1
Purple-blue ii -1t,6 39,6 20,3 46,0
Purple 42 -10,9 33,7 20,8 4t,i
White i3 9,5 3,4 40,6 41,8
Mean radius 42,7
Standard deviation 2,~ ~
Variabilit , ~ 5,8
Coefficien~ of correlatie I 0,982
~4~
FOR OFFICIAT, :'SE ONLY
APPROVED FOR RELEASE: 2007/02/08: CIA-RDP82-00850R000300080043-3

APPR~VED FOR RELEASE: 2007/02/08: CIA-RDP82-00850R000300080043-3
FOR OFFICIAL USE ONLY
In order to solve this problem, each configuration of color points is rotated ortho- -
gonally so as to meet the following conditions:
1) A positive direction of the X1 axis characterizes the green Fhase of _
- red-green opponent function; the maximum of this phase is in the color
range with a wavelength of 505 nm [13, 14]; consequently, in our case,
the X1 axis must pass between blue-green and green (495 and 520 nm, res-
pectively). A negative direction of the same axis corresponds to red or
purple-crimson, according to Hartridge [9], phase, and for. this reason must
correspond to the position between purple-red and red (620 nm).
2) A positive direction of the X2 axis analogously characterizes the yellow
phase of blue-yellow function, which has a maximum in the color range with a
wavelength of 575 nm [11]; consequently, this axis must, for the colors we
took, pass between yellow-green (560 nm). A negative direction of the X2
axis must, accordingly, pass between the points representing purple-blue
and b lue (480 nm) .
3) The color points with maximum value (in positive or negatiive direction)
- for the first axis must have a minimum value for the second axis of coordi-
nates and vice versa. This condition is applicable only to colors with
about the same saturation; in our case, these are blue, green and red. _
4) Use of white among the stimuli makes it possible to add one more condition:
the point for white must have a maximum value for the third axis of coordinates
and minimum value for the first two axes, since a positive direction of the
X3 axis characterizes the white phase of white-black opponent function.
These conditions are valid for subjects with trichromatic color vision. In the case
of dichromatism, either the first or second condition is eliminated depending on
the type of dichromatism.
These conditions determine unequivocally the only system of coordinates for each
configuration of points. The values of these coordinates are listed in Tables 4b-
6b. They characterize the opponent systems separately for each subject.
_ 4. Standardizing the chromatic sphere: The nonzero variability of radii of the
color sphere, which in our opinion is a consequence of errors in the subjects'
appraisals, leads to the fact that, in each instance, the color sphere has a cer-
tain "thickness" in a radial direction, which makes it difficult to measure colar
functions. Moreover, the mean. radius of the sphere fluctuates from subject to
subject also, and this makes interindividual comparison difficult. We are not
discussing here the intrinsic interpretation of radius of the color sphere, and in
order to eliminate fluctuation thereof, we standardized the color sphere and thus
transformed it into a unique [singl~~ one: X12 + X22 'F' X32 = 1, where Xir - Xir~Ri'
i= 1, 2, 13; r= 1, 2, 3 and RZ�is the radius of the ith point.
Standardization of the color sphere results in some shift of the color points in
relation to one another, i.e., some change in interpoint distances. In order to
assess this shift, we again calculated the coefficient of correlation between the
base matrix of interstimulus differences and interpoint distances on the single
sphere. They are listed in Table 7. A comparison. of these data to the coefficients _
- of correlation before standardization shows that the changes were very negligible,
49
F(1D l1FFif iAr iTCC nNi.V
APPROVED FOR RELEASE: 2007/02/08: CIA-RDP82-00850R000300080043-3

APPR~VED FOR RELEASE: 2007/02/08: CIA-RDP82-00850R000300080043-3
FOR OFFICIAL USE ONLY
within the range of 0.01-0.02. This confirms the hypothesis of random fluctuation
of the radius of the color sphere.
Table 7. Coefficients of correlation between matrices of initial differences and
distances between points on single sphere
J Conti uration on sin le s h re
Matrix o~ i,nitial�differences left right 1 binocular
, eye eye ~
Subject M. V. (left) 0,954 0,92f ~ 0,9i2 ~
(right) 0,92i 0,959 0,896
(binocular) 0,932 0,9f5 0,950
Subject P. R. (left) 0,970 0,957 0,954
(right) 0,964 0,97i 0,953
(binoc. ) 0,948 0,950 0,963 '
Sui~ject S. A. (left) 0,972 0,94i 0,939
(right) 0,956 0,969 0,927
(binoc. ) 0,923 0,9i8 0,969
Configuration on sphere
Mean for 15 subjects (trichromat ) 0,978
Mean for 3 matrices (subj. P.R.) 0,972
Mean for 3 matrices (subj. S.A.) ~
After standardization, the construction of the spherical model of color discrimina-
tion is considered completed. All subsequent calculations and graphs based on
them were made for the standardized and single coardinates of color points.
~ Results of Experiments
We have submitted our results in three different form~ that reflect the successive
stages of the diagnostic procedure.
As we have already indicated, the primary materi al, i.e., evaluation of paired dif- ~
ferences between color stimuli, is submitted in the form of triangular matrices of
differences in Tables 4a-6a. Table 4 lists data that were av~raged for 15 matrices.
They refer to both monocular and binocular evaluations, but all of the data were
obt~ined on subjects with normal color vision. This did not include the data for
sub~ act M. V., which are listed by us as an exainple of a normal trichr~mat. Table
5a ~ists data for the deuteranomalopic subject, P. R., averaged for three matrices,
while Table 6a analogously lists data for protanomalopic subj ect S. A., averaged =
_ for three tests.
_ Tables 4b-6b list the results of multidimensional scaling of each of these matrices
of differences. In all of the tables, the colors used have the same names and
numbers as in Table 2. The Euclidean coordinates of each point after rotstion of
space and shift of the sphere to the start of the coordinate axes are listed in
columns X1-X3 in Tables 4b-6b. At the end of each table, the coe~ �icient of
- correlation between initial evaluation of differences and Euclidean interpoint
distances is given.
The conatant distance bet~aeen each point to the center (radiu~) is proof of the
sphericity of the obtained configuration of points. The values of the ~adii are
listed in Tables 4b-6b, in column ~t. Below are the values of the mean radius for
50
- FOR OFFICIA ~
APPROVED FOR RELEASE: 2007/02/08: CIA-RDP82-00850R000300080043-3

APPROVED FOR RELEASE: 2007/02/08: CIA-RDP82-00850R000300080043-3
FOR OFFICIAL USE ONLY
all points of standard deviation from the mean, and variability, the value of which
characterizes the relative scatter of radii as a percentage of tfie mean.
These data define the subjective color space, regardless of the spherical model of
color discrimination.
The final answer, in terms of the spherical model, are given in the form of drawings
of spherical diagrams of chromatism and opponent functions. These data are suffi-
cient to illustrate the potential of the proposed method of evaluating color vision.
Diagnostic Analysis of Data
Comparison of resuZts to data for monochromatic colors: Unlike monochromatic stimuli,
which are traditionally used to plot chromatic functions, the color stimuli we u~ed
here to evaluate color vision have a complex spectral composition, as a result of
which they have less saturation than monochromatic colors; they are characterized
by wavelength, like monochromatic colors, with regard to color tone (Table 2).
Evaluation of color vision for this group of subjects in terms of a spherical model
can be based only on a comparative analysis of configurations obtained in one test.
It is of no significance here whether the sfimuli have a complex spectral composi-
tion or simple, monock~romatic one. But in order to be able to compare data ob- _
tained at different times, with the use of different sets of stimuli, it makes sense
to relate specific configurations of color points to the configuration of monochro-
matic colors in each instance. This would make it possible to distinguish between
changes caused by specific sets of stimuli from those obtained due to the distinc- ,
J tions of color discrimination. `
Such a comparison is drawn in Figure 4. The projection of the sphere on plane X1X2 _
is illustrated in Figure 4a. The thin solid line shows the trajectory of monochro-
matic colors on the surface of the sphere, plotted from the mean data for normal
color discrimination. The tra~ectory ends arbitrarily in the region of purple
colors with a geodesic line that passes through the point~ for tlue and red (440 and =
660 nm, respectively). The wavelengths of the colors are shown near some of the
points. The interval between points is 10 nm. On the same graph, the configura-
tion of points representing the colors used in this study is shown by the large
dots connected by heavy lines. These data were also obtained from averaging those
referable to the 15 subjects with normal color vision.
Quite obviously, the configv.ration of nonspectral c~lors is "inscribed" in the con-
figuration of spectral ones, which is the consequenc~ of lower saturation of non~
spectral colors. Blue, green, yellow-green, green-yellow, yellow, orange and red
are closed to the spectral colors. One can assess the accuracy of calorimetric
measurements in terms of the chromatic MOK-31 diagram according to the position of ~
these colors in relation to the spectral line. In general, all of the wavelengths
_ calculated calorimetrically and listed in Table 2 correspond to ~:he wavelengths of
the points that represent these colors on the sphere. Yellow is an exception;
according to calorimetry it has a wavelength of 585 nm, but it is Iocated on the
sphere in the region that is characterized by a wavelength of ~595 nm, according to
its spectrum.
Now the configuration of nonspectral colors on the spilere serves as a sort of
standard, according to which we evaluate the subjects'` individual data. The -
51
Fl1D l1FFT!'T A T i rcC (~Ni V
APPROVED FOR RELEASE: 2007/02/08: CIA-RDP82-00850R000300080043-3

APPROVED FOR RELEASE: 2007102/08: CIA-RDP82-00850R000300080043-3
FOR OF'FICIAL USE ON1.Y
graphs of red-green, blue-S~ellow and white-black functions, plotted for nonspectral
colors accordi.ng to the averaged data for 15 subjects with normal color vision,
u~hich are illustrated in Figure l~b, are an analogous standard for plotting opponent
color functions. The wavelength of the color is plotted on the x-axis and the
value of coordinates X1 (red-green function), X2 (blue-yellow function) and X3
(white-black function) are plotted on the y-axis. They are compared to the data for
monochromatic colors, in the same way as in Figure 4 a. These opponent functions
are not plotted for all colors, but only those which, as shown in Figure 4 a, are
closest to spectral. This was done so that the opponent functions that we obtained
for nonspectral coZors would retain the cusromary form of traditional sgsctral oppo-
nent functions. The discrepancies we see in Figure 4b are the result of differences -
in sets of stimuli, analogously to what we discussed, rather than discrepancies
of color discrimination.* ~
Figure 4.
Spherical model of color discrimination ,
for normal trichromatic, protanomalopic
and deuteranomalopi~ vision _
a) a) spherical diagram of chromatism. The
,
~,t circles connected by a solid line repre-
' ~ sent the configuration of colors for
n? normal vision; triangles connected by a
nK r~+ C3 ~ dash line represent configurati.on for pro-
tanomalopic vision and the squares con-
y~" eb�~ X~ nected by dash-dot lines deuteranomalopic
. ~
-~o i ~ 6 ~ ~ vision ,
K;~o_ 3 b~ ~pponent functions calculated in accord- '
o ~w3 3* ance with data for normal, protanomalopic
' ; and deuteranomalopic vision; black symbols
~ ; ,
~ ~ refer to red-green functions X1i white to
� v------
, \ ~ ~ ' 'y~31 3~~ "3 blue-yell~w X 2
"~,ot-' c) white-black rpponent functions for the
same forms ~f color vision (designations
of forms of color vision are identical in
all three figures) .
b~ , . Key for this and ~
;~-~---~.---U--- Figures 5a, 6a and 7a:
, ~
; XJ ~ II) purple
0,5 r,a ,0. IIC) purple-blue
~ ^-~i{>- ~,:-~o d' ~.b,.d~ IIK) Purple-red
saor ss~- , ~ C) blue
� ~ r=~= ~on s o,s P) light blue
~ ~ C 3) blue-green
i' pZ' ~ % ; ti!/
~ tj ~ ; i~' B) white
-o,s ~ ~ ~ ~ S00 sso 600 ~ ~C 3) yellow-green
~ trichromatistq 3Xf) gr~en-yelloj~r
~ � Normal
~z~ xotano~ua-� . Protanomal. orange
o~ ~e~teranc:- s ~ Deuteranom. xf) yellow
` malopia K) red
3) green
*Trans~.ator's note: I~ the source, reference is made throughoutthis.sectionto Fig. 3;
however, it appears it should be Figure 4.
~
APPROVED FOR RELEASE: 2007/02/08: CIA-RDP82-00850R000300080043-3

APPROVED FOR RELEASE: 2007102/08: CIA-RDP82-00850R000300080043-3
- FOR OFFIC[AL USE ONLY
Of special interest are the curves illustrated in Figure 4c for white-black opponent
function, which characterizes the whiteness of equally bright colors in the spheri-
cal model. In general, this function coincides with the analogous one obtained for
spectral colors; but since colors with wavelengths in the range of 490-510 nm, where
_ white-black function has an additional peak, are not represented among the non-
spectral stimuli, the graph of white-black function shows a ~reak in this place. "
In some cases, when it is necessary to describe the behavioi of colors in this
~ range, we shall draw upon additional data for blue-green (495 nm); but it must be
borne in mind that it is significantly shifted itself from the trajectory of
. spectral colors (Figure 4a) in the direction of white.
Evaluation of Color Vision According to Averaged Data
The traditional methods of evaluating color vision, which we discnssed above, per-
form classification chiefly according to forms of color vision as, for example,
shown in Table 1. In this section, we shall discuss how an analogous classif ication
is made in terms of the spherical model of color discrimination. For this analysis
we took the average data separately for each of the subjects representing three
different forms of color vision. Averaging was done for three matrices of base
differences: binocular, monocular for the left and right eyes. Each of these
matrices is known to characterize the same form of color vision, since they were
obtained for the same subject. And each subject is characterized only by one form
of color vision for both eyes, as can be seen in Table 1. For this reason, averaging
should only emphasize the typical characteristics of each form as a result of leve~-
ing off individual differences within each form.
Table 5a (deuteranomalopia) and Table 6a (protanomalopia) list the averaged matri-
ces of diff?rences for anomalous forms. As an example of normal color vision, we
used the data discussed above referable to 15 subjects (Table 5a). The results of
multidimensional scaling of these matrices are listed in Tables 4b, Sb and 6b, res-
pectively. There too, the characteristics of sphericity of the obtained configura-
tions are given: variability of radii and coefficient of correlation for distances
_ between points in relation to initial d ifferences.
The spherical color diagram ~'igure 4a) serves as the principal material for evalua-
tion; it illustrates the configuration of color poinr_s characterizing each fo mn of
color vision. The points connected by solid lines show the configuration for
normal color vision; squares connected by a dash-dot line illustrate deuteranomalopic
vision, while the triangles connected by a dash line represent protanomalopic vision.
The differences between forms of configurations are very substantial, and they can
be used for error-free determination that we are dealing with deviations from normal
- color vision. A comparison to the standard configuration of normal color discrimina- -
tion shows that in deuteranomalopic subject P. R. there is approximation of points
in relation to the red-green X1 axis, which is particularly significant in the
blue-purp].e part of the diagram. Approximation of color points in terms of the
spherical model indicates worsening of sensitivity to color discrimination. As com-
pared to normal color discrimination, there is poor differentiation here between
blue-green and white; all three purple colors are also poorly distinguished f rom
, white. The entire series of colors, from red to green, is differentiated worse
than with nermal color discrimination. However, discrimination is less reduced in
the long-wave part of the spectrum than the short-wave ona. In addition to changes _
in co~or discrimination on the red-green axis of the spherical diagram, the
53
~nu n~~rrr ~ T T ~c~ n~r v
APPROVED FOR RELEASE: 2007/02/08: CIA-RDP82-00850R000300080043-3

APPROVED FOR RELEASE: 2007/02148: CIA-RDP82-44850R000300084443-3
FOR OFFICIAL USF. ONLY
deuteranomalope presented compression of color configuration on the blue-yellow X2 -

~.=LL~ axis also, and it affects chiefly blue. This may be a feature only of this subject,
P. R.; however, it is more likely that this is a manifestation of a distinctive
~ feature of deuteranomalopia in general, as alsa indicated by comparing these
changes to those characterizing protanomalopic vision. As seen in Figure 4a, the
color configuration is also deformed in two directions, red-green and blue-yellout
_ with protanomalopic vision. But while there is compression in the red-green direc-
tion, as in the deuteranomalope, the changes are differ.ent in the blue-yellow direc-
tion; they af.fect chiefly yellows, rather than blues, and the colors are shifted
away from white, rather than to it. These opposite tendencies of changes referable
to the blue-yellow axis of the spherical diagram in the deuteranomalope and protano-
malope can be interpreted as additionsl classification tags for different forms of _
color vision.
r
The general findings referable to deuteranomalopic vision are indicative of the
- systematic shifts of color poj.nts from the periphery of the spherical diagram toward
its center, toward the point for white. The changes are generally different in
the protanomalope. Here we can readily distinguish two small regions of space, in
which virtually all colors are grouped: purple and blue-light blue in the blue
- zone, orange-red and yellow-green in the yellow zor?e. These zones are localized
along the blue-yellow axis of the space on difrerent sides of white. It may be -
assumed that the pole of the sphere represents another, third zone, in which purple-
" crimson and blue-green should be located. This is canfirmed by the fusion of
white and blue-green, which is seen on the diagram.
The differenc2 between deuteranomalopic- ~nd protanomalopic vision shows that the _
traditional prohlem of considering anoms:: ~is forms as merely quantitative devia- ,p~
tions from normal color vision is solved ~i~ferntly here. For the deuteran.omalopic
subject this hypothesis coincides with the general changes in subjective space of
color discrimination, and for the protanomalopic subject it is not confirmed,
although this conclusion, of course, requires cons-3erable statistics for each form.
As we have already stated, in addition to a general description of color vision in
- terms of the spherical model, one can also make comparison5 of various special
characteristics. To illustrate this thesis, we shall compara different forms of
, color vision according to color opponency functions [Figure 4b and c). rrom the
_ standpoint of opponent colorstheory, the change related to each type of anomaly
sr~uld be manifested by specific changes in opponent functions. Deuteranomalopia
l~ads to reduction of red-green function and a shift of the point of intersection of
blue-yellow function and the zero x-axis in the direction of the long-wave end of
the spectrum, while protanomalopia leads to reduction of red-green function and a
shif t of the point of intersection of the second ~unction and zeru axis in the
short-wave direction. The degree of change is related to the degree of anomalous-
ness of color vision. In the case tritanopia and tetartanopia, the reduction
is referable to blue-yellow function, while red-green is retained.
A comparison o� opponent functions af anomalous and normal form.: ~f color vision
plotted within the f ramework of the spherical model generally conforms with the
theses of opponent theory. For both the protanomalope and deuteranomalope, the red-
_ green function is reduced, as compared to normal vision. In the protanomalopic
subject, this reduction of red-green function is more marked, and this characterizes
his greater anomalousness, as compared to the deuteranomalopic subject. In most
cases, it is impossible to make such comparisor. between different forms by the
- 54
APPROVED FOR RELEASE: 2007/02/08: CIA-RDP82-00850R000300080043-3

APPROVED FOR RELEASE: 2007/02108: CIA-RDP82-00850R000300080043-3
FOR OFFICiAL USE ONLY
' traditional anomaloscopic methods, since the coefficients of anomalousness of the
deuteranomalope and protanomalope are parameters on two unrelated ordinal scales.
The change in blue-yellow function differs somewhat from the theoretically assumed
one. Since colors in the range of 480-510 nm are not represented here, the presence
or absence of a shift of the point of intersection of the functions with the zero
axis cannot be determined, particularly for the deuteranomalope, although by extra-
polation the shift in the short-wave diYection is very obvious in the protanomalope.
At the same time, there is also a significant change in opponent function: it in-
creases in the yellow phase in the protanomalope and decreases in the blue phase
- in the deuteranomalope.
Considerable changes can also be demonstrated in the fo~ of white-black function,
and each phase of color vision affects this function in its ow~x way (Figure 4c).
With deuteranomalopic vision, the value of the short-wave part of the function
' grows appreciably, whereas with protanomalopic vision it decreases in the range of
average waves. The general difference between anomalous and normal vision is
manifested by a drastic increase in values of the function in the local 490-510 nm
se~ent. One cannot demonstrate this change in overt form, because the blue-green
color used in this study is poorly saturated, as compared to other colors (Figure
- 4a). However, one can assess this local change indirectly, even on this poorly
saturated color, an~ for this reason ~.ts position is indicated in Figure 4c sepa-
rately from white-black function by the white symbols. The symbols correspond to
the black symbols of color vision (circles--normal, triangles--protanomalopic and
squares--deuteranomalopic vision).
- Thus, the combination of general features of color discrimination (in the form of
configu?-ation of color points on the spherical diagram) and special features (in the
form of opponent functions) makes it possible to unequivocally differentiate between
forms of color vision. In this case, evaluation can be made both visually, accord~
ing to the overall configurations, and formally, using the amplitudes of each phase
of opponent functions as indicators. A~ we have already seen, the results of our
evaluations are cou}~letely analogous to the results of the most refined of the
~ procedures used in pra~tice, i.e., anomaloscopy. In the next section we shall
- discuss evaluation of color vision witnin each form, and we shall demonstrate that
they are readily demonstrable, even upon visual analysis of configurations of
color points on the spherical diagram.
Comparison of Individual Characteristics of Color Vision
Formal description of individual spherical diagrams. Before we discuss the meaning-
fu1 aspect of individual differences, we shall briefly analyze the extent to which
different conf.igurations of points on a single sphere differ on the basis of a
formal solution, when the obtained data merely represent a certain geometric shape.
Since the configuration of points is formally given to us as the distance between
points, the subject of our analysis will be the matrice of distances between pairs
of points. The problem is to check the extent to which the interpoint di~tances
for a given configuration are closer totheir base matrix of differences ~han they
are to some "alien" matrix. The coefficient of linear correlation serves as the -
gage of closeness. And we compare only data, for which the probability of confusion
of coafigurations is particularly high (such as data for the le�t and right eye of
the same sub~ect, monocular and binocular presentation, etc.), since the differ-
ences between a normal and anomalous sub~ect are already quite obvious from the
55
F(1R (1FFif'TAr iTC~, nNT.V
APPROVED FOR RELEASE: 2007/02/08: CIA-RDP82-00850R000300080043-3

APPROVED FOR RELEASE: 2007/02108: CIA-RDP82-00850R000300080043-3
FOR OFFICIAL USE ONLY
base data. Table 7 lists the coefficients of correlation, which were calculated to L
compare each configuration to each base matrix of differences. The calculations
indicate that, in all cases, "its own" matrix is better correlated with the ob-
tained configuration than a"foreign" one.
Normal trichromatism: Let us discuss the spherical chromatic diagram illustrated
in Figure 5a, where configurations are shown that were plotted from the data refer-
able to subject M. V. The triangles connected by a dash line refer to the right
- eye, while the white circles connected by a thin solid l.ine refer to binocular -
color discrimination. Data for the averaged matrix (boldface Points connected
with boldface line) are given against the backgrouna of the individual graphs.
A c~mparison of individual data *_o one another and with the averages shows that
there is no systematic difference between them. Both the left and right eye of
subject M. V. have generally the same spherical diagrams, that are very close to
the averaged data. Since both eyes of subject M. V. present the same color dis-:
crimination, binocular color discrimination is also in the range of the averaged
s~andard diagram, as we had assumed. It must be noted that, as can be seen on the
spherical diagram, this sub~ect is characterized by rather wide scatter, as com-
pared to averaged data, i.e., considerable experimental "noise." This can also be
seen from the magnitude of variability of scatter of radii and coefficient of cor-
relation between the answer and base evaluations. While variability constitutes
6.1% for the mean data, with 0.983 coefficient of correlation, in subject M. V.,
who presents less correlation (0.963 binocularly, 0.960 right eye and 0.959 left
eye), variability is greater (11.1 binocular, 4.4 right eye, 7.8~ left eye). The
principal cause of this noise is the ss "1 amount of statistics: each element of the
~ matrix of differences for subject M. V. ti..: mean of only 10 ratings, while the
averaged matrix of differences is given foc 150 ratings. Moreover, this subject
is characterized as one of the "noisiest," even in comparison to others for whom
there is also a small amount of statistics (or even less). The chief reason for
including expressly his data, instead of a better s�tbject, is that M. V. was tested
in the same group of subjects with anomalous color visi~n that represent the
main data in our experiment. According to the findinqs on subject M. V., which
are illustrated in Figure 5a, he can be unequivocally characterized as having normal
colo.r discrimination in both the left and right eye.
Th.s is ~~~ident just as clearly on the graphs of three opponenc functions illus-
tr ~ed in Figure Sb: red-green X1, blue-yellow X2 and white-black X3. The tri-
ar.gles, squares and circles refer to data for the left and right eye, and binocular
vision, respectively~
No changes were demonstrable that could be related to some anomaly of color vision
on the curves illustrated in Figure 5b. They coincide entirely with the curves
that characterize our standard for normal color discrimination (Figure 4b). Thus,
according to the results illustrated in Figure a and b, sub~ect M. V. is charac-
terized as a normal trichromat. This diagnosis conforms entirely with the results
of both anomaloscopy and evaluation with the polychromatic tablc.,
Deuteranoma]opia: Very different color discrimination was found in subject P. R. -
Figure 6a i]iustrates the spherical diagram with configurations characterizing the
color discrimination space for the left and right eyes, and binocular vision (see
key for Figure 5a). The position of the color points, like for the averaged data,
-
r,;;.
~
- 56
APPROVED FOR RELEASE: 2007/02/08: CIA-RDP82-00850R000300080043-3

APPROVED FOR RELEASE: 2007102/08: CIA-RDP82-00850R000300080043-3
FOR OFFICIAL USE ONLY
. ~ 10 X indicates a consistent deviation from
- the configuration characterizing nor-
' ~ mal color discrimination in the di-
~ rection of white, i. e. , there appears
~cr', ' 1~ to be reduction of color saturation.
'nK/ nC~'0,5 ~8 However, the magnitude of the shift
~ C3 is not the same for the left and
~ . j ~6 ~.1 right eye: for virt~sally all colors,
,-!,0 ~-QS I ' 0, ~ I,0 the right eye presEnts poorer color
� b ~ discrimination than the left. The
K~ 0 * ~ 3 configuration for binocular vision
_ps ~3 3m ~ is in an intermediate position, be= -
tween monocular, which is indicative
of equal participation of each eye
_ in binocular vision, i.e., conformity
-1, with the simplest (though, of course,
_ � not the only possible) form of bi-
- ' a nocular interaction.
- 1,0 ,
If we analyze the functions of this
sub~ ect (Figure 6b and c) , it is ,
, ~ also easy to see that the right eye
~s (squares) presents more change than
the lef t, whereas in binocular color -
I X3 ~ discrimination the function or.cupies
an intermediate position between the
~ ys0 S00 SSO 600 6,~p 1ef t and right eye . The change in
- red-green function affects mostly the
green phase, and to a lesser extent
X~ the red; but we cannot state this
-O,S o-lef t with complete certainty, since the
M.V.o-right red phase of this function is only
o-t~inoc. partially representad. There are .
X~ corresponding changes in blue-yellow
� 2 function of subject P. R. The yellow
, ~ b phase of the function remaining un-
Figure 5. changed, there is distinct reduction _
Spherical color diagram (a) and opponent of the blue phase; this reduction is
more marked for the right eye than
functions (b) for subject M. V. The the left. These changes indicate that
circles refer to data for binocular subject P. R. is characterized, as
vision, triangles for left eye and squares
- is usually the case, by a mixed
- for right eye in (b) [also see key for form of anomaly, deuteranomalopia
_ Figure 4] with some worsening of the short-
wave mechanism also, rather than a pure form of anomaly (from the standpoint of
oppanent theory). The white-black function for subject P. R. is illustrated in
Figure 6c. Under all observation conditions, it differs from no~al white-black
- function in that there is a very marked el,evation in the short-wave region of the
spectrum. The white-black function of the deuteranomalopic sub~ect is virtually
the same as in normal vision in the long- and short-wave regions of the spectrum.
This confirms the thesis that a change in the short-wave region of the spectrum can
serve as a distinctive feature of deuteranomalopia, in addition to the main ,
features manifested by changzs in red-green function.
57
_ FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/08: CIA-RDP82-00850R000300080043-3

APPROVED FOR RELEASE: 2007102/08: CIA-RDP82-00850R000300080043-3
FOR OFFICIAL USE ONLY
(a) ~ ~o x - C~~ ~
, C � 1,U X
_ � ~ ' ~ -
� as; ? ; , ~ ~ ~ . .1 ',p~
; n ' 1 �ca . ~ ~ ~ -
~ n nc o ` . ~ - ~ ~ ,~S ,
i ~ 6 X~ i
-l, ~.~qJ /0 ~ . ~1 Q ~ ~ C3 ' ~
. ~o
~ 6 0 ~ A" ~
� K -------~---a` -gsj `1~ Gi5 l,a
~ 0 ~ -QS ~3 3~ ~ i
~
j ~~-QS ~
. , , s
~ K ' ~
t ~ -1,0 , . ~ Nf3 3~
l,0 , ' ~ -1
~h~ /0 .
fb).
as . . .
. .
. ,ran ~ .
~ yS0 SSO 6A~ 6S0 s ~ SSO ,
~ ~ ~ yS0 600 6S0
_gs ..o -lef"t
Pd~ ; _`t�' _as ~-2eft
Tii~. SoAe a-~~o ,
"mean o- n.
-~o . of 24, d~~r~ ~
-io
~o ,
~ ;
~ ~ , ~ . ,
~
ae / a ~ .
~ , ~~R o_ left ~ ' S.A �~_~lefh .
. o-~' ~ ~ ~~o~..
. e � � -_.tg~a$ , p ~ ne n
yS0 S00 SSO 600 6d0 � 4J0 i0~ SSD 600 6S0 '
Figure 6. ~ Figure 7.
S~:erical color diagrams (a) and opponent Spherical color diagrams (a) and opponent
functione (b~ c) for eubject P. R. functions (b, c) for sub~ect S. A.
(For both above ~igures, see key to Figure 4 and caption to Figure 5)
- The res~ilts obtained conform entirely with the evaluation made in Table 1 for this
sub~acr; the difference between the right and left eye, diagnosed only by means of
anomaloecopy, i0 also very distinctly manifeeted as a result of ++aing our diagnostic
method. Z'he right eye of aub~ect P. R., which has a higher coe.fficient of anomalous-
nees~. i~~cha~a~c~ertzed by greater anomaly in terms of the spl?erical model of color
dtacrimination as well. Moreover, the method we propose permits evaluation of bi-
nocular color diecrimination also, and as we have seen from the data submitted for
eubject P. R., the la~ter does not necesearily coincide with either the left or
58
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/08: CIA-RDP82-00850R000300080043-3

APPROVED FOR RELEASE: 2007/02148: CIA-RDP82-44850R000300084443-3
. ~
FOR OFFICIAL USE ONLY
right eye; of course, one could have aesumed the binocular characteristics as well
on the basis of monocular evaluations, but it is easy to predict the variant of
monocular differences when it alone cannot be used for even an approximate evalua-
tion of binocular vision (fo~ example, when one eye is normal and the other is
anomalous, etc.).
It must be noted that there was consiclerable less "noise" for the results we obtained
on sub~ect P. R. than the subject with normal color discrimination; while variabi-
lity of sphere radii was the same, the coefficient of correlation to base evalua-
tions was higher for the former.
Protanomalopia: The new variant of configurations obtained for subject S. A. is
illustrated in Figure 7a (see the key to points on the spherical diagram in Figure _
4). In the case of protanomalopia, the configuration of color points differs sub-
stantially from both the configuration of the normal subject and deuteranomalope.
In sub~ect S. A., the differences between configurations characterizing the right
and left eye were even more significant than j:n sub~ect P. R. The left eye of S. A.
discriminates very poorly between green and yellow-green. We see on the graph that
the points representing these colors have virtually coincided and they shifted
toward the yellow half of the X2 axis. At the same time, light blue and blue
shifted toward the blue ha.lf o� the X2 axis. Orange-red colors are also shifted
toward the yellow half of the X2 axis, but they are~~~stinguished from one another
considerably better than yellow-green colors. The ~ame can be said about purples,
as compared to blue and ]Lght blue. The overall features of the color space for the
left eye of subject S. A. are the same; for the averaged data for protanomalopic
vision (Figure 4a), the color space is compressed on the first spatial axis, and
all colors are grouped in three small loci situated along the second axis. The same
tendency of marked separation of all colors into three loci is also inherent in
the right protan.analopic eye of subject S. A. But while the left eye forms a con-
figuration that is more compressed in the green half of the red-green axis and less
so in the red half, the opposite is observed for the right eye: more compressed red
half of the axis than the greex~, i.e., the red and green phases of this opponent
system may change differently, depending on the individual distinctions of the color
analyzer.
Subject S. A. differs from the others in bionocular color discrimination also; in
sub~ect P. R. it could be interpreted as average, intermediate between the two
forms of monocular color discrimination, whereas in subject S. A. binocular color
discrimination constitutes a summary result, and the configuration of color points
_ on the spherical diagram (Figure 7a) representing binocular color discrimination ~
contains both monocular configurations.
The opponent functions for sub~ect S. A. are illustrated in Figure 7b and c. The
chromatic opponent functions (b) are drastically deformed, and in general this
deformity is consistent with the predictions of opponent colors theory. However,
there are also differences that are referable to one and the other opponent func-
tions. -
The red-green function (Figure 7a, white symbols) is reduced in both the red and
green phase; however, the extent of this reduction in these phases is not propor-
tionate. For the left eye there is more reduction of the green phase and for
the right, the red phase of opponent.function, as compared to normal vision. We
had already partially observed this asymmetry in subject P. R., but for him the
59
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/08: CIA-RDP82-00850R000300080043-3

APPROVED FOR RELEASE: 2007102/08: CIA-RDP82-00850R000300080043-3
FOR OFFIC[AL USE ONLY
green phase presented more reduction than the red in both eyes. This asymmetry ex-
plains why evaluation of the forms of color vision--protanomalopia and deuterano-
malopia--should not be made according to opponent function in general, but separately
according to each of its phases. As shown by the data on subject P. R., in the
case of deuteranomalopia there are more marked changes in the green phase ~f red- .
green function; they are correlated with the degree of anomalousness (i.e., reci- -
procals of the coefficient of anomalousness) as well as the general characteristic
of color vision which, as we have already stated, is the region of color discrimina-
tion space on the chromatic diagram. For protanomalopic sub~ect S. A., it is ex- -
pressly the variability of the red phase of red-green function that is correlated
to both a change in region of color discrimination space (its area is smaller for
the right eye than the left) and degree of anomalousness (the anomaly coefficient
is higher for the left e}e than the right) (Table 1). The chan~es in the green
- phase with protanomalopia, like those in the red phase with deuteranomalopia, can be
urilized for a more comprehensive analysis within each form of anomalous trichroma-
tism. Under all observation conditi~ns, the graph of blue-yellow opponent function
varies significantly in the yellow phase and is constant in the blue for subject
S. A. (Figure 7b, black symbols). This change is the opposite of what we observed
in deuteranomalopic subject P. R. (Figure 6b), where only the blue phase varies.
At the same time, from the change in this opponent function there is nothing we
can say about differences within each form, protanomalopia or deuteranomalopia,
since blue-yellow opponent function is the same for both eyes, in spite of the
appreciable differences between the left and right eye of subject S. A. (as was the
case for subject P. R.).
In the protanomalopic subject, the white-black function diminishes in the yellow-
orange region, and this decline is gene--ally unrelated to the eye that the sub~ect
uses. This distinguishes th~ white-blaci. =uc~.:tion of a protanomalope from that of
a deuteranomalope, where the change affect~ mainly the short-wave part of the spec-
trum. As in the case of deuteranomalopia, we submitted the break in this function
in the range of 490-510 nm, since we cannot overtly demonstrate a significant rise
of the peak, which is distinctly visible in the whi=e-black function of monochro-
matic colors (Figure 2b).
Analysis of configuration of color poirtson the spherical diagram and opponent func-
tions of protanomalopic subject S. A. shows that diagnosis ot this form of anomalous
coi.or vision in terms of the spherical model of color discrimination coincides with
tht~ re5ults of anaJ.ysxs that we established by traditional metiiods. At the same
ti: it is apparent that the spherical model of color discrimination permits finer
ditferentiation between the color vision of this subject and others. The advan-
tages of the spherical model are even more evident when we examine the individual -
characteristics within this form of color vision, as we see from the data of
comparing monocular and binocular color discrimination.
The minimal "noise" in the results of sub~ect S. P., as compared to others, serves
as confirmation of the reliability of qualitative analysis of his data. With
variability of radii of 7-10%, the coefficient of correlation with in3tial evalua-
� tions reaches a maximum for values of 0.97-0.98.
Conclusions
1. A spherical model of color discrimination is proposed, according to which the
entire set of equally bright colors can be designated with points on the surface of
60
FOR OFFICIAL vSE ONLY
APPROVED FOR RELEASE: 2007/02/08: CIA-RDP82-00850R000300080043-3

APPROVED FOR RELEASE: 2007/02148: CIA-RDP82-44850R000300084443-3
FOR OFFICIAL USE ONLY
a sphere in three-dimensional space, with the color white on the pole and spectral
colors nearing ttie planes of the equator to different degrees.
2. The entire psychological diversity of colors is defined by angles: horizontal
that co rrespond to the color hue and vertical that correspond to saturation. The
physio logical characteristics of color discrimination are three orthogonal systems
of coordinates representing the red-green, blue-yellow and white-black opponent
systems of neurons.
- 3. The subjective difference between colors is measured with the small arc of a
large circle drawn through the points representing colors on the sphere.
4. Knowing the matrix of subjective differences between colors, one can calculate
the coordinates of points corresponding to colors in thrEe-dimensional space on the
surf ace of the sphere and obtain an individual diagram of color vision for a given
sub,j ect .
5. The subjective diversity of colors in cases of protanomalopia and deuteranoma-
lopia is also situated on the surface of the sphere in three-dimensional space;
however, the configuration of the color diagram differs from normal.
- 6. A comparison of individual diagrams of color discrimination makes it possible
to sep arate and describe quantitatively the different forms of color vision dis-
_ turbanc es. The data obtained by the method of multidimensional scaling coincide
with the evaluations made on the basis of anomaloscopy and polychromatic tables.
However, the method of plotting individual diagrams of color vision permits more
accurat e differentiation between individual distinctior?s of color vision.
7. One can obtain individual diagrams of color discrimination simultaneously for
an entire group of subjects by the method of automatic evaluation of color vision,
_ which is based on a computer-controlled color television. The procedure of determin-
ing the properties of color vision is reduced to a simple system of evaluating sub-
jective differences, while data about the properties of color discrimination are
- obtained with the use of a computer.
BIBLIOGRAPHY
1. Brusentsov, N. P.; Maslov, S. P.; and Ramil Alvares, H. "The 'Nastavnik'
Automated Teaching System," in "Vychislitel'naya tekhnika i voprosy
kiDernetiki'~ [Computer Techno~ogy and Problems of CyberneticsJ, Moscow, Vyp 13,
1977, pp 3-13.
2. Judd, D., and Vysheski, G. "Color in Science and Technology," Moscow, 1978.
3. Iz~aylov, Ch. A., and Sokolov, Ye. N. "Metric Characteristics of a Spherical
Mode1 0~ Coloz D~tscrimination," VESTN. MGU [Vestnik of Moscow State University],
Series ~,4: "Psychology," No 2, 1978, pp 19-28.
~ 4. Rabki.n, Ye. B. "Polychromatic Tables for Testing Color Perception,'.' Moscow,.
1,971,
5. Sokolov, Ye. N.; Izmaylov, Ch. A.; Izmajlova, T. V.; and Zimachev, M. M. "A
Spherica7. Model of Co1or Vision,t' VESTN. MGU, Series 14: "~sychology,~' No 1,
19 77, pp 45-52.
61
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/08: CIA-RDP82-00850R000300080043-3

APPROVED FOR RELEASE: 2007/02108: CIA-RDP82-00850R000300080043-3
FOR OFFICIAL USE ONLY
6. Sokolov, Ye. N.; Zimachev, M. M.; and Tzmaylov, Ch. A. "A Geometric Model of
Subjective Space of Color Stimul3," TR. VNIITE, Ergonomika [Works of the All-
Union Scjentific Research Institute of Esthetic Styling in Engineering,
Ergonomics], No 9, 1975, pp 101-102.
7. Terekhina, A. Yu. "Metric Multidimensional Scaling," Moscow, 1977.
8. Idem, "Nonmetric Multidimensional Scaling," Moscow, 1977.
- 9. Hartridge, G. "Current Advances in Physiology of Vision," Moscow, 1952.
10. Boynton, R. M., and Gordon, J. "Bezold-Brucke Hue Shift Measured by Color-
Naming Technique," J. OPT. SOC. AMER., Vol 55, 1965, pp 78-86.
~ li. Jameson,D., and Hurvich, L. M. "Theoretical Analysis of Anomalous Trichromatic -
Color Viaion," Ibid, Vol 46, 1956, pp 1075-1089.
12. Judd, D. B. "Interval Scale, Ratio Scales and Additive Scales for the Sizes
of Differences Perceived Between Members of a Geodesi c Series," Ibid, Vol 57,
" 1967, pp 380-386.
13. Hurvich, L. M., and Hameson, D. "Some Quantitative Aspects of an Opponent-
Colors Theory. I. Chromat~c Responses and Spectral Saturation," Ibid, Vol 45, -
1955, pp 546-552. �
14. Hurvich, L. M., and JamESOn, D. "Some Quantitative Aspects of an Opponent-
Colors Theory. II. Brightness, Sa,~-~ation and Hue in Normal and Aichromatic
Vision," Ibid, Vol 45, 1955, pp 602-~.6.
15. Hurvich, L. M. "Color Vision Deficiencies," ir~ "Handbook of Sensory Physiology,"
ed. by D. Jameson and L. M. Hurvich, V., VII/4, New York, 1972, pp 582-624.
16. McAdam, D. L. "Visual Sensitivities to Color Diff^_rences in Daylight,"
J. OPT. SOC. AMER., Vol 32, 1942, pp 247-274.
17. Snepard, R. N. "The Analysis of Proximities: Multidimensional Scaling With ~
Unknown Distance Function," PSYCHOMETRIKA, Vol 27, 1962, pp 125-129, 219-246.
18. Idem, "Metric Structures in Ordinal Data," J. MATH. PSYCHOL., Vol 3, No 6, 19
1966, pp 287-315. _
19. Young, F. W., and Torgerson, W. S. "Torsca, a Fortrain 4 Program for Shepard-
Kruskal Multidimensional Scaling Analysis," BEHAV. SCI., Vol 12, No 6, 1967,
PP 216-222. -
20. Wyczecki, G., and Stiles, W. "Color Science, Concepts and Methods, Quantitative
Data and Formulas," New York, 1967.
COPYRIGHT: Izdatel'stvo "Nauka", "Psikhologicheskiy zhurnal", 1980
- [93-10,657~
10,657
CSO: 1840
- , 62
FOR OFFICIAL USE ONLY ~
APPROVED FOR RELEASE: 2007/02/08: CIA-RDP82-00850R000300080043-3

APPROVED FOR RELEASE: 2007/02/48: CIA-RDP82-44850R000300084443-3
FOR OFFICIAL USE ONLY
OPTICAL METHODS OF TRANSFORMING VISUAL FEEDBACK
Moscow PSIKHOLOGICHESKIY ZHURNAL in Russian No 3, 1980 pp 85-94
[Article by V. A. Barabanshchikov, B. I. Belopol'skiy and N. Yu. Vergiles]
[Text] For the last few years, a wide circle of specialists has given its close
attention to studies of quantitative and qualitative distinctions of eye movementse
Among the numerous aspects of analysis of oculomotor activity, a special placz
belongs to the study of principles and mechanisms of function of the system that
regulates eye movements under different observation conditions.
It must be noted that the existing conceptions of function of the human oculomotor
system (OMS) are largely related to the choice of inethodological research procedures.
Thus, the use of several new experimental procedures--presentation of "pulsed,"
dual "pulsed" stimuli [14, 15], stimuli that are stabilized in relation to the
retina [2, 5, 6]--yielded some new data on OMS function that do not always by far
- conform with the conventional conceptions.
Our objective here was to describe the "battery" of special methods of studying the
OMS, which are based on the same principle: transformation of visual feedback.
Let us consider the main elements of the OMS. A signal of discrepancy between the
position of the fixed object on the retina and center of the fovea is delivered tn
- the input of the system. The output of the system is a turn of the eye. The pur-
pose of regulation is to minimize the input signal. Thus, the system is established
in a stable position only when the pro~ection of the object of fixation on the
retina coincides with the fovea. Since each turn of the eye in the direction of
the object of fixation is associated with attenuation of the error [discrepancy]
signal, the human OMS can be described (by analogy with technical systems of
- automatic regulation) as a tracking system of positional monitoring [control]
with negative feedback [1, 8].
The fact that there is f eedback in the OMS is of utmost methodological importance,
since transformation of output-input relations opens up new avenues for analysis
_ of the patterns of OMS function and its extrasystemic links. (Let us recall that
a change in feedback is one of the principal investigative methods for technical
systems of automatic regulation.)
- We can distinguish three main parameters of transformation of visual feedback of
the OMS: magnitude, sign and direction. The magnitude of visual feedback is the
ratio of angle of eye movement to angular amplitude of displacement of the projection
63
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/08: CIA-RDP82-00850R000300080043-3

APPROVED FOR RELEASE: 2007/02108: CIA-RDP82-00850R000300080043-3
FOR OFFIC[AL USE ONLY
of the object of fixation over the retina. Under ordinary conditions of OMS func-
tion, one can consider the visual feedback to equal one. However, in principle
this parameter may be either >1 (visual angle by which the pro~ection of the
ob~ect shifts is larger than the angle of eye movement) or