Apter - 1969 - Cybernetics and Art PDF
Apter - 1969 - Cybernetics and Art PDF
Apter - 1969 - Cybernetics and Art PDF
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms
The MIT Press is collaborating with JSTOR to digitize, preserve and extend access to Leonardo
D
JSTOR
This content downloaded from
132.208.246.237 on Sun, 13 Sep 2020 19:17:29 UTC
All use subject to https://about.jstor.org/terms
Leonardo,Vol. 2, pp. 257-265. Pergamon Press 1969. Printed in Great Britain
MichaelJ. Apter*
Abstract-This article is a non-technical introduction to cybernetics, the study of
'control and communication in the animal and the machine'. A number of funda-
mental cybernetic concepts are described including some of those involved in
information theory (like the notion of a message and amounts of information), in
control theory (like homeostasis, negative feedback and servomechanisms) and
in automata theory (like algorithms, Turing machines and networks). The
structure of cybernetics is outlined showing the relationship between work on the
further development of the theory, on understanding organismic processes in
cybernetic terms (especially through model building) and on constructing more
purposeful hardware systems. Bionics, cybernation and artificial intelligence
(which especially involves heuristic programming) are all fields concerned with
the development of such hardware systems.
It is argued that cybernetics is related to art in three ways: it may be used by the
scientist in studying art, it may be used by the artist in creating works of art-and
may have been one of the influences behind the development of the idea of machines
as works of art and machines for creating art, as well as the increasingly process-
oriented nature of contemporary art-and finally cybernetics may itself be
regarded in certain respects as an art form in its own right.
257
This content downloaded from
132.208.246.237 on Sun, 13 Sep 2020 19:17:29 UTC
All use subject to https://about.jstor.org/terms
258 Michael J. Apter
D D
Message
Encoding D Decoding D
D
I
Noise
Fig. 1. Information transmission situation.
Negative feedback
certain limits. For example, pressure cookers admits steam or fuel to the engine. What happens
maintain pressure within certain limits, the human is that as the speed of the engine increases, so the
body maintains a large set of physiological variables weighted arms move at an increasing speed and
(like temperature) within limits, viable business therefore rise. As they do so they operate the valve,
organizations maintain certain economic variables closing it in proportion as the speed of the engine
within certain limits and so on. Homeostasis may mounts. The result of this arrangement is that the
be achieved by means of negative feedback. more the engine exceeds a given speed, the less it is
Let us look at this in terms of one of the first, and supplied with steam, so that it slows down again.
certainly one of the most famous, homeostatic But as it slows down, so the valves are opened again
systems in the history of engineering: the Watt to admit more steam, and so it can speed up again.
governor (cf. Fig. 3). An engine using the Watt This negative feedback counteracts the speed of the
governor is homeostatic in that it is able to maintain engine as it goes above or below the homeostatic
its speed of operation within limits. Part of the out- point and the engine soon settles down to a stable
put of the engine is to turn a set of weighted arms speed. It is important to note that it is the speed of the
which are mounted on pivots so that they rise by engine itself which in effect controls the speed. This
centrifugal force as they revolve. This small part of situation is represented diagrammatically in Fig. 4.
the output is used to control the engine speed auto- A thermostat controlling the temperature of a
matically because the arms operate a valve which room represents a slightly more complex system in
Power
I
I
Engine
speed
I
I Output
that not only does it control a heating system seem trivial and yet it is of great philosophical
(through negative feedback) but it maintains the importance. For centuries people have been able to
heat of the room in relation to variations in the argue that there is a fundamental difference between
environment. Not only is it homeostatic, but also living and non-living entities in that animals,
adaptive-the room is kept at the same heat whether including man, are able to be purposeful, to act
it is snowing outside or the sun is shining. It does this towards the future in some mysterious way. Non-
by continually monitoring the temperature of the living entities, on the other hand, were supposed to
room and automatically turning on the heat when be always subject to cause-and effect principles, to be
the temperature drops below a specified level and explicable in terms of previous rather than future
turning it off when it goes above a higher specified events. In terms of the examples we have just seen,
level. This is an example of what is known in engin- this particular argument is clearly not valid.
eering as a servomechanism [5]. Purposeful behaviour by animal or machine can be
Let us look at a biological example of a similar explained in identical 'cause-and-effect' terms.
process. A man walking along a pavement normally
tries to stay on the pavement: as long as he does so Automata theory
we can say that he is maintaining homeostasis Although in terms of the definition of cybernetics,
in this respect. Any change of direction which causes automata theory may seem to be less central than
him to move away from the pavement causes him to information and control theory, in practice it has
note that his direction of movement is no longer been closely bound up with the development of
correct and to move in an opposite direction from cybernetics.
the direction of error. So movement by him away Its nature can perhaps be most easily understood
from the desired direction of movement results in by contrasting it with information theory. If infor-
action being taken which corrects this error. This mation theory is essentially a part of probability
allows him not only to stay on a straight pavement theory and is concerned with quantities, automata
but to adapt to changes in direction of the pavement theory is part of metamathematics and is concerned
itself. Here the negative feedback is in part (seeing with proof. We could perhaps call automata theory
himself move away from the desired position) more 'the logic of systems'. The term automaton means in
obviously informational (in the everyday sense of this modern sense essentially a system or machine
the term) than the negative feedback in the Watt which is deterministic and to which probability
governor example. And the error correction is less notions do not have to be applied. In automata
inexorable. But the principle is the same. theory an automaton is an ideal, an archetype,
Many systems incorporate the converse of negative rather like a diagram in Euclidean geometry in
feedback-positive feedback. With positive feed- which we do not have to take into account the
back, part of the output of the system is fed back into thickness of the lines.
the system in a positive direction. This has the Automata theory is concerned with what kinds of
effect of increasing rather than reducing deviation behaviour can and cannot, in principle, be achieved
in the system. Typically this is unadaptive: for by different kinds of systems. For example, what are
example, the more anxious one becomes the less the minimum properties which a system would need
able one may become at performing a certain skill in order to carry out a computation of a particular
which in turn leads to increased anxiety and further kind? An especially striking question of this sort is:
decreased performance. On the other hand, positive Is there any logical contradiction in the idea of a
feedback can be related to negative feedback so as system being able to reproduce itself and, if not,
to increase the effectiveness of the latter. For what are the requirements? This particular question
example, power brakes amplify pressure on a in automata theory has received much attention, the
brake, braking being a· form of negative feedback classic work being done by von Neumann [6].
controlling the speed of the vehicle. Metamathematics, the field which automata
A system may control its behaviour, using theory derives from, is concerned with studying the
negative feedback, not only to maintain a state it is notion of proof in mathematics. That is, it is con-
normally in but also to achieve some future state cerned with questions like: In what sense are mathe-
that it has not yet been in. For example, the man matical statements true? How can we justify the
walking along the pavement may be going to the methods of deduction used in mathematics?
shops: the whole activity may be regarded as error- In order to carry out investigations in this area,
correction in terms of the desired end-state of being Alan Turing [7], a British mathematician, proposed
at the shops. A ground-to-air missile homing on its that for certain purposes (those relating to a general
target would be another example of a future- problem described by Hilbert and Ackermann [8]) it
directed, error-controlled system. The point may might be useful to think in terms of a theoretical
A
Mathematical/logical
theories :
Information theory
Control theory
Automata theory
B
Induction f C
Applied science
Deduction t Pure science
especially: especially:
Engineering - Simulation Psychology
Bionics -
Management Biology
machine, since called a Universal Turing machine, cine, psychology, psychiatry, anthropology, socio-
for carrying out mathematical computations by logy, economics, education and business manage-
means of well defined programmes (sequences of ment. In fact, there is something of a two-way
instructions which specify how the machine will act). process involving abstract ideas from cybernetics
When automata theory developed, the use of Tur- becoming applied in various concrete situations, the
ing's archetype computer became a widely used way latter sometimes giving rise to further developments
of carrying out investigations in the area. It also, inci- of theory.
dentally, stimulated the realization in hardware form There is also something of a two-way process at
of the modern general purpose digital computer. the level of concrete systems in that cybernetics is in
A successful programme for a Turing machine- part, as implied by its definition, a comparative
that is, a programme which inevitably leads to a study of animals and machines. On the one hand,
given end-product being achieved-is an example biological processes are studied in order to discover
of what are called 'effective computational pro- ways of designing more sophisticated (purposeful,
cedures' or algorithms. Much of automata theory, intelligent, adaptive, flexible) machines for various
therefore, is concerned with discovering whether purposes. On the other hand, study of the way in
there are algorithms for achieving different mathe- which engineering systems (like computers) achieve
matical and logical ends. various ends (and also the building of special-purpose
Another type of theoretical machine used in hardware models) may give insight into the way in
automata theory is the network. This type of auto- which organisms achieve similar ends. In this way,
maton was first introduced into the theory by continual interaction between the study of engineer-
McCulloch and Pitts [9]. A network consists of a ing and biological systems is mutually beneficial to
set of interconnected elements, information being both. Typically, however, this interaction will be
able to flow under appropriate conditions between mediated by ideas at the more abstract level of
elements so connected. The network as a whole acts cybernetic theory. In doing this cybernetics is not
to transform information inputs of various spatial so much reducing man to the level of 'machines'
and temporal configurations into various outputs. (in the conventional sense) as elevating machines to
The algorithm of a network is not represented by a the level of man.
programme as it is with a Turing machine but is The whole situation, representing the relations
implicit in its structure. between the three different poles of interest of cyber-
netics, is represented in Fig. 5. The six relationships
that are involved are listed in the table below this
ID. THE STRUCTURE OF CYBERNETICS
figure.
So far the impression may have been given that Let us look at some examples of cybernetic re-
all of cybernetic research is devoted to developing search directed towards the ends of pure and applied
various mathematical and logical ideas for their own science.
sake. If this was the case, cybernetics would not A widely used technique in pure science is that of
have the pervasive influence which it is coming to simulation, of making a model of the process being
have across a wide variety of subject-matters studied in order to understand it better. A cyber-
including engineering, biology, physiology, medi- netic model of a biological or psychological process
the work of art in turn communicating to an audience feeling that cybernetics is in some way bound up
through another noisy channel (e.g. incorrect with exciting intellectual developments in other
interpretation in music and drama, poor translation fields and that in some obscure way it points to the
in literature, damage to a piece of sculpture, poor world of the future, and also possibly a feeling that
lighting of a picture and the various shortcomings a rapprochement between science and art would be
of the audience itself, including the noise caused by beneficial to both and that cybernetics represents an
preconceptions, physical incapacities, span of ideal vehicle for such a rapprochement. None of this
attention, etc.); finally it becomes involved in the explains why artists have become attracted to some
interacting emotional and intellectual processes in cybernetic ideas, like the idea of feedback, rather
the nervous systems of the spectators themselves. than other ideas. One cannot help but feel that if the
And this whole situation is one which involves many idea of algorithms had been adopted by artists in-
feedback processes including those between the stead, then a quite different, more ordered, more deli-
artist and the work of art he is in the act of creating, berate, kind of cybernetic art would have emerged.
between the work of art and its audience, and In this section I shall, then, be simply reporting on
between the audience and the artist through criti- the ways in which cybernetics happens to have
cism in the short-term and, in the long term, through excited and influenced artists up to the present,
the giving or withholding of financial support. A rather than on what one might consider rigorous
more precise and detailed understanding of this inferences to be made from cybernetics to art. It is
whole control and communication situation might possible to list these developments under three
help the artist himself in due course to improve his closely related headings.
technique. 1. The idea of machines as works of art. One way
As can be seen, then, some of the concepts of in which cybernetics may have influenced art is
cybernetics are directly applicable to art. Mean- through helping to break down the widespread
while much cybernetic-oriented research is helping feeling that there is some necessary antagonism
to unravel the mysteries of human thinking and com- between art and machinery. If it had not been for the
munication through, for example, computer models demonstration by cyberneticians of how intelligent
of problem-solving, pattern-recognition and so on. and purposeful machine behaviour can be and how
This may in due course help to provide a more arbitrary the living/non-living distinction is for
substantial insight into art. A few workers like many purposes, it is conceivable that the negative
Pierce [13], Goldacre [14], Mueller [15] and, connotations towards machinery might have been so
especially, Moles with his book on Information great that, for example, kinetic art would have taken
Theory and Esthetic Perception [16] have already longer to develop.
started to investigate art directly using cybernetic 2. Machines to create works of art. Some artists
ideas and techniques. In such cases the temptation have thought in terms of not only creating works of
to try to reduce art to something simpler has been art in the form of machines but creating machines to
avoided: it is accepted as complex in its essence. create works of art. Many examples of this were
The Gestalt view is one which is entirely consistent included in the exhibition of 'Cybernetic Seren-
with cybernetics, except that cybernetics now dipity' which took place in London in 1968 and
provides techniques unavailable to the Gestaltists for represents a high point in the early development of a
a precise and objective treatment of what one might self-consciously cybernetic art [18]. Examples
term 'non-decomposable' systems. Indeed, I have include pendulum drawing machines and Tinguely's
argued elsewhere [17] that the work of art itself may painting machines. With Tinguely's machines both
be conceptualized from a cybernetic point of view the machine in action and the results of this action
as a complex dynamic system which in certain of its may be regarded as art. Another notable example of
aspects may be similar to the functioning of a a machine-artist is the robot painter of Hoenich
developing organism. [19].
In this category we must include computer pro-
Creating works of art grammes written to simulate (and even improve
We now come to the relationship of immediate upon) the work of the artist. Examples of this in-
interest to practising artists: the use of cybernetics clude computer-generated visual patterns, e.g. Noll
in the creation of works ofart. Although cybernetics [20] and Mitchell [21], programmes to write poetry,
has been used in a technical sense in, to give the major e.g. McKinnon Wood and Masterman [22] and
example, programming computers using informa- programmes to compose music. As far as the latter is
tion theory to compose music, its main influence has concerned, the work of Hiller, Isaacson and Baker
been rather different. That is, it appears to have [23-25] represents probably the most detailed and
captured the imagination of many artists, in a serious work yet on computer-generated art of any
general way, and so stimulated new works of art or kind and is related to their analysis in statistical and
even new types of art. (It may also have changed the information theory terms of different types of music.
attitude of the spectator to some degree.) The Many other examples of computer art will be found
reasons for all this are none too clear. But one can in the catalogue of the 'Cybernetic Serendipity'
list the following speculative reasons: the promise exhibition [18).
that cybernetics holds out of taking art seriously (as 3. Art as a process. The emphasis of cybernetics
mentioned at the beginning of this article), the on process and change may have been one of the
factors generating an increasing feeling among involve elements of art. But cybernetics appears to
artists that art should be regarded as a process rather generate art even in its pure science aspects. It
than as the production of static objects. This feeling produces, especially through the process of model
has manifested itself in a number of ways including building, entities which often seem to possess
the production of works of art which are imper- aesthetic as well as scientific value.
manent, the advent of the' happening' as an art form Of course it could be argued that pure science and
and the deliberate and creative utilization in some art are in any case not as incompatible as they are
works of kinetic art of the participation of the often assumed to be and that the same processes are
spectator, i.e. offeedback between the spectator and involved in both. Koestler's idea of 'bisociation'
the work of art. Ascott has argued forcefully [26] for example demonstrates that creative activity in
that the aim of art should be the process-oriented all fields may be essentially the same [27]. Never-
one of producing further artistic activity in the theless there could be a particularly close relation-
observer, and that this can be achieved by both the ship between cybernetic model-building and artistic
stimulation of art objects and also by certain kinds creation in that the cybernetician as modeller (at least
of social interaction, especially a type of social in hardware) is creating as well as discovering, is
interaction which constitutes what he calls a expressing his ideas in concrete terms and, in doing
'cybernetic art matrix'. so, has a flexible choice of methods and materials
One of the consequences of the developments which to some extent allows him to express his own
under these three headings is the blurring of some preferences and individuality. It is my own impres-
of the traditional distinctions between the work of sion that a successful model often involves a
art and the system which creates the work of art, combination of elegance and satisfying complexity
and between the work of art and the system which which together are irresistibly pleasing. It could be
observes the work of art. How far cybernetics has said that models like Ashby's Homeostat [28) and
really been instrumental in bringing this about and Grey Walter's mechanical tortoises [IO] were among
how far other major factors have been involved is the first, and are still among the most satisfying,
impossible to estimate. pieces of kinetic art. It is perhaps not without
significance that in introducing the exhibition of
Cybernetics as art cybernetic serendipity, Jasia Reichardt was able to
As well as the two kinds of relationships outlined point out how difficult it was, without checking the
above between cybernetics and art, there is possibly notes relating to the works, to know whether one
a third and even more intimate kind of relationship: was looking at something made by an artist, en-
it may be the case that cybernetics is, in part, an gineer, mathematician or architect and she could
art form as well as being a science. In a sense most easily have added biologist and psychologist to the
applied branches of science (e.g. civil engineering) list.
ACKNOWLEDGEMENT
I should like to thank Mr. Alvin Lewis for his drawing of a Watt governor shown in Fig. 3.
REFERENCES
1. N. Wiener, Cybernetics (New York: John Wiley, 1948).
2. C. E. Shannon and W. Weaver, The Mathematical Theory of Communication (Urbana:
University of Illinois Press, 1949).
3. H. Quastler (Ed.), Information Theory in Biology (Urbana: University of Illinois Press,
1953).
4. M. J. Apter and L. Wolpert, Cybernetics and Development 1. Information Theory, J.
theoret. Biol. 8,244 (1965).
5. L.A. MacColl, Fundamental Theory of Servomechanisms (New York: Van Nostrand,
1946).
6. J. von Neumann, The General and Logical Theory of Automata, in L. Jeffress (Ed.),
Cerebral Mechanisms in Behavior (New York: John Wiley, 1951).
7. A. M. Turing, On Computable Numbers with an Application to the Entscheidungs-
problem, Proc. Land. math. Soc. 42, 230 (1936).
8. D. Hilbert and W. Ackermann, Grundzuge der Theoretischen Logik (Berlin: Springer,
1928) Chap. 3.
9. W. S. McCulloch and W. Pitts, A Logical Calculus of the Ideas Immanent in Nervous
Activity, Bull. math. Biophys. 5, 115 (1943).
10. W. Grey Walter, The Living Brain (London: Duckworth, 1953).
11. Newell, A. and Simon, H. A., GPS, A Programme that Simulates Human Thought, in
E. A. Feigenbaum and J. Feldman (Eds.) Computers and Thought (New York: McGraw-
Hill, 1963).
La cybernitique et /'art
Resume-Cet article se veut une introduction non technique a la cybernetique, une
etude 'des controles et des communications dans l'animal et dans la machine'. Un
certain nombre de principes cybernetiques fondamentaux sont decrits, y compris ceux
qui traitent de la theorie de l'information (par exemple les notions de message et de
quantite d'information), de la theorie des controles (par exemple l'homeostase, la
contre-reaction et les servomecanismes) et de la theorie des automates (par exempleles
algorithmes, les machines de Turing et les reseaux). L'auteur definit la structure de la
cybernetique, en montrant les relations qui existent entre les travaux realises sur la base
des recents progres faits sur le plan theorique, de la comprehension des processus
organismiques en termes cybernetiques, en particulier par la fabrication de modeles,
et de la construction de systemes 'hardware' capable d'une meilleure adaptation en vue
d'un but precis. Ces systemes 'hardware' sont utilises surtout en bionique, pour
l'application industrielle de la cybernetique, et pour l'intelligence artificielle, laquelle
necessite une programmation heuristique.
L'auteur pense qu'il existe trois sortes de rapports entre la cybernetique et l'art: le
savant peut se servir de la cybernetique pour etudier l'art, l'artiste peut l'utiliser pour
creer des reuvres d'art-la cybernetique est peut-etre meme a l'origine de la notion de
machine en tant qu'reuvre d'art et de machine creant l'art-et enfin la cybernetique
peut etre dans une certaine mesure consideree elle-meme comme une forme d'art a
part entiere.