Meeting on Cybernetics and Systems Research
and the Art of Living
Ernst von Glasersfeld
I had intended to begin
this talk, as one so often does to break the ice in plenary
sessions, by being funny about some trivial matter. But Professor
Trappl's sad announcement has put me in a state
Having been traveling
for the last two weeks, I had not heard of Gordon Pask's death
until this moment - and the loss is far too great to grasp.
Gordon had generously supported me some thirty years ago when I
entered academia in the United States, and there is no end to the
help his ideas gave me in developing my thinking. Now I am
crushed by the profound regret that a friend has gone whom I have
not thanked nearly enough for the inspiration he provided.
So let me start my talk
by saying I sincerely hope that a knowledgeable scholar will
collect and record the history of the first ten or fifteen years
of cybernetics before too many of the fascinating personal and
intellectual details of its inventors are irrevocably forgotten.
I, unfortunately, am anything but a knowledgeable scholar. I only
had by chance some opportunities to witness the development of
this revolutionary discipline as an enthusiastic outsider who was
profoundly influenced by it.
The point that struck me
at the outset was that the founding fathers, especially Norbert
Wiener and Warren McCulloch, thought of their enterprise not
merely as a technique but also as a new and powerful approach to
philosophy. But the two interests quickly separated, and the
spectacular mathematical and technological successes of
cybernetics have until recently all but obscured the
Judging by the list of
symposia announced in our present program, the focus of this
conference, too, is largely upon technical innovations and new
applications in a variety of domains. I am therefore taking
something of a risk by talking about cybernetics and the art of
living. I can only hope that in the end you will forgive me.
I want to begin by
recalling a statement Warren McCulloch made in a lecture at the
University of Virginia in 1948. "To have proved an hypothesis
false," he said, "is indeed the peak of knowledge." The "real"
world does not show us when we are right, but when we are wrong.
All we experience are the constraints that prevent us from acting
in certain ways.
When I read McCulloch's paper in
the early sixties, I had long been thoroughly dissatisfied with
traditional epistemology. The statement was a revelation. A
little later I came across Gregory Bateson's paper on "Cybernetic explanation"
(1972), in which he explained that what makes cybernetics
different from other scientific enterprises is the fact that it
operates with constraints rather than with efficient causes. He
cited the theory of evolution as a prime example, because natural
selection only eliminates what does not fit. The
properties that allow an organism to survive are not created by
selection but are the result of random variations.
The theory of evolution
applies to species and to the heritable properties that
characterize them. Species have no knowledge, they are what they
are, and the organisms that compose the species either have the
properties that enable them to survive, or they don't.
But there are many
organisms that we call intelligent because they are able to learn
from their experience. What they learn, of course, is not
heritable - but it may help them to survive. They
learn to avoid some constraints of the world which they
experience. In other words, they learn to fit better between the
obstacles their environment puts in their path.
If one takes this idea
of fitting and applies to the problem of how we gain the
knowledge on the basis of which we try to lead our lives, one
comes to a theory of knowing that is radically different from
most of the epistemologies of traditional philosophers.
Because Vienna is
something of a stronghold of "Evolutionary Epistemology", I
want to stress that the model of cognition I am talking about
does not fit that mold. The main reason is that knowledge in the
cybernetical model is never knowledge of a real world. It is
knowledge of what one can or cannot do. The obstacles that
manifest themselves as constraints are merely the limits of the
space that is accessible to experience. They are relative to the
organism's way of experiencing, not representations
of an independent reality. Knowledge, in this theory, is
therefore not a picture of reality, but a repertoire of actions
and thoughts which in past experience have turned out to be
In this sense, this
theory of knowing replaces the notion of true
representation with the notion of viability. - Rather
than go into the details of that theory which is laid out
elsewhere (cf. Glasersfeld, 1995), I shall give you tangible
examples of how I see it.
I spent the last ten
days in Chamonix and the mountains around Montblanc. It was a
nostalgic experience, because until forty years ago I spent many
a spring skiing on the glaciers of the Alps. In those days there
were no cable cars and other mechanical devices to bring
thousands of skiers to the tops of mountains. You were alone
there, and if you wanted to ski down a mountain, you first had to
In retrospect, it struck
me as a good example of dealing with constraints. If you wanted
to go up or down a mountain, you had to look at it rather
carefully. You wanted to reach the summit - but it would have been a mistake
simply to look for an easy way up. As an experienced mountaineer,
you first of all figure out where you must not go.
You try to see possible avalanches, ice breaks, crevasses, and
other fatal constraints. Only when you have, so to speak, blocked
out the treacherous parts of the mountain, would you begin to
plan your way up. At this point, you do make choices, but you
make them within the space left between the
mountains constraints. To "know" a mountain means to know where,
on its slopes, you are relatively safe; it means to have learned
the viable paths.
One can come to the
notion of viability in many ways. One of them is the principle
that Leibniz and Maupertuis formulated a long time ago: The
principle of least action or, respectively, of least resistance.
Water will follow the
pull of gravity as far as it can. When it rains on a hill, the
rain water runs down wherever it finds a way. If it is stopped,
it will collect and eventually flow over or around the obstacle.
This, in turn, changes the shape of the hill, makes new paths
viable, and encounters new constraints.
But let me return to the
origins of cybernetics. Besides shifting the focus of attention
from causes to constraints, it brought about another fundamental
change by launching the theory of communication.
When Claude Shannon
formulated the theory mathematically, he was careful to state in
the first two pages of his seminal paper that what he called "information" had
nothing to do with semantics. The impulses that travel in a
communication channel from a source to a sink are changes of some
form of energy. They are "signals" only to those who are in
possession of the relevant code. The code itself is not part of
the transmission. The signals are instructions to select specific
parts of the code. And "information", in the theory of communication,
does not refer to the meaning of the coded elements, but is
simply the measure of how many or how few of the pre-established
elements the signals select.
You probably know all
this - but it is good to remember it when
someone is speaking.
Norbert Wiener provided
the marvelous example of the flower shops that use their own
economical system of communication. If a young man in Vienna
spent a few happy days with an American tourist and, now that she
has left, he wants to deepen the impression he made on her, he
might go to a flower shop and arrange for a dozen red roses to be
sent to her in Los Angeles on her forthcoming birthday. The
flower shop then cables the address, the sender's name,
and a specific number, say 54. By means of two simple digits the
number instructs the receiving shop to select 12 red roses and
the message "Happy Birthday".
Don't think that I have done this so
often that I know the florists' code by heart. I just invented
the number 54. But florists do use such numbers for the election
of specific flowers and all sorts of good wishes and condolences.
My point in mentioning it here is simply that the number is
meaningless unless it is interpreted or decoded by
a person who knows the code. The "information" the number carries in this
context is no more than that it indicates and thus selects a
particular item out of all the messages the florists' code
condition was unfortunately disregarded by nearly all the
linguists who went into a frenzy of excitement when they heard of
Shannon's theory, and they promptly compounded
their confusion by speaking of "Information Theory",
while Shannon had deliberately called it "Theory of Communication".
Human language is, of
course, a communication system and it is therefore quire
enlightening to apply Shannon's theory to it. But language is
also different from all artificial or technical communication
systems. The crucial difference is that in language we do not
start out with a pre-established code, but each of us to learn it
by using it.
You may say this is
nonsense, because we have dictionaries that tell us the meanings
of words. True, we have dictionaries - but how do they tell us the
meanings of the words we look up? They use other words. Just
think for a moment how far you would get if you had to learn
Morse code by trying to use it.
A one-year old child, in
the process of acquiring language is almost in as difficult a
position. I say "almost", because there is an important
difference: the language the child has to learn is constantly
being used by the speakers within the child's field of immediate experience.
An example may help to show what I mean. Let us say a mother says
to her child: "It's time for your bottle. "She
goes to fetch the bottle, puts the nipple to the child's mouth,
and says: "Drink your milk!"
The child begins to suck
(because it sucks everything that touches its lips) and it feels
the liquid in its mouth. No doubt it will form some association
between the sound of the words, the touch of the nipple on its
lips, and the feeling in its mouth. But it will take a great many
other experiences with drinking water or orange juice, with cups
and glasses, and with many other word-sounds, before the child
has sorted out approximate meanings for "bottle", "drinking", and "milk". And the most important aspect
of this learning situation is this: the experiences of bottle,
drinking, and milk, with which the child associates the sounds of
these words, are the child's subjective impressions. They
are neither the mother's nor anyone else's. Nor are
they "things in themselves" or instances of independent
objects in a real world. They cannot be anything but the
impressions this child-subject happens to experience.
Clearly, in the case of
words that are frequently used in everyday language, these
subjective impressions become more or less intersubjective in the
course of linguistic interactions with other speakers. But one
can show that even the commonest words retain a margin of
subjective meaning for each individual speaker. Linguists and
philosophers of language usually subsume this margin under the
term "connotation", and they claim to be able to
separate the subjective component neatly from "objective
denotation". From our point of view, this claim rests
on the illusion that words refer to things in a
real world. In our theory - which we of course consider to
be more adequate -- words, as I hope to have shown, refer to
subjective experiences of the individual language user. The
separation between denotation and connotation thus no longer
involves objectivity, but becomes a question of greater or lesser
fit with the usage of other speakers.
The result of our
investigations in this area is that the meanings of words and
longer segments of language are never "shared"
- with others in the
sense that they could be considered the same for all members of a
language community. All one can say is that among proficient
speakers of a language, meanings are at best compatible
, i.e., they function similarly in most situations.
This is an important
difference from the artificial, technical communication systems.
There, the code that bestows significance to the signals is
established and distributed to the communicators before any
communication takes place. In contrast, a child acquiring its
human language enters into a system that is already in action and
as newcomer it has to establish a code for him- or herself. This
is a laborious process that involves countless trials and errors
and leads at best to viable approximations. In fact, it is an
endless process. No matter how old you are and how long you have
been speaking your language, every now and then you discover that
you have been using a particular word in a way that is not
compatible with the accepted usage of your community. You were
unaware of the idiosyncrasy, simply because the situation in
which the discrepancy becomes relevant had not occurred in your
On the surface, this may
seem to have little to do with the art of living, but it is
obviously a factor in the art of reading. How often do you open a
book on an intellectual subject, and in the first few pages you
come across a statement that seems quite nonsensical. If you are
an impatient reader, you may say to yourself, this author is a
fool - and you put the book away. But if,
instead, you keep in mind that the meanings of words are
essentially subjective constructs of individual language users,
you will tend to withhold judgement. The author, you will say to
yourself, is supposed to be intelligent and therefore it is
likely that what he or she has written makes sense to the writer.
In this case you will make an effort to find out what this sense
could be. Quite often such an effort is worth making, because it
may lead to the realization that the text not only uses some
terms in an unfamiliar way, but also that this new way makes good
sense. Whenever this is what you manage to find, you have leaned
something new - and that, after all, is the deeper
purpose of reading.
How many philosophical
debates could be turned into productive discussions, if only the
participants were not quite so convinced that the meanings they
have associated with words are the only legitimate ones.
How many quarrels
between lovers could be avoided, if one of the two considered
that what the other says may not mean what it appears to mean.
All this, of course,
raises the question of what we mean when we say we
have understood a piece of language. There is still the
wide-spread view that words contain their meaning
the way a book contains pages. If, however, words are printed or
spoken signals in the linguistic communication system, they
cannot convey a fixed meaning. They can only point to and select
whatever the reader or listener has associated with them. And the
conceptual structures that this person has associated with the
given words are abstractions from that individual person's
experience, not from the experience of the writer or speaker. No
doubt each language user's associations have been adapted
and honed by years of linguistic interactions with others, but
the material of which they consist is under all circumstances
What, then, is
I want to suggest that
understanding depends on the sense you make of what is said or
written. If the concepts the words have called up in you, and the
way the sentences have prompted you to relate them, yield a
conceptual network that fits the context created by what came
before and is not countermanded by anything the speaker says or
does now, then you assume that you have understood what he or she
This, of course, is a
simplification. What I called context is usually a hierarchy of
different contextual levels, such as your past experiences with
the speaker or author, your own construction of the experiential
world in general, certain expectations you have formed, and other
things as well. However, the point I want to emphasize is that on
all levels it is a question of fit - not a
question of receiving or reproducing conceptual structures that
originated in the speaker's head. What a speaker or author
wants to say is forever inaccessible - you can only interpret what he
or she actually said.
You may wonder why I
spent so much time talking about language. I had two reasons. The
first is that much of our living is done in conjunction with
others, and language is inseparable from the social context. I
therefore feel that a coherent model of how linguistic
communication works is a great help in managing our social
interactions and thus our life.
It would make our lives
more pleasant, for instance, if every time one is about to shout
at someone: "But I told you so!" one remembered that telling does
not guarantee being understood.
The second reason is
that the principle of interpretation and viable fit that I have
outlined with regard to understanding language is equally
applicable to understanding the world in which we find ourselves
living. We have no more access to an ontological reality than to
the thoughts of another person. All we have to go on is our
experience. In both cases we interpret what we see, hear, and
feel, and we construct models that should enable us to make
At an earlier edition of
this conference, some years ago, I suggested that if the model we
have constructed of the person we live with has served us well
for some time, we tend to believe that it has captured how that
person really is. But sooner or later our companion does
something that we did not expect. This may irritate us, and we
reproachfully say: "You have changed!" - Often
this is not at all the case. The other has merely shown an aspect
we had not incorporated in our model, because no prior situation
has brought it to the fore. Our surprise and our irritation would
be greatly mitigated if we kept in mind that the other we know is
not the other as he or she is, but a model we have constructed on
the basis of nothing but our own experience.
The very same happens to
the scientist who constructs a model of, say, the planetary
system or the universe. If that model works well and provides
useful answers to the questions that are asked, it comes to be
regarded as a true description of reality. But sooner or later
something incompatible is observed, a precession of Mercury or a
beam of light that does not follow a straight line. Such
observations constitute constraints that demolish the viability
of the accepted model. It no longer fits the scientists' expanded
experiential world. At first there usually is some incredulity
and a great deal of reluctance to accept such failure.
Eventually, however, a new model is constructed with the help of
new concepts that make the shocking observations seem normal and
All this is in harmony
with the fundamental principles of our discipline, for
cybernetics is the art of creating equilibrium in a world of
possibilities and constraints. - And I would suggest that this is
also a viable definition of the art of living.
(1972) Cybernetic explanation. In Bateson, Steps to an
ecology of mind, (pp.399-410). New York: Ballantine.
E. von (1995) Radical constructivism: A way of knowing
and learning. London: Falmer Press.
(1948) Through the den of the metaphysician. Printed in
McCulloch, Embodiments of mind, (pp.142-156)
Cambridge (Massachusetts), M.I.T. Press, 1965.
(1948) The mathematical theory of communication. Bell
Systems Technical Journal, 27, 379-423 &