The Second Cybernetics: Deviation-Amplifying Mutual Causal Processes (Maruyama)
The Second Cybernetics: Deviation-Amplifying Mutual Causal Processes (Maruyama)
The Second Cybernetics: Deviation-Amplifying Mutual Causal Processes (Maruyama)
By MAGOROH MARUYAMA
Since its inception, cybernetics was more or less identified as a science of self-regulating
and equilibrating systems. Thermostats, physiological regulation of body temperature,
automatic steering devices, economic and political processes were studied under a general
mathematical model of deviation-counteracting feedback networks.
But they have one essential feature in common: they are both mutual causal systems, i.e.,
the elements within a system influence each other either simultaneously or alternatingly.
The difference between the two types of systems is that the deviation-counteracting
system has mutual negative feedbacks between the elements in it while the deviation-
amplifying system has mutual positive feedbacks between the elements in it.
1
Since both types are systems of mutual causal relationships, or in other words systems of
mutual feedbacks, they both fall under the subject matter of cybernetics.
But since the deviation-counteracting type has predominantly been studied up till now
under the title of cybernetics, let us consider its studies the first cybernetics, and call the
studies of the deviation-amplifying mutual causal relationships "the second cybernetics."
The deviation-counteracting mutual causal process is also called morphostasis", while the
deviation-amplifying mutual causal process is called "morphogenesis." (For a technical
treatment of the subject, see my article "Morphogenesis and Morphostasis," Methodos, IS,
no. 48,1960)
Though the second cybernetics as defined here is lagging behind the development of the
first cybernetics at the present moment, the germination of the concept of deviation-
amplifying mutual causal process is not entirely new. The concept was formulated in
some fields even before the advent of cybernetics and was applied fruitfully. The field of
economics is a good example.
For many years the economists had claimed that it was useless to try to raise the standard
of living of the lower class, because, they argued, if the income of the population in the
lower class should increase, they would produce more children and thus reduce their
standard of living to the original level; the poor stay poor and the rich stay rich. This was
a morphostatic model of mutual deviation-counteracting between, the income level and
the number of children. This theoretical model led the policy makers to the action of
laisser-faire policy. On the other hand, it was also known that "the more capital, the more
rapid the ratio of its increase"; in other words, the poor become poorer and the rich
become richer - This was a morphogenetic model of deviation-amplifying process.
Subsequently J. Tinbergen and H. Wold have given more elaboration and mathematical
sophistication to the theory of mutual causal process in the theory of economics [1]. More
recently G. Myrdal has pointed out that, while in the economically well-developed
countries the regional, social, and hierarchical differences in economical level tend to
decrease, in the economically underdeveloped regions the difference between the poor
and the rich increases. In an economically well-developed society, transportation,
communication, education, insurance systems, and welfare programs equalize the
economical level throughout the society. In an economically underdeveloped society, on
the other hand, under the laisser-faire policy and free play of market forces, the few
privileged people accumulate more wealth and power while the living standard of the
poor tends to fall. Low standard of living, poor health, and low efficiency in work
aggravate one another. Racial or social discrimination, and other social, psychological and
cultural factors may be added in the "vicious circle." Likewise, between nations, world
free trade is profitable for rich countries and detrimental for poor countries. This
morphogenetic reformulation of the economic theory affects the public policy toward the
direction of planned economy within economically underdeveloped countries and
controlled international trade pi.
2
Myrdal |2) further points out the importance of the direction of the initial kick, which
determines the direction of the subsequent deviation amplification in the planned
economy. Furthermore, the resulting development will be far greater than the investment
in the initial kick. Thus, in the economically underdeveloped countries it is necessary not
only to plan the economy, but also to give the initial kick and reinforce it for a while in
such a direction and with such an intensity as to maximize the efficiency of development
per initial investment. Once the economy is kicked in a right direction and with a
sufficient initial push, the deviation-amplifying mutual positive feedbacks take over the
process, and the resulting development will be disproportionally large as compared with
the initial kick.
Development of a city in an agricultural plain may be understood with the same principle.
At the beginning, a large plain is entirely homogeneous as to its potentiality for
agriculture. By some chance an ambitious farmer opens a farm at a spot on it. This is the
initial kick. Several farmers follow the example and several farms are established. One of
the farmers opens a tool shop. Then this tool shop becomes a meeting place of farmers. A
food stand is established next to the tool shop. Gradually a village grows. The village
facilitates the marketing of the agricultural products, and more farms flourish around the
village. Increased agricultural activity necessitates development of industry in the village,
and the village grows into a city.
This is a very familiar process. But there are a few important theoretical implications in
such a process. On what part of the entire plain the city starts growing depends on where
accidentally the initial kick occurred - The first farmer could have chosen any spot on the
plain, since the plain was homogeneous. But once he has chosen a spot, a city grows from
that spot, and the plain becomes inhomogeneous. If a historian should try to find a
geographical "cause" which made this spot a city rather than some other spots, he will fail
to find it in the initial homogeneity of the plain. Nor can the first farmer be credited with
the establishment of the city. The secret of the growth of the city is in the process of
deviation-amplifying mutual positive feedback networks rather than in the initial
condition or in the initial kick. This process, rather than the initial condition, has
generated the complexly structured city. It is in this sense that the deviation-amplifying
mutual causal process is called "morphogenesis."
3
Aspects of Causality
A sacred law of causality in the classical philosophy stated that similar conditions
produce similar effects. Consequently, dissimilar results were attributed to dissimilar
conditions. Many scientific researches were dictated by this philosophy. For example,
when a scientist tried to find out why two persons under study were different, he looked
for a difference in their environment or in their heredity. It did not occur to him that
neither environment nor heredity may be responsible for the difference - He overlooked
the possibility that some deviation-amplifying interactional process in their personality
and in their environment may have produced the difference.
In the light of the deviation-amplifying mutual causal process, the law of causality is now
revised to state that similar conditions may result in dissimilar products. It is important to
note that this revision is made without the introduction of indeterminism and probabilism.
Deviation-amplifying mutual causal processes are possible even within the deterministic
universe, and make the revision of the law of causality even within the determinism.
Furthermore, when the deviation-amplifying mutual causal process is combined with
indeterminism, here again a revision of a basic law becomes necessary. The revision
states:
A small initial deviation, which is within the range of high probability, may
develop into a deviation of very low probability or more precisely, into a deviation
which is very improbable within the framework of probabilistic unidirectional
causality.
Not only the law of causality, but also the second law of thermodynamics is affected by
the deviation-amplifying mutual causal process. Let us return to the example of the
growth of a city in an agricultural plain - The growth of the city first increases the internal
structuredness of the city itself. Secondly, it increases the inhomogeneity of the plain by
its deviating from the original prevailing condition. Thirdly, the growth of a city at a spot
may have an inhibiting effect upon the growth of another city in the vicinity, just as the
presence of one swimming pool may discourage an enterpriser from opening another pool
right next to it, and just as the presence of large trees inhibits with their shades the growth
of some species of small trees around them. A city needs a hintergrund to support it, and,
therefore, cities have to be spaced at some intervals. This inhibiting effect further
increases inhomogeneity of the plain.
4
state. The second law of thermodynamics is in this sense a law of decay of structure and
of decay of inhomogeneity.
Evolution
Any process such as biological growth which increases structuredness and inhomogeneity
was against the second law of thermodynamics and was an embarrassing problem for
scientists. This embarrassing question was temporarily ignored by the argument that an
organism is not an isolated system. But what process and principle make it possible for an
organism to increase its structure and to accumulate heat was never squarely answered.
Now, under the light of deviation-amplifying mutual causal process this mystery is
solved.
First, there is deviation-amplifying mutual process between the mutations and the
environment. For example, suppose that some mutants of a species can live at a lower
temperature than the "normal" individuals. Then the mutants may move to an
environment which is colder. Further mutations occur - Some of the mutants are unfit for
the low-temperature environment and die off. But some other mutants are able to live in a
much colder climate than their parents. They move to a still colder environment. The cold
climate eliminates any new mutants that are unfit for cold climate. The "average"
individuals of the survivors are then fit for cold climate. The chances of the species, or at
least some members of the species, to move to a still colder environment are now greater
than before. Thus the selection of, or accidental wandering into, a certain type of
environment and the direction of survivable mutations amplify each other.
Not only the organism may move into a new environment, but it may also create its own
environment. Homo sapiens is a typical example. A "civilized" man lives in an
environment which he created, and which is relatively free from the bacteria of certain
diseases such as typhoid. His resistance against typhoid decreases as a result of living in
such an environment. The decreased resistance necessitates him to make his environment
more germ-free. This decreases his resistance further.
5
Thirdly, the intraspecific selection has a deviation-amplifying effect, for example many
animals prefer supernormal (above-average) members of its species to normal members
for mating and for carrying on other cooperative activities. By giving more responses to
supernormal stimuli than to normal stimuli, the members of blic species amplify, by
favoring supernormal mutants, the deviation in the direction of super-normality. The
deviation in turn may increase either the number and the intensity at the members'
responses to the supernormal stimuli, or the level of the supernormality of the members'
characteristics.
In any case, roan is responsible for his own evolution because of his capacity to create his
own culture which is his environment and to choose his criterion of supernormality.
Incidentally, to "create" is no longer a concept which violates the law of physics. As we
have seen, creation ex nihilo, or rather almost ex nihilo, is scientific ally possible because
the secret is no longer in the Prime Mover or in the Creator, but in the process of
deviation-amplifying mutual positive feedback network.
Many times cooperation between individuals facilitates life [4], and therefore the
existence of species facilitates the life of the individual. The sexual reproduction, as
compared with asexual reproduction, acts as a stabilizer of the species.
6
Here we have seen that mating may amplify the deviations or stabilize the species. This is
not a contradiction. Whether mating is deviation-amplifying or deviation-counteracting
depends on whether the inbreeding component or the interbreeding component is
predominant. S. Wright [5] has made this relationship clear. When the mutation rate is
high as compared with the population size, random matings will produce more inbreeding
than interbreeding, and the deviation-amplifying effect is predominant. On the other hand,
when the population size is large as compared with the mutation rate, random matings
will produce more interbreeding than inbreeding, and the deviation-counteracting effect is
predominant. The direction of deviation in a small population with a high mutation rate is
unpredictable since it depends on the initial kick which is random. But once the deviation
started, it is systematically amplified in the same direction.
When the mutation rate is neither too high nor too low as compared with the population
size, neither inbreeding nor interbreeding predominates. The result is neither deviation-
amplifying nor deviation-counteracting, but a combination of both which results in
random drift. At one time, a random initial kick produces a deviation in a certain
direction. Deviation-amplification takes over and this deviation is amplified consistently
in the same direction. But this does not last very long. Soon, deviation-counteracting takes
over, and the population becomes fixed at a certain point of deviation. After a while,
another random initial kick produces a deviation in a new direction. Deviation-
amplification takes over again and the population drifts consistently in this direction for a
while. But soon, deviation-counteracting takes over and the population becomes fixed at a
new point in the new direction of deviation. Then another initial kick starts a deviation in
a new direction. The process repeats itself with unpredictable drifts.
The maximum speed of evolution is found, not in one colony with a high rate of mutation
and a small size of population, but in the interaction between colonies which have a
moderate mutation rate. When the mutation rate is too high as compared with the
population size, the mutant characteristics may amplify themselves at a speed beyond the
possibility of finding a new environment and new intraspecific and interspecific
ecological conditions which are suitable for the mutant characteristics, and beyond the
possibility of allowing other variations of mutants, which have characteristics greater in
survival value, to develop. The species may become extinct, or it may reach the limit of
mutability and become fixed there, or the mutant characteristics may become so dominant
and homogeneous as to become deviation-counteracting.
A moderate mutation rate produces a more viable and changeable evolution. Moreover,
when there are occasional exchanges of immigrants between colonies, the introduction of
a new strain, which has proved to be viable in one colony, into another colony has the
same effect as producing viable mutants, and tends to favor evolution.
This seems to be true of human cultures also. When there is enough separation, not
necessarily geographical, between cultures to allow them differentiation and variety, with
enough exchange between them to allow mutual enrichment and new combinations, the
human civilization seems to progress most efficiently.
7
As we have seen, evolution is deviation-amplifying in several ways. For this reason we
called evolution "phylogenetic morphogenesis." But more traditionally, "morphogenesis"
is used for ontogenesis, i.e., for the development of the embryo into an adult individual. Is
the new usage of the word "morphogenesis," meaning deviation-amplifying mutual causal
process, in any way contradictory to the traditional usage? By no means so. As we see in
the following, the development of the embryo also involves deviation-amplifying mutual
causal process.
There is one basic difference between ontogenesis and phylogenesis. Mutation and natural
selection, which are the basis of phylogenesis, are absent in ontogenesis. In fact, while
phylogenesis has the randomness of mutation, ontogenesis lacks this randomness and
seems to be based on a strictly detailed, deterministic planning. But this detailed planning
is generated within the embryo by a deviation-amplifying mutual causal process in a
deterministic scheme.
Let us imagine, for the sake of simplicity, a two-dimensional organism. Let us further
imagine that its cells are squares of an equal size. Let us say that the organism consists of
four types of cells: green, red, yellow, and blue. Each type of cell reproduces cells of the
same type to build a tissue. A tissue has at least two cells. The tissues grow in a two-
dimensional array of squares. Let us give a set of rules for the growth of tissues:
2. Both ends of a tissue grow whenever possible, by reproducing one cell per unit time in
a vacant contiguous square. If there is no vacant contiguous square at either end, that end
stops growing. If there are more than one vacant contiguous squares at either end, the
direction of the growth is governed by the preferential order given by Rules 3, 4, and 5.
3. If, along the straight line defined by the end cell and the penultimate cell (next to the
end cell) there are less than or equal to three cells of the same type (but may be of
different tissues) consecutively, the preferred direction is along the same straight line. If
that direction is blocked, follow Rule 5.
8
4. If, along the straight line denned by the end cell and the penultimate cell, there are
more than or equal to four cells of the same type (but may be of different tissues)
consecutively, the preferred direction of the growth is a left turn. If a left turn is
impossible, make a right turn.
5. If, when a straight growth is preferred, the straight growth is impossible because the
square ahead is already occupied, do the following:
If the square to which the straight growth would take place is filled with a cell of the same
type as the growing tissue, make a left turn. If the square ahead is filled with a cell whose
type is different from that of the growing tissue, make a right turn.
6. The growth of the four types of tissues is timewise out of phase with each other: green
first, red second, yellow third, and blue last within a cycle of one unit time.
Fig. 1
9
Rules 2, 3, 4, and 5 can be diagrammed together as in Figure 1. Using these rules, we can
compute the growth of the tissues when the locations of the initial tissues are specified.
For example, let us say that, in Figure 2, only the squares marked by 1's are filled with
cells of the types indicated by the colors. At the end of the second unit time, the squares
marked by 2's are filled. And at the end of the nth unit time, all squares marked by
numbers less than n are filled. At the end of the 44th unit time, all tissues have completed
the growth, and the organism has attained its full differentiation.
Fig. 2
In this example we started with four tissues of the minimum length, one tissue of each of
the four types. But already the result is a fairly complex structure. If we start with a
slightly larger number of tissues at the beginning, the resulting pattern becomes
disproportionately more complex.
The amount of information to describe the resulting pattern is much more than the amount
of information to describe the generating rules and the positions of the initial tissues. The
pattern is generated by the rules and by the interaction between the tissues. In this sense,
the information to describe the adult individual was not contained in the initial tissues at
the beginning but was generated by their interactions.
Besides generating information, this type of process has two additional features. First, it is
strictly deterministic. When the locations of the initial tissues are identical in two
embryos, the resulting adults, no matter how complex, will be exactly identical.
10
Secondly, it is in most cases impossible to discover the simple generating rules after the
pattern has been completed, except by trying all possible sets of rules. When the rules are
unknown, the amount of information needed to discover the rules is much greater than the
amount of information needed to describe the rules. This means that there is much more
waste, in terms of the amount of information, in tracing the process backwards than in
tracing it forward. A geneticist would waste much time and energy by trying to infer the
characteristics of the embryo from the characteristics of the adult organism. It would be
more profitable to perform experiments in embryonic interference and embryonic
grafting. The same is true also for the study of other deviation-amplifying mutual causal
processes such as history or mental illness.
Since information is generated by the interaction between various parts of the embryo, it
is not necessary for each part of the embryo to contain information regarding the body
part it is destined to become. It partly receives the information from other parts of the
embryo and from its relationships to them. For example, in the embryo of certain species,
if the part which would become an eye is transplanted at an appropriate stage of the
embryonic development into the part which would become skin, the eye-tissue becomes
skin. It receives information for its growth from its surroundings.
We have discussed mainly the structure-generating aspect of the interaction between the
parts of the embryo. But the interaction has also a structure-stabilizing aspect. Let us look
at the example of the grafted eye tissues again. When they were grafted on skin tissues,
the skin tissues made the would-be-eye tissues into skin tissues, against the possibility
that the would-be-eye tissues might become an eye. This process is "morphostasis" in the
traditional terminology.
Thus, the usage of the words "morphogenesis" and "morphostasis" in the sense of
deviation-amplifying mutual causal process and deviation-counteracting mutual causal
process, respectively, does not contradict their traditional usage in the sense of
ontogenetic structure-generation and structure-stabilization. The new usage not only
extends the old usage, but gives a functional definition in terms of deviation-generating
and deviation-counteracting, and in terms of positive and negative feedback networks.
11
For example, Eisen Iron Co. makes iron from iron ore. Dexter Tool Co. manufactures
tools made of iron. Dexter buys iron from Eisen, and Eisen buys tools from Dexter. There
is some mutual relationship between the two companies. But suppose Dexter buys iron
from several companies. When Eisen's output drops, Dexter's purchase of iron from other
companies increases. When Eisen's output goes up, Dexter's purchase of iron from other
iron companies decreases. The amount of tools Dexter can supply to Eisen does not
depend on the amount of iron Dexter buys from Eisen. Furthermore, Dexter has other
customers besides Eisen. Whether Eisen buys no tools or 10,000 tools from Dexter does
not matter much to the operation of Dexter. In this- case, though there is traffic of
merchandise in both directions between Eisen and Dexter, the amounts of traffic in two
directions have no mutual causal relationship.
Suppose that, suddenly, some ship industry develops in the vicinity, and both iron and
tools are in great demand. Consequently, both Eisen's output and Dexter's output increase
simultaneously. But this simultaneous increase was not caused by a mutual causal
relationship between Eisen and Dexter, but by a third element which is the ship industry.
On the other hand, if the amount of Dexter's output depends on the amount of Eisen's
output and varies with it either in the same or opposite direction, and the amount of
Eisen's output depends on the amount of Dexter's output and varies with it either in the
same or opposite direction, then there is a mutual causal relationship between Eisen's
output and Dexter's output.
Mutual causal relationships may be defined between more than two elements. Let us look
at the following diagram. The arrows indicate the direction of influences. + indicates that
the changes occur in the same direction, but not necessarily positively. For example, the +
between G and B indicates that an increase in the amount of garbage per area causes an
increase in the number of bacteria per area. But, at the same time, it indicates that a
decrease in the amount of garbage per area causes a decrease in the number of bacteria
per area. The – between S and B indicates that an increase in sanitation facilities causes a
decrease in the number of bacteria per area. But, at the same time, it indicates that a
decrease in sanitation facilities causes an increase in the number of bacteria per area.
12
Figure 3
As may be noticed, some of the arrows form loops. For example, there is a loop of
arrows from P to M, M to C, and C back to M. A loop indicates mutual causal
relationships. In a loop, the influence of an element comes back to itself through other
elements. For example, in the loop of P-M-C-P, an increase in the number of people
causes an increase in modernization, which in turn increases migration to the city, which
in turn increases the number of people in the city. In short, an increase in population
causes a further increase in population through modernization and migration. On the other
hand, a decrease in population causes a decrease in modernization, which in turn causes a
decrease in migration, which in turn decreases population. In short, a decrease in
population causes a further decrease in population through decreased modernization and
decreased migration.
Whatever the change, either an increase or a decrease, amplifies itself. This is so when we
take population as our criterion. But the same is true if we take modernization as a
criterion: an increase in modernization causes a further increase in modernization through
migration and population increase; and a decrease in modernization causes a further
decrease in modernization through decreased migration and decreased population. The
same holds true if we take the migration as the criterion.
In a loop, therefore, each element has an influence on all other elements either directly or
indirectly, and each element influences itself through other elements. There is no
hierarchical causal priority in any of the elements. It is in this sense that we understand
the mutual causal relationships.
13
Let us take next the loop P-G-B-D-P. This loop contains a negative influence from D to P.
In this loop, an increase in population causes an increase in the amount of garbage per
area, which in turn causes an increase in the number of bacteria per area, which in turn
causes an increase in the number of diseases, which in turn causes a decrease in
population. In short, an increase in population causes a decrease in population through
garbage, bacteria and diseases. On the other hand, a decrease in population causes a
decrease in garbage, bacteria and diseases, and thus causes an increase in population. In
this loop, therefore, any change in population is counteracted by itself. Likewise, any
change in the amount of garbage per area is counteracted by itself. The mutual causal
relationship in this loop is a deviation-counteracting mutual causal relationship. Such a
deviation-counteracting process may result in stabilization or oscillation, depending on
the time lag involved in the counteraction and the size of counteraction.
Let us further consider the loop P-M-S-D-P. This loop has two negative influences. An
increase in population causes an increase in modernization, which in turn causes an
increase in sanitation facilities, which in turn causes a decrease in the number of bacteria
per area, which in turn causes a decrease in the number of diseases, which in turn causes
an increase in population. This is therefore a deviation-amplifying loop. Two negative
influences cancel each other and become positive in the total effect.
Not only are there deviation-amplifying loops and deviation-counteracting loops in the
society and in the organism, but also under certain conditions a deviation-amplifying loop
may become deviation-counteracting, and a deviation-counteracting loop may become
deviation-amplifying. An example is the principle of diminished return. An increase in
investment causes an increase in capital, and an increase in capital makes more
investments possible. Before the profit reaches a certain level the effect of income tax is
negligible. But, as the profit becomes greater, the influence of income tax becomes
greater and eventually stabilizes the size of the capital.
A culture may follow a similar process. Sometimes one may wonder how a culture, which
is quite different from its neighbouring cultures, has ever developed on a geographical
background which does not seem to be in any degree different from the geographical
conditions of its neighbours. Most likely such a culture had developed first by a
deviation-amplifying mutual causal process, and has later attained its own equilibrium
14
when the deviation-counteracting components have become predominant, and is currently
maintaining its uniqueness in spite of the similarity of its geographical conditions to those
of its neighbours [g].
The second cybernetics is useful also in such fields as psychiatry. In many cases,
interpersonal conflicts are generated by mutual deviation-amplification between persons
and are later maintained at the deviated (but not necessarily deviant) pattern. Mutual
amplification may occur within a person, for example, between loss of self-confidence
and poor performance in a neurotic person. The established pattern, no matter "how"
deviated, is not necessarily "pathological" if it enables a constructive life. But, if the
pattern results in a reduction of conflict-free, constructive energy, then a therapy becomes
necessary. The therapy aims at breaking the stabilized unsatisfactory pattern and at
initiating a new deviation-amplification in the direction of developing a satisfactory
pattern.
The second cybernetics will be useful also in the technological fields such as in the design
of a machine which invents. A trial-and-error machine is inefficient because it has no
directivity. But it has a great flexibility. A deviation-amplifying inventing machine, on the
other hand, works in the direction specified by the initial kick, and, for this reason, is
efficient. It is not built for any specific direction, because the direction is a variable which
is specified by the initial kick. In this sense it is flexible. But in another sense it is not
flexible because once the direction is set, it will persist in that direction. A machine that
incorporates randomness, deviation-amplification and deviation-counteracting may be
both efficient and flexible. It can search for all possibilities. It can try to amplify certain
ideas in various directions. It can stay at a relevant idea (which may change from time to
time during the invention) and bring back to it other ideas for synthesis. In fact, openness
to strange hunches, ability to elaborate on them and to bring them back to a synthesis are
what is found in the process of human creative minds [7].
The elaboration and refinement of the second cybernetics belong to the future, and we
may expect many fruitful results from them.
15
REFERENCES
-(1932). Roles of mutation. Proc. 6th intemat. Congr. Gen., 1, pp. 356-366.
6. Maruyama, M. (1961). The multilateral mutual causal relationships among the modes of
communication,
- Sociometric pattern and the intellectual orientation in the Danish culture. Phylon, 22, no. 1,
pp. 41-58.
-(1957). Originality in Relation to Personality and Intellect. J. Personality, 25, no. 6, pp. 730-
742.
16