Comments 155

in the absence of such a dictionary, Shannon's law would become a meaningless formal
exercise.
While the meaning of the 'dictionary' is clear enough in the case of human communi-
cation, it is not at all clear in the analogous biological case of the 'genetic code'. It seems
highly inadequate to say that the code eventuates in a set of enzymes; these are very far
from a living organism. It seems clear enough that whatever the dictionary is that corre-
sponds to the genetic code, it must refer to the relation of the code to morphology. Such a
phrase covers up all the problems that stand in the way of a deeper understanding of the
dynamics of living organism; but it merely covers them, it does not solve them.

R eferen ces
Brillouin, L. (1956). Science and Information Theory. London: Academic Press.
Shannon, C. & Weaver, W. (1949). The Mathematical Theory o]" Communication. Urbana,
IL.: University of Illinois Press.

Walter M. Elsasser
Department of Earth and Planetary Sciences
The John Hopkins University
Baltimore, Maryland 21218, USA

Comment by Humberto R. Maturana

The author of this article adequately criticizes the use of the notion of information by
biologists in general, and by sociobiologists in particular. He, however, does not go far
enough; and this, presumably, because, judging by two questions that he asks, he thinks
that there is some fundamental significance in the notion of information that is still hidden.
His questions are ' . . . how the discernible events described by the strict sense of infor-
mation (restriction imposed upon a set of possibilities) interact with those described by the
meaning (metaphorical) sense of information?'; and ' . . . there remains more fundamental
problems, one of which is to understand how the two senses of information interact and,
in fact, become one another?'.
Before Shannon the word information referred to the action of giving form, or of
instructing with knowledge, as if in the interactions each interacting participant specified
what happened to the other participant it acted upon. It was this connotation that Ashby
(1956) rejected when he preferred to use the word variety instead of the word information
in his writings about cybernetics, and it was this connotation that attracted biologists when
they borrowed the word information from the newly developed communication theory
(Shannon & Weaver, 1949) to use in genetics and neurophysiology. Biologists did not under-
stand then that the word information was used by communication engineers to describe
a particular kind of interaction between organizationally isomorphic and structurally
complementary systems, and took it to refer to a mechanism of instructive determinism
that allowed for the formalization and quantification of phenomena in which meaning
seemed to play a central part. They did this even though every biologist knows that instruc-
tive interactions do not take place, seduced perhaps, by the belief that information theory
would permit them such formalization and quantification. Although I had great doubts from
the beginning, I also used the notion of information until I rejected it completely in the late
nineteen-sixties. Whatever the case, many biologists do not seem to have realized yet that
even in communication theory the notion of information is only a descriptive notion and
not an explanatory one.
156 Comments

Let me make this explicit:

Information, in the strict sense indicated in the article that I am commenting, refers, in
communication engineering, to a situation designed and constructed such that a message
may, (i)go between two organizationally isomorphic and structurally complementary
systems, and, (ii)select, from a set of events in the receiver identical to an equivalent set
of events in the emitter, a subset of events identical to the subset of events to which it
corresponds at the emitter. Any mismatch that may occur between the subset 'of events
specified (coded) at the emitter and the subset of events selected (decoded) at the receiver,
is the result of some interference (noise) with the realization of the processes involved in
the communication, and not part of its design. If the expected selection of alternatives is
realized at the receiver, the communication engineer says that total information has been
transmitted, even though no instructive interaction has taken place. In fact, no instructive
interactions can take place in the physical space (see Maturana, 1978).
A description in terms of the strict sense of information, can be applied only to fully
specified situations of interaction between systems with organizational isomorphism and
structural complementarity that can be assimilated to the engineering situation of communi-
cation. Such a description, however, does not provide an explanation for the occurrence of
the organizational isomorphism and structural complementarity between systems that makes
possible the phenomena connoted by the description, and which is what in biology requires
an explanation. In human-made communication systems such a situation arises through their
design and intentional construction. In biological systems such a situation is the result of
phylogenic and ontogenic histories of interactions in which the selective mutual triggering
of structural changes with the conservation of the mutual structural coupling (mutual
adaptation) of the involved organisms (or systems), is the mechanism for its origin. Infor-
mational descriptions do not grasp this.
The metaphorical sense of information (meaning) is also descriptive, and refers to a
relation of correspondence between non-intersecting phenomenal domains seen by an
observer. Therefore, the notion of meaning, because it does not refer to a mechanism
capable of giving rise to the seen relations, has no explanatory value either; it only connotes
them in a domain of descriptions. Meaning is a relation specified through a descriptive
reflexion made by an observer and, therefore, can only take place in language. Professor
Fedanzo is right when he says that in my article with G. Guiloff (Maturana & Guiloff, 1980)
consensus can be related to meaning if one contemplates it. Yet, we do not talk in that
article about information in either sense, and we do not do so because it is not pertinent.
In fact, we do not use consensus as an explanatory notion; we only point to consensuality
as an expression of the phenomenon of intelligence as a case of structural coupling. Further-
more, it should be apparent in that article that consensus and structural coupling lie in two
non-intersecting phenomenal domains. The condition of structural coupling lies in the actual
domain of structural operation of the organisms in interactions with each other and with
their medium, where a continuous structural selection takes place in them through their
interactions with conservation of adaptation and class identity (see Maturana, 1980). A
consensus, or a consensual domain, lies in the domain of descriptions of the observer that
sees the organisms in coordinated conduct while they operate in their domain of reciprocal
ontogenic structural coupling.
When the observer considers the organisms in their actual operation as members of a
domain of reciprocal structural coupling, he sees them interacting under the conditions of
a pre-existing structural complementarity that appears to him comparable to a situation of
communication designed by an engineer, and to which he may apply a description in terms
of information transfer in the strict sense. When the observer considers the organisms as
 Comments 157

operating in a consensual domain, he sees them in a coordination of conducts that he can

describe in semantic terms, that is, as if the interacting organisms were using information in
the metaphorical sense (meaning) for their actual operational coordination. Neither of these
two uses of the notion of information, however, has explanatory value for the biological
phenomena involved. For this reason I think that the notion of information should be
totally avoided in the domain of biological explanations; in that domain the notion of
information (in whatever sense) is both unnecessary and misleading. In fact, the use of the
notion of information has interferred with the understanding of language and cognition by
obscuring the separation of the two phenomenal domains mentioned above. This is par-
ticularly apparent in relation to the mind-body problem where the notion of information
through its instructive connotations seemed to offer the possibility of bridging the gap
between the abstract (mind) and the concrete (body). Information, however, as a descriptive
notion cannot do the trick, and its two senses maintain the separation between the abstract
and the concrete.
If one gives up any attempt to use the notion of information, abandoning any hidden
notion of instructional determinism, one remains in a domain of exclusive structural
interactions and the situation changes completely. In fact, by doing this one accepts:
(i) that an interaction can take place only if there is between the interactants a structural
complementarity that allows for their reciprocal triggering of structural changes in their
encounters; (ii) that the circumstances or relations that lead to an actual interaction between
two systems select which of their possible structural changes take place by selecting which of
their possible interactions in fact occur; and (iii) that relations in a domain of descriptions as
relations in a domain of ontogenic structural coupling (consensual domain) may recursively
select the structural changes of the organisms that produced them through their operation in
language. The general result of this is that one accepts the legitimacy of generatively depen-
dent but not intersecting phenomenal domains like body and mind as well as the selective
consequences that the mind may have on the dynamics of structural changes of the body of
the observer through his operation in the domain of language where the mind exists. In other
words, metaphorically, by abandoning the notion of information, the mind becomes
comparable to the moir6 patterns that an observer sees when he beholds two differently
regular systems of dots on transparent papers that are superimposed and moved with respect
to each other. Without the transparent papers and dots upon them no moir6 pattern can be
observed. Yet, one cannot say that the papers and the dots cause the moir6 patterns because
these, as relations established by an observer in his observation, exist in a phenomenal
domain that does not intersect the phenomenal domain in which the dots and pages exist
as physical entities that reflect light. Furthermore, if the observer as a result of observing a
moir6 pattern moves one of the sheets of paper so that the pattern changes, it cannot be said
that the moir6 pattern as an abstract entity, that is as a system of relations in the domain
of descriptions of the observer, caused its own change or caused a change in the structure
that made it possible. We can say, however, that certain relations given in the interaction
between the pages with dots and the observer, and which this describes by saying that he
sees a moir6 pattern, constituted a configuration of interaction that triggered (and thus
selected) a structural change in the observer, who, in turn, through his interaction with the
pages, triggered a structural change in the system of pages that resulted in a change in the
moir6 pattern that he observed. This generative dynamics, in which non-interesecting
phenomenal domains arise as metadomains when two independent systems interact, as
well as the non-causal selective participation of relations, that is so central to biological
phenomena, are lost in a description in informational terms through its instructional
connotations.
158 Comments

Finally, a comment on evolutionary questions (see Maturana, 1980) that are at the core
of sociobiological arguments in the domain of human values:
The genetic constitution of a population is the result of the evolution of the lineages to
which the organisms that integrate the population belong, not the cause of th~ evolution of
these lineages. The genetic constitution of an individual organism is part of the organism as
a living entity until it dies, as a result of the organism's conservation of its adaptatio n (struc-
tural coupling to its medium) through its phenotypic realization as an organism. It is the
organism, the living being, the autopoietic unity, that which is adapted and lives or is not
adapted and disintegrates, not the genes; it is autopoiesis and adaptation that are conserved
in the course of biological evolution, not genes. In fact, the evolution of living systems is
the continuous drift of the lineages of organisms that conserve their structural coupling in
their media (adaptation) up to reproduction (see Maturana, 1980; Maturana & Varela, forth-
coming). Evolution, therefore, takes place only to the extent that the variations produced in
each generation result in phenotypes that conserve the adaptation of the organisms in their
medium. Only if a character, a function, a conduct, is there as a relation of the organism
with its medium, that is, only if a character is there as a phenotypic feature involved in the
conservation of the adaptation of the organism, can an evolution of the genetic/environment
determination of the character (function or conduct) take place through the conservation of
the character in a lineage. Only if a character is constitutively involved in the conservation of
the adaptation of the organism that exhibit it, does it participate in the conservation of the
lineages of these organisms. Adaptation is not the result of evolution (or natural drift, as I
prefer to call it); given reproduction, adaptation (the condition of structural coupling to the
medium) is the condition that makes evolution possible. In the absence of conservation of
adaptation the organisms die or do not reproduce, and their lineages come to an end. As a
matter of fact, without conservation of adaptation natural drift does not take place.
The description of hereditary events in terms proper to communication engineering
through the use of the notion of information in either sense, has contributed to the confusion
of the phenomenon of heredity (that results from reproduction) with a partial description
of its genetics (that refers to the genealogy of the distribution of characters), particularly
through the notion that the nucleic acids contain the genetic information of the organism.
This confusion has lead to the modern belief that the genes are the central elements in the
natural drift (evolution) of living systems, and that they may be adaptive, deletereous or
neutral, in a falacious argument that loses the organism as the evolutionary unity by reducing
it to a mosaic of characters subordinated to the conservation of its genes. An organism is
not a mosaic of characters, nor is it a conglomerate of genes. An organism is an entity
defined as a unity by its organization as a living system, and its characteristics (characters)
are necessarily partial descriptions of it made from the perspective of an observer who
unavoidably destroys its unity by making them.
Similar confusions have occurred in sociobiology, sociology and psychology, where,
frequently, the use of informational descriptions have reduced human beings to mosaics
of characters, ideas or values. Yet, it is not the engineer's notion of information our problem,
but rather it is our misuse of it in the domain of biological phenomena.

References
Ashby Ross, W. (1956). An Introduction to Cybernetics.
Maturana, H. R. (1978). In (G. A. Miller & E. Lenneberg. A.P., Eds.) Psychology and
Biology o f Language and Thought, New York: Academic Press.
Maturana, H. R. (1980). In (M. Zeleny, Ed.) Autopoiesis, Dissipative Structures and
Spontaneous Orders. A A A S Selected Symposium 55.
 Comments 159

Maturana, H. R. & Guiloff, G. D. (1980). J. sociol biol. Structures 3, 135-148.

Maturana, H. R. & Varela, F. J. (forthcoming). Evolution: NaturalDrift through the conser-
vation of adaptation (submitted to the J. social biol. Struct.).
Shannon, C. &Weaver, W. (1949). The Mathematical Theory of Communication. Urbana,
IL: University of Illinois Press.

Humberto R. Maturana
Departamento de Biologia
Facultad Ciencias Basicas y Farmaceuticas
Universidad de Chile
Casilla 653,
Santiago, Chile

Comment by Mario Bunge

Fedanzo is quite right in distinguishing two concepts of information and in stressing their
difference. One is the vulgar concept of meaning, the other is the information theoretic one.
I am glad he shares my view that neither has a place in biology (or in social science). In
particular, Shannon's concept of information is out of place in genetics and in social
science - at least as long as those who say that they use it do not teach us how they manage
to evaluate the quantity of information said to be locked in a gene or in a social system. [For
further reasons see Bunge (1979).]
In addition to the two senses of 'information' discussed by Fedanzo one may discern
another three, and it is possible that a closer examination of the current scientific and tech-
nological literature will yield further concepts designated by the same word. Let me list
those other concepts.
One is the concept of signal. A signal may be defined as a process - physical, chemical,
or biological - that, though carrying a small quantity of energy, is capable of triggering a
far more energetic process - as when a signal propagating down a nerve contracts a muscle,
or when we open the garage door with a radio wave. This concept is often, and unfortunately,
called 'information' in neuroscience and in psychology, although the word 'signal' is more
precise.
Another concept of information is that of knowledge, as when we say that we just
received information about the political misuses of sociobiology by the French right wing,
or when we transmit to receive an instruction or a request. From a psychobiological view-
point, knowledge is a state or process in some of the plastic subsystems of the vertebrate
nervous system. Hence it cannot be transmitted directly, i.e. without the intermediary of
some external signal. [For this view of mind see Bunge (1980).]
A third concept of information is that of communication, or transmission of a knowledge
item, be it datum or conjecture, question or command. Obviously, knowledge is conveyed
by artificial signals of various kinds. I submit that this is the concept o f information relevant
to sociobiology and the social sciences. Indeed, communication, together with other social
relations - such as those of grooming, feeding the young, courting, playing, cooperating, and
fighting - constitutes the structure of a social system, i.e. the collection of couplings or
bonds that keep it together. [Cf. Bunge (1979; 1981) for a detailed analysis of the concept
of social structure.]
We have then, altogether five different concepts o f information: information I = meaning;
information 2 = signal; information 3 = quantity o f information carried out by a signal in a
system; information 4 = knowledge; and information 5 = communication o f information 4
by social behavior involving a signal (information 2].