Reproduced from: https://cepa.info/584
Table of Contents
EVERYTHING SAID IS SAID BY AN OBSERVER
MATURANA H. R.
Cite as: Maturana H. R. (1987) Everything said is said by an observer. In: Thompson W. I. (ed.) Gaia: A way of knowing. Lindisfarne Press, New York: 65–82.
First of all, before I get to what I want to say about cognition, I need to point out that I am not after an explanatory principle. In part, I think that principles do not work, that whenever one has an explanatory principle, one invents a mechanism to conceal what one wants to explain. So what I propose to do is to specify a problem, and to specify also what I understand an explanation to be; then I shall discuss, from my perspective, a way of addressing the problem. In a way, I am asking you to accept as a problem what I shall propose as a problem, to accept as an explanation what I shall propose as an explanation, and, finally, to accept as an answer what I shall propose as an answer. But I am being explicit, and to make it clear that I am being explicit, I will write here in italics,
|Everything is said by an observer.|
And beside this I draw an eye.
Now, what is the problem? I want to think about cognition, so the problem is, then, to grasp what is the problem in cognition. I think that whenever we want to know whether or not somebody knows something about something, we ask him or her a question; and the question demands that he do something. If you want to know if somebody knows architecture, then you ask him how he would build a building, how he would proceed to make a building with certain characteristics. If he shows a way of doing this which is satisfactory to the questioner, then the questioner can say that he knows architecture. The same thing applies to biology, physics, Buddhism or religion of any kind, anything. So the problem is to identify adequate conduct. What constitutes adequate conduct, that is, a conduct which the questioner will accept as adequate? If I ask somebody if he knows biology, and he says, “Yes, I know biology; I am a specialist in such and such a thing,” and I next ask a question to which he responds by saying or doing something which I recognize as adequate conduct in that domain, then I can say that, Yes, he knows. And I think that this is what we always do. Actually, we have no other way of assessing knowledge. Therefore, I take “adequate conduct” as an expression of knowledge. Hence, if my problem is cognition itself, or knowledge, and I recognize knowledge by seeing adequate conduct, then my problem will be to identify adequate conduct, or to show how adequate conduct arises.
What, then, would be an explanation? Usually, whenever you ask a question – ask somebody else to explain something – you expect that person to produce an answer which is satisfactory. What does satisfactory mean? It means that you don’t continue questioning. When a child comes to his or her mother and asks, “Where do I come from?,” the mother provides an explanation. Now, throughout history the answers provided by mothers have changed. When I was a little boy, mothers used to tell about the bees and the flowers and things of that sort, and we children would go away to play, completely satisfied – until the next day. That was an explanation for a child, at least until the next day, when either the same question or a new one arose, because the explanation that had been given was no longer satisfactory. So the listener, then, the questioner, is the one who decides what an explanation will be: what will satisfy his curiosity? This means that if I am going to talk about cognition, I must provide an explanation which has to do with adequate knowledge, and I must be very clear about what I shall accept as an explanation.
Now, I am a biologist, a scientist, so I shall accept as an explanation only a scientific explanation. But what is a scientific explanation? Usually people think that scientific explanations have to do with predictability, that answers or propositions which allow us to predict are scientific explanations. But in my opinion this is not the case. Scientific explanations do not have to do with predictability; predictability may appear, but it is not the central point. The central point of a scientific explanation is the proposal of a mechanism. You have a question – for example, how does a horse move? A horse’s movement includes trotting, and you want this explained. The scientific explanation would be a description that would imply several things, but it would have to contain a description of the mechanism which generates the horse’s movements. If you want to explain lightning, you have to present a mechanism which generates lightning. This mechanism will be presented in terms of certain ideas that you have about clouds, friction, electrostatic charges, and things of this sort, but what you are actually proposing in one central idea is a mechanism that generates the phenomenon you want to explain. First, you observe the phenomenon that you want to explain, which is the question; second, you have to provide the mechanism. There is no scientific explanation if you do not propose a mechanism. But that alone is not sufficient. What is also needed in order to make the explanation a scientific explanation – and here is where the problem of predictability arises – is that the proposed mechanism generate not only the phenomenon that you want to explain, but other phenomena that you may observe as well.
Taking other observed phenomena into account is a requirement for a scientific explanation because scientists claim that what they say has something to do with the world in which we live, and that the phenomena they want to explain are the phenomena of the world. They claim that the propositions they make have a particular relation with the mechanisms that generate the phenomena because there is some isomorphism, some correspondence in structure, between the mechanisms proposed and the mechanisms in the world that generate the phenomena they want to explain. But since one can invent many possible mechanisms to generate a particular phenomenon, the scientist must select from among this multitude one in which he has more confidence because it seems to have to do with the world in which we live. This is why he also looks for some other phenomenon that will be generated by his explanatory mechanism and that pertains to the same domain as the phenomenon he wishes to explain.
So, as a scientist, I propose a mechanism. I say, “Aha! This mechanism generates this phenomenon.” Of course, I have proposed the mechanism specifically because it will generate the phenomenon which constitutes my question. But then I look at this mechanism and realize that it may also generate some other phenomenon, for example, phenomenon A, which is different from the one I am explaining. That is, another phenomenon may occur in the same domain in which the phenomenon I am explaining occurs. So I look around, and if I find this other phenomenon, then I can say, “Aha! My explanation has been validated, my hypothesis has been validated. This is a scientific explanation.” And that is it. It is not more, it is not less. How long will this explanation last? Until I find other phenomena which are not generated by it. At that time, I must realize that my explanation is no longer a scientific explanation. I have to drop it and invent a new mechanism which will generate not only the former phenomena but also other ones in the same domain which I consider to be important, but which are not being generated by the original mechanism. Therefore, if I want a scientific explanation of cognition, I must provide a mechanism which will generate adequate conduct – animal and human – as well as other phenomena which I can observe in the same domain. If I can do that, then, by all scientific standards, I have proposed a scientific explanation of the phenomenon of cognition – if you accept that the phenomenon of cognition is properly grasped by stating the problem in terms of adequate conduct.
What I should do next, then, is to show how adequate conduct arises in any system. This can be done, provided that we have language which is adequate to do so. First, I am going to make a couple of clarifications. An entity, anything that we can distinguish in some way, is a unity. How do we distinguish such a unity? There are many ways. For example, I could make a concrete distinction, in terms of picking it up, or a conceptual distinction, in terms of specifying a certain procedure which will carve out this unity from the background – which is at the same time specified by making the distinction. That is, when I say that something is a unity, I am also specifying all the rest of the background. This is what we do continually. If I were to ask you how many cushions there are in some room, you would count them. And in counting, you would be distinguishing cushions, performing the operation of distinction that carves these things out from the background. You may agree or disagree with someone else who is counting, but if you disagree it means that the two of you are applying different procedures of distinction. You are distinguishing different things. But if you agree, if you have the same procedure of distinction, you will count the same number of cushions, or chairs, or lamps, or whatever it is – persons, dogs, fleas, whatever. I had the honor, when I was a student at Harvard, of being the only student in a course on arthropods who knew personally fleas, mites, and all sorts of parasites. It was very interesting. I was the only one to be able to make those distinctions.
A second clarification is that we can and we do distinguish two kinds of unities – those which are simple, and those which are composite. Whenever we distinguish something as a whole and do not decompose it into parts, we distinguish it as a simple unity. Ideally the word “atom” means exactly that. If I distinguish my watch as a simple unity, then it is an atomic watch, if you wish. And the simple unity, in the moment in which you distinguish it, is specified for the operation of distinction in terms of certain properties. You can move it around, for example, or use it to point because it is long, or so on. The operation of distinction specifies or indicates the properties that characterize the simple unity. But we also distinguish composite unities. We say that the watch is made out of so many parts, things that can be separated. The components of the unity are those parts which are separate. Actually, the atom was an atom for many, many years, until the discovery of radiation allowed it to be decomposed, and then it was no longer an atom. We continue calling it an atom, but it turned out there were procedures to treat the atom as a composite unity, or a composite entity.
Now, when the unity distinguished is simple, the task is simple. One specifies properties, and that is sufficient. But when the unity distinguished is composite, there is a problem with the components, with their relations. There is a problem of composition – how are the parts put together? Here I make a distinction, which applies only to composite unities. I distinguish two features of composite unities, and I claim that we all do this. One has to do with the organization of a composite unity, which refers to the relations between the components that make the unity what you claim it is. For example, a chair is a composite unity. The relations between the parts that make it a chair are the organization. If I saw it into pieces and separate these pieces, would you say you still have a chair? No, you would not say that. You would say, “Why did you disorganize my chair?” I destroyed the chair by disorganizing it. The relations between components, then – that which makes a chair a chair – are its organization. A unity is a composite unity of some kind only as long as its organization is an invariant. A chair will be a chair only as long as it has the organization of a chair. If the organization changes, you no longer have, a chair. This is why, by the way, I do not think I should ever use the notion of self-organization, because that cannot be the case. Operationally it is impossible. That is, if, the organization changes, the thing changes. A chair is a chair, a composite unity of a particular kind, only as long as its organization is an invariant.
The second feature of composite unities has to do with structure. By structure I mean what most people mean by structure – the components and the relations that make a particular unity. A particular chair is made in a particular manner with particular components with particular relations between them. Another chair belongs to the same class, is a chair, is called a chair, because it has the same organization. But it has a different structure. The kinds of components that make up one chair are different from ‘the kinds of components that make up another chair. So the organization is invariant and is common to all the members of a particular class of composite unities, but the structure is always individual. Each particular unity has a structure which realizes the organization, and which is comprised of its particular components and the concrete, particular relations that make it a particular unity. But not only that. If I were to come with a knife and secretly carve little notches on your chair, you would not ask me why I disorganized your chair; rather, you would ask why I changed your chair. I would have modified the chair, but it would still be a chair.
So the structure of a composite unity can be changed without its organization being destroyed. If you destroy the organization, you no longer have the unity, but something else; however you can change the structure without changing the unity in terms of its class identity, in terms of the kind of unity you have. Now, this is very interesting because we all know it. If we come home and find that our children have carved the corners off the table, we say, “What have you done to the table?,” but it goes on being a table. Similarly, you continue giving the same names to your children all your life; there is something constant in the children even though they grow, and the name applies to this invariant, which is the organization, although the structure changes. Actually, in dynamic systems such as living systems, the structure is continually changing. You are changing your structure now. When I move, I change my structure, because the structure is both the components and their relations. Hopefully I can change my structure without losing my organization. As long as I can do that, or that happens to me, I am alive. But you can see that this is a very interesting situation, because when we look at things in this manner, which is what we do in everyday situations, we open an avenue to talk about change and invariance in living systems. Now, biologists know this, and when they speak about growth and evolution, they are speaking about conditions under which something remains invariant – the organization of the entity they are talking about – and something changes – the structure of those things they are talking about.
However, we still have a problem. If the explanation that a scientist proposes has to be a mechanism – and, as I have said, a scientific explanation implies a mechanism – then this explanation or hypothesis must satisfy the characteristics which make something a mechanism. That is, it must be the description or the construction of an entity whose structure – the relations and changes of relations of its actual components – determines what happens to it. To put it another way, since a scientific explanation entails the proposition of a mechanism which will generate the phenomenon, then this proposition of a mechanism means that whatever happens to the system, which itself is being proposed by the hypothesis of a mechanism that will generate the phenomenon, is determined by its structure. It is determined by the kinds of components and the relations between the components that constitute the system. This means that whenever you have a structure-determined system, or a mechanism, and you do something to it, what happens to it does not depend on what you do to it. What happens to it depends on it. If you have a refrigerator, for example, the changes that it undergoes in its dynamic aspects do not depend on what you do to it; they depend on how it is made. We know this very well from using any of those pushbutton machines, in which if you push a button something happens – it washes, it glows, it plays music – which is not determined by your pushing the button, but, rather, is triggered by the pushing of the button.
So in structure-determined systems, in mechanisms or systems that are defined and constituted structurally, what happens to the system depends on how it is made. The interactions that the system undergoes can only trigger changes in it. You do not instruct a system, you do not specify what has to happen in the system. If you start a tape recorder, you do not instruct it. You trigger it. And living systems, if they are to be explainable, must be treated as structure-determined systems, defined by certain organizations. Hence they must be systems in which whatever happens to them is determined in them by their structure. The interactions they undergo will only trigger changes in them; they will not specify what happens to them. This is a very serious point, one that should not be taken lightly. What I am saying is that to a structure-determined system, nothing can happen which is not determined by it – by how it is made, its structure. You are forced to accept this if you want me to provide a scientific explanation of living systems, because I cannot provide a scientific explanation of systems that are not structure-determined. That is, I cannot provide a scientific explanation for systems which do not admit mechanistic experimental hypotheses. So if you want me to provide a scientific explanation which has to do with something that living systems do, such as adequate conduct, then you are asking me to treat the organism or living system as a mechanism, as a structure-determined system.
For a system to change its dynamics of state, then, for it to change what it does, even though it maintains its identity and we still call it the same name, means that its structure must change. If I have a friend who turns from Catholicism to Buddhism, his behavior will be different, so there must have been a structural change. He could not change his behavior if his structure has not changed. But his structure is changing anyway, because he is a dynamic system, so in a way that is not the problem. The problem is, what structural change took place so that he changed from Catholic to Buddhist. Our problem is indeed to explain adequate conduct, to show how adequate conduct arises. This is a problem of showing how the structure of a living system changes in a manner such that we see a particular adequate conduct which we did not see before, or we go on seeing adequate conduct even through we know the structure is changing and the medium in which the system exists is changing also. The problem is to handle the problem of structural change and to show how an organism, which exists in a medium and which operates adequately to its need, can undergo a continuous structural change such that it goes on acting adequately in its medium, even though the medium is changing. Many names could be given to this; it could be called learning. But we also have the question of how the organism originally has an adequate conduct in the place where we find it. I shall answer this question first.
Why does an organism, a living system, a person have the conduct that it has where we find it or her or him? Why am I behaving the way I am behaving? This is a question that also has to do with evolution, in that in order to understand what is taking place in evolution, one must understand what is taking place in the individual through its life history, in the ontogeny. I shall answer this question of behavior in general terms. If I have a living system – and, although I will not go into it here, I depict a living system in this manner because it is a closed system, a system which only generates states in autopoiesis – then this living system is in a medium with which it interacts. Its dynamics of state result in interactions with the medium, and the dynamics of state within the medium result in interactions with the living system. What happens in interaction? Since this is a structure-determined system – and I cannot speak as a scientist if I do not treat systems in this manner – the medium triggers a change of state in the system, and the system triggers a change of state in the medium. What change of state? One of those which is permitted by the structure of the system. There are, of course, many changes of state that the structure of a particular system would permit, and the one that occurs depends on the particular circumstances. So in the interaction of a living system and its medium, although what happens to the system is determined by its structure, and what happens to the medium is determined by its structure, the coincidence of these two selects which changes of state will occur. The medium selects the structural change in the organism, and the organism, through its action, selects the structural change in the medium. Which structural change takes place in the organism? One that is determined by structure. Which structural change takes place in the medium? One that is determined by structure. But the sequence of these is determined by the sequence of interactions. The medium selects the path of structural transformation that a living organism undergoes during its life.
There are structural transformations, it is true, that result from a system’s own dynamic, but those which have to do with its medium are selected through the interaction with the medium. Two organisms ideally equal in the initial state, but in different media, will undergo different sequences of interactions. Hence they will have different personal histories, individual histories, different histories of structural change. When I was a medical student, other students always went to sleep in anatomy lectures. And so the professor used to say, “Please wake your friend. I think he will be an anatomy professor when he grows up; he is sleeping now.” I did not sleep in the lectures, so I never became an anatomy professor. So in the particular relation of two systems which have different structures and independence with respect to the interaction, each selects in the other the other’s respective path of structural change. If this history of interaction is maintained, the outcome is an unavoidable one. The structures of the two systems will have coherent histories, although in each of them the structural changes will be determined by the structure. So after a certain history of interaction, we as observers will observe a certain correspondence in the structures of the two systems, and this correspondence is no accident. It is the necessary result of this history, the ontogeny of the individual in this medium. None of us is here by accident. All of us are here as a result of our particular histories of interactions in our media. So this congruence that one observes is no accident. This by itself, in principle, explains the most apparent features of adequate conduct. Adequate conduct is conduct which is congruent with the circumstances in which it is realized. Conduct is something that one sees, the changes of state of an organism in a medium, as seen by an observer, by the eye, this fellow who sees and describes these changes of state of the organism in its medium as conduct.
What I am saying, then, is that the life history of every organism is a history of structural change in coherence with the history of structural changes of the medium in which it exists, as realized through the continual mutual selection of the respective structural changes. The congruence between an organism and its medium is, hence, always the result of its history. This is valid for each individual, for each organism. Each organism begins its existence as a cell, and as a cell it has certain initial structures. Now the initial structure of each organism at the beginning of its individual history is itself the result of another history, which is the history of the phylogeny – the sequence of reproduction leading to that cell which is the beginning of a particular organism. And in that history of the phylogeny, the following has taken place: In each reproductive step of each life previous to an individual organism, the then existing organism reproduces at least two other organisms of the same kind, and the one that can realize itself and reach the stage of reproduction participates in the lineage. The other one – let us suppose one does not reach that stage – does not participate in a lineage. Here the participation or non-participation in a lineage, the reaching or not reaching of the next stage of reproduction, depends, of course, on whether the ontogeny is realized or not. If the ontogeny is realized, that is, if this organism lives until reproduction, it is realized only because the organism maintains invariantly its correspondence with its medium. Its structure is changing, and the medium is changing, but the coherence with the medium is maintained invariantly. Adaptation is an invariant. If adaptation were not an invariant, it would stop, and the organism would disintegrate, die.
So every cell is itself the result of a long history, which implies millions of years, a history of successive, successful reproductions, and every cell pertains to one of the many lineages which possibly come from one common point in some faraway past. But along this history, the phenomenon of the organization of the cell, the condition of living, has remained invariant, and adaptation has remained invariant. Structures of the organism have been changing as a result of a continuous selection through the structural changes, through the interactions of the organism with the medium. So not only are we here now as a result of our personal histories, but we are here now as a result of the history of our ancestors. In a way, we all have the same age, and all our cells have the same age – millions of years – if we see not only our own individual ontogenies, but also the phylogenies, the history which is responsible for the structural changes that have led to our particular kind of coherence. That particular kind of coherence appears expressed in adequate conduct.
Now, I realize that you may think that there is a trick in this problem of adequate conduct, so perhaps I can illustrate the idea by relating an interesting anecdote which I read in Time magazine some years ago. A young student had to take an examination in physics. The professor handed him an altimeter and told him to determine the height of the campus tower with the altimeter. The student went to Woolworth’s, bought a piece of string, went up the tower, tied the altimeter to the string, lowered it to the base of the tower, and then measured the string: 32 meters, 50 centimeters. Flunk. But the student appealed, and the commission of education, or whatever, decided that he had the right to take the examination again. So the professor handed him the altimeter and told him to determine the height of the tower with it. This time, the student got a goniometer, which measures angles, went to some distance from the tower, and used the height of the altimeter to triangulate the tower. Flunk. Again petition, again concession, again determine height of tower with altimeter. Now, the tower happened to have a beautiful helicoidal staircase, so the student went along each step with the altimeter, determined the path of the screw, and again came up with a figure. This perverse student invented seven ways of determining the tower height without reading the altimeter! Of course, the question is, did he know physics or not? Did he have an adequate conduct? When the professor flunked him, it appeared that he did not have an adequate conduct. He failed to show adequate conduct under the circumstances in which the question was asked. So if the crucial thing was the opinion of the teacher, he failed. But the education commission had a different opinion, and so he did not fail.
Now, the teacher that determines this fundamental adequate conduct for us is life. If we remain alive, we have adequate conduct – whichever way we manage to remain alive. And if we reproduce, we participate in a lineage. However, if the criterion is determined by the professor, we have adequate conduct only to the extent that we satisfy the demands of the professor. Can what I have said in very general terms that obviously apply to our fundamental call, which is to live, also be used to explain adequate or inadequate conduct in front of the professor? Yes, and I shall show you how. Suppose that instead of considering only the medium which I have previously posited – an inert physical medium, something that we would not call living – I put in another living system. Then the situation will be this: We shall still have the previous interactions taking place, but other interactions will also take place. But my argument about interactions will still apply for these new interactions, because the phenomenon of selective interaction, of selecting the structural change in the other, does not depend in any way on the characteristics of the agent with which the change is made, provided interaction takes place. In fact, the organism specifies what it admits as an interaction. Each one of you specifies what you admit as an interaction – To other things you are transparent. You don’t understand what I say when I use an unknown language; you specify what languages you understand.
So there is no restriction on what other things one can interact with, but if it happens that the other entity is a living system, then we have an adaptation that involves another living system. And the invariance of adaptation involves another living system. When this occurs, I claim, although I shall not go into its full development, that we have a linguistic domain. Whenever we have organisms that through a history of interaction continue interacting with each other, we have a linguistic domain. But notice that adaptation, invariance of adaptation, is a structural coherence, meaning that the structure of the system can be described as having mutual correspondence in a dynamic manner. I have called this structural coupling. The same thing happens between organisms. If there is coherence in the history of interaction, they are mutually adapted. And they will continue interacting with each other as long as there is coherence, as long as they remain mutually adapted, because each interaction will result in the selection of a particular structural change. And again, whenever this takes place, a linguistic domain is established. If this linguistic domain allows for a recursion in linguistic interaction, then we have a language, but I shall not go into that. Certainly when a professor and a student have a history of interaction, the adequate conduct of the student will reveal a coherence in the domain of interaction with the professor. If such coherence is interrupted at some moment, then the student will not have adequate conduct in the eyes of the professor. But the professor and the student do select in each other the path of structural changes as long as they maintain the relation.
To the extent that I have shown you the mechanism by which adequate conduct is generated, I have answered the question I proposed about the phenomenon of cognition. Remember that I did not ask, what is cognition? I only asked, under what circumstances do we recognize cognition? I have shown the circumstances that generate phenomena in which we recognize cognition, but I have also done something else. I have made an identity between cognition and living, at least in the absolute general terms that have to do with us as living systems. There are other, much more restricted domains of cognition, and, with respect to those, I have said that in any given domain of concentrality that we establish with another organism, the other organism will observe in us cognitive behavior – will observe in us adequate conduct. The phenomenon of cognition, from what I have said, is necessarily relative to the domain in which one observes structural coherences which are the result of the histories of interactions of the organisms.
Finally, I will present a couple of interesting ideas, although I will not be able to develop them here. When you have language, what you have is the possibility of behavior that the observer can describe as recursions in a linguistic domain of consensus. These recursions can take place because there is a very interesting peculiarity in the nervous system. The nervous system is a closed system, a closed network of components which interact with each other, and in which the dynamics of state is a continuous change of relations of activity that generates relations of activity in the same network. Which relations of activity and which changes of relations of activity take place? Those that are determined by the structure of the nervous system. Therefore, one can show that in terms of description, because the description is conduct in this linguistic domain of mutual coherences, language is not in the brain or in the nervous system, but rather in the domain of mutual coherences between organisms. When the observer observes that this takes place, and that the distinctions realized here can be recursive, can be distinctions on distinctions in this domain, then we have a language. But that can only occur because everything is taking place in a closed system. For the system, for us in our nervous system, the act of picking up a piece of paper is a particular series of changes of relations of activity in our nervous system. To drink water is again another series of changes of relations of activity in our nervous system. To speak is yet another. From the point of view of what takes place inside the organism, everything takes place inside the organism in a closed manner. But for the observer, the coherences appear as language, or linguistic interaction, things of that sort. And this is what allows, finally, for your own dynamics of state in a linguistic domain to operate as a selector of your dynamics of state.