maturana and varela - autopoiesis

8/8/2019 Maturana and Varela - Autopoiesis

1/85

BOSTON STUDIES IN THE PHILOSOPHY OF SCIENCEEDITED BY ROBERT S. COHEN AND MARX W. WARTOFSKY

VOLUME 42

HUMBERTO R. MATURANA and FRANCISCO 1. VARELA

AUTOPOIESISANDCOGNITIONThe Realization of the Living

With a preface to 'Autopoiesis'by

Sir Stafford Beer

D. REIDEL PUBLISHING COMPANYDORDRECHT: HOLLAND I BOSTON: U .S .A .

LONDON: ENGLAND


2/85

Libra ry of Congress Catalog ing in Pub lica tion DataMat urana, Humber to R. 1928-

Autopoiesis and cognition.(Boston s tudies in the phi losophy of science ; v, 42)Bibliography: p.Includes indexes.1. Biology=Philosophy,

I. Varela, Francisco J., 1945-Q174.B67 vol. 42 [QH331]

2. Cognition. 3. Life (Biology)joint author. II. Title. Hl, Series.

SOls [574'.01] 79-24724ISBN 90-277-1015-5ISBN 90-277-1016- 3 pbk.

PUblished by D. Reidel Publishing Company,P .O . Box 17. 3300 AA Dordrecht , Hol land

Sold and d is tr ibuted in the U .S .A . and Canadaby Kluwer Boston Inc ., L inco ln Bui ld ing,

160 Old Derby Stree t, H ingham , MA 02043 , U .S .A .In a ll o ther countries , sold and d is tr ibutedby Kluwer Academic Publishers Group,

P .O . Box 322 , 3300 AH Dordrecht , Hol landD . Reide l Pub lish ing Company isa member of the K1uwer Group

Autopoiesis: The Organ izat ion o f the L iv ing was origina lly pub lished in Chi leunder the t it le De Maquinas y Seres Vivos, 1972 by

Editorial Universitaria S.A.All Rights Reserved

Copyright is) 1980 by D. Reide l Pub lish ing Company , Dordrecht , Hol landNo p ar t o f t he ma te ri al p ro te ct ed by t hi s c opyr igh t no ti ce may b e r ep ro du ce d o r

u ti lized in any form or by any means , e lect ronic or mechanica l,inc luding pho tocopying, recording or by any informational s to rage andret rieval sys tem, w ithout wri tten permiss ion from the copyr ight owner

Printed in The Netherlands

EDITORIAL PREFACE

Thi s i s a bold, br il li an t, p rovoca ti ve and puzz li ng wo rk . It demands a radicalshi ft i n s tandpo int , an almos t p aradox ica l pos ture in wh ich l iv ing sys tems a rede scr ib ed in t erms of what l ie s out si de the doma in of de sc ri pt ions . P rofesso rHumberto Maturana, with his col league Franc isco Varela, have undertaken theconstruct ion of a systematic theoret ical b io logy which a ttempts to define l iv ingsystems not as they a re obje ct s o f observat ion and descr ip ti on , nor even a s in-terac ting systems, but asse lf-conta ined uni ties whose only reference isto them-selves. Thus, the standpoint of descrip tion of such uni ties from the 'outside' ,i .e ., by an ob se rve r, al re ady seems to v io lat e t he fundament al r equi rementwhich Maturana and Va rel a pos it for t he charact er iza ti on of such sys tems -namely, tha t they are autonomous, self-re fe rr ing and self-construc ting c losedsys tems - in sho rt , autopoietic systems i n t hei r t erms . Yet , on the ba si s o f sucha conceptu al method , and such a theory o f l iv ing sys tems, Maturana goes on todef in e cogni ti on as a biolog ica l phenomenon; as , i n ef fec t, t he very nature o fal l l iv ing sys tems . And on thi s ba si s, t o generat e t he very doma ins o f i nt er ac -t ion among such systems which const itute language, descrip tion and thinking.

The rad ical shi ft i n s tandpo int h ere r equi re s an imaginat iv e l eap and theabandonment a t the outse t of the standard character izat ions of l iv ing systemsin te rms o f funct ion o r purpo se, o r o f organi sm-envi ronmen t r el at ion s, or o fc au sal i nt er ac ti on s wi th an exte rn al wo rld, or even in t erms o f informat ion,coding and transmission. In effec t, Maturana and Varela propose a theoret icalbiology which is topological, and a topology in which elements and theirr el at ions cons ti tu te a c losed sys tem, or more rad ica ll y s ti ll , o ne wh ich f romthe 'po int o f v iew' of t he sys tem i tsel f, i s e nti re ly sel f- ref er en ti al and has no'outside' , Leibniz ian for our day.

The work demands and de serves ca reful r ea ding . I t i s t ec hn ica l, formal ,d ifficult , phi losophical and boldly imagina tive . It is rigorously constructed,and insofar asi t isa theoret ical b io logy, i t remains uncompromisingly abstrac tand formal. Yet it smells of the medical laboratory and of the working do-ma in of t he neu rophys io logi st . Where the inte rp ret at ion o f t he forma l t heorymaps it i nt o the doma in of t he nervous sys tem , the ins igh ts a nd sugges ti on sfor fur th er i nt erpret at ion a re exci ti ng indeed. And we expe ct no thing le ss ,h ere and to come.

v


3/85

vi EDITORIAL PREFACEMaturana is perhaps best known to the philosopher and the scientific

nonspecialist as a co-author of the classic \959 paper 'What the Frog's Eye'Yells the Fro\! ,'s Brain' "with l ..et tvin, McCu\loch ann 1'i .t ts ). S i.nce then, hehas worked on the anatomy and neurophysiology of vision, especially oncolor v ision . He has al so been teaching medica l studen ts . The problems andpuzz les which emerged in hi s re search and teaching led Maturana to developa dis tinctively alternative theoretical f ramework in order to answer the ques-t ions, 'What i s a Living Sys tem? ' and 'What i s Cogni tion?' The consequenceof his investigations, and of his construction of living systems asself -making,self- referr ing autonomous unities , is that he discovered that the two questionshave a common answer. He writes, "Living systems are cognitive systems, andliving as a process isa process of cognition."

We are very p leased to int roduce thi s ma jor theoret ical work in the BostonStudies in the Philosophy of Science. The integration of biological theory,fo rmal cons truct ion , epi stemology (and, fu rther, Maturana' s suggest ions ofthe nature of interacting systems as a kind of biological sociology, and hissketch of the ethical implications of such a construction) - all mark thesetwo studies as among the most original attempts at a systematic biology indecades , and as a profoundly philosophical work.

Centerfor Philosophy andHistory of ScienceBoston UniversityJuly, 1979.ROBERT S. COHENMARX W. WARTOFSKY

GENERAL TABLE OF CONTENTS

vEDITOR IAL PREFACE

GENERAL TABLE OF CONTENTS viiixFOREWORD

INTRODUCTION (by Professor Maturana) xi

BIOLOGY OF COGNITION

DedicationTable of Contents

I. IntroductionI I. The ProblemIII . Cognitive Function in General

A. The Obser ve rB. The Living SystemC. Evolut ionD. The Cognit ive Proces s

IV. Cognitive Function in Par ticularA. Nerve Cell sB. ArchitectureC . Funct ionD. RepresentationE. Descr iptionF. ThinkingG. Na tura l LanguageH. Memory and Learn ingI. The Observer

V. Problems in the Neurophysiology of CognitionVI. Conclusions

VII . Post Scr iptum

2357889111315151921222 62930353 S414~5~


4/85

viii GENERAL TABLE OF CONTENTSAUTOPOIESIS: THE ORGANIZATION OF THE LIVING

Preface (by Sir Stafford Beer)Introduction

I. On Machines, l iv ing and OtherwiseI. Machines2. living Machines

I I. Dispensabil ity of TeleonomyI. Purposelessness2. Individuality

I II . Embodiments of Autopoies isI. Descriptive and Causal Notions2. Molecular Embodiments3. Origin

IV. Dive rsi ty 0f Autopoiesis1. Subordination to the Condition of Unity2. Plasticity of Ontogeny3. Reproduction, a Complication of the Unity4. Evolution, a Historical Network5 . Second and Third Orde r Autopo ie ti c Systems

V. Presence of Autopoies is1. Biological Implications2. Epistemological Implications3. Cognitive Implications

Appendix: The Nervous SystemGlossaryBIBLIOGRAPHY

INDEX OF NAMES

637377777885858788889093969698

10 010 210 711211 2115118124135139141

FOREWORD

Everything that I say in the introduction is my exclusive responsibility.Fr anci sco Vare la has been generous enough to grant me thi s freedom inwha tconcerns the essay that we wrote together. His views about it he expressesfully and independently in his book Principles of Biological Autonomy,published by Elsevier-North Holland, New York, 1979.

Also, I wish to acknowledge the subtle debt which the contents of thisbook owe to the many illuminating conversations about all the topics herecontained that I have had over more than four teen yea rs with my dea r fr iendProfessor Felix Schwartzmann, of the Univers ity of Chile .

ix


5/85

Representat ion of the cel lu la r autopoiet ic network .A ll a rr ow s t ha t d o not c ro ss t he b ound ar y o f t he r ep re se nt ed u ni ty i nd ic at e p ro du ct io nr elat ions. The unif or ml y shaded ar eas, i ncludi ng the boundar y l ine and t he wedges,together w ith the names , ind icate const itut ive relat ions . The general form of c losure w ithrespect to product ion and const itut ion in a recursive network reali zed as a concret euni ty t hr ou gh t he p re fe re nt ia l r el at io ns o f t he c omponent s w it hi n t he n etwo rk , i nd ic at eorder rel ati ons and t he cons equent cleavage of the network as a simple uni ty f rom it smedium. Th e whol e r ep re se nt s a c lo se d n etwo rk o f p ro du ct io ns , b ut t he a rr ows a cr os sthe depict ed const it uti ve boundary of t he network i ndicat e the necess ary mat erial

openness of the system asit real izes the physical space.

INTRODUCTION

In this introduction I wish to make some comments in relation to the twoessays that make this book, but i n order to do so Imust, at least i n p art, writeabout how they came into being .

HISTORYSince my childhood I have been interested in animals and plants, and I fre-quen tly asked myse lf what made them l iv ing. Thus, in 1948, in my fi rst yea ra s a medical student , I wrote a poem whose fi rst st anza was :

"i .Que es Ia muer te para el que Ia mira?i .Que es lamuer te para e lque la s ien te?Pesadez ignota, incomprensible,do lo r que e lego lsmo t rae , para ese ;s ileneio, paz y nada, para este.S in embargo el uno sienteque su orgul lo s e r ebela, que sumenteno soporta que t ra s Ia muer te nada quede,que tras Iamuer te este Iamuerte.EI o tro , en su paz , en su si lencio,en su majestad inconseiente siente,nada siente, nada sabe,porque Iamuer te es la muertey t ra s lamuerte es ta lav idaque sin Iamuer te solo esmuer te ."

What is death for the beholder? I What is death for the dying? I A weightbeyond knowledge or understanding, I A pain f or the se lf-a ss er ting ego, fo rthe one; I For the othe r, s il ence, peace, and noth ingnes s. /I Ve t the one feel shi s p ride in ange r I And in his mind he does not accept I That beyond deathnothing should arise, I And that beyond death I There should be only death./I The othe r, in h is si lence , I In his unknowing majesty feels , I He feels noth-ing , he knows noth ing, I Because death is death I And life without death isonly emptiness. II

xi


6/85

xii INTRODUCTIONThe poem is not a verygood one, yet it contains the implicit question: 'Whatkind of systems are livingsystems that they may die,and how come that theycognize?'In 1954 I went to s tudy biology in England and the U.S .A., and when Ireturned to Chi le in 1960, after s ix years of s tudy and research abroad, Ibegan to work at the MedicalSchool of the Universityof Chilein Santiago asa research associate in the Department of Biology.There I was involved intwo kinds of activities; I collaborated in teaching a course ingeneralbiologyfor the medical students, and I did research in the fieldsof neurophysiologyand neuroanatomy. In my teaching my responsibil ity was to convey to themedical students some understanding of the organization of livingsystems asautonomous entities, as well as some understanding of their possible originon ear th. In my research I wanted to apply to the s tudy of form and colorvision in birds the same approach that J. Y. Lettvin and I had used in thestudy of form visionin the frog.I soon discovered through my teaching that the central question that thestudents would always ask was: 'What isproper to livingsystems that had itsorigin when they originated, and has remained invariant since then in thesuccession of thei r generat ions? ' At the same t ime I soon realized in myresearch that my central purpose in the study of color visioncould not bethe study of a mapping of a colorful world on the nervous system, but ratherthat i t had to be the understanding of the par ticipat ion of the ret ina (ornervous system) in the generation of the color space of the observer.

As a result of these different activit ies I entered a situation in which myacademic life was divided, and I oriented myself in search of the answers totwo questions that seemed to lead in opposite directions, namely: 'What isthe organization of the living?' and 'What takes place in the phenomenon ofperception?'Let me speak about how I faced them.First Question: What is the organization of the living?Whenthis question was first asked by the students, although it was the samequestion that had been lurking in my mindfor many years, I could not answeri t. I had prepared mysel f for this moment, but when j t came and I t ried toanswer it in a manner sati sfactory for the s tudents I realized that I had tothink everything anew. I could speak about form and function and astonishmy students and myself at the harmony ofnature, exalt ing the fitness of theenvironment and the fitness of the individual. I could claimthat the questionwas a very difficult one and that it could not yet be answered due to our

INTRODUCTION xiiiinsufficient knowledge. We had to accept that we could recognize livingsystems when weencountered them, but that wecould not yet say what theywere. I could enumerate features of living systems such as reproduction,heredity, growth, irritability, and so on; but, how long a listwas necessary":when would the listbe completed? In order to knowwhen the listwascom-pleted I had to know what a livingsystem was, whichwas,infact,the questionthat I wanted to answer in the f irst placeby producing such a l is t. I couldspeak about adaptation and evolution, about development anddifferentiation,and showhow all these phenomena were tied together by the phenomenon ofnatural selection; but the question: 'What was the invariant feature of livingsystems around which natural selection operated?', remained unanswered.Every approach that I could attempt and that I did attempt lef t meat thestarting point.Yet I obviouslyhad some inkling of what wasthe correct answere,becauseI rejected the unsatisfactory ones. After several years of these various at-tempts I realized that the difficulty was both epistemological and linguistic,and that both my wife and myoid professor, J. Z. Young,were right: one canonly say with a given languagewhat the languagepermits. I had to stoplook-ing at livingsystems as open systems defined in an environment, and I neededa language that would permit me to describe an autonomous system in amanner that retained autonomy asa feature of the systemor entity specifiedby the description. In other words,any attempt to characterize livingsystemswith notions of purpose or function wasdoomed to failbecause these notionsare intrinsically referential and cannot be operationally used to characterize.any system asan autonomous entity. Therefore, notions of purpose, goal,useor function, had tobe rejected, but initially I did not know how. Accordingly,I tried in my lectures several approaches in order to find a way of speakingabout livingsystems in a manner that would grasp their autonomy as a phe-nomenon of their operation as unitary systems. Thus, eventually, I made adistinction between what I called self-referred and allo-referred systems, adistinction that separated systems that could only be characterized withreference to themselves, such as livingsystems, from systems that could onlybe characterized with reference to a context. I did thisin order to emphasizethat whatever took place in livingsystems as living systems, took place asnecessarily and constitutively determined in relation to themselves becausetheir being defmed as unities through self-reference was their manner ofautonomy; and that whatever took place in other systems took place asconstitutively determined in relation to the context with respect to whichthey were defined as unities. This way of speakingwasnot fully satisfactory


7/85

xiv INTRODUCTIONbut led me to realize that what wasindeed needed wasthe characterization ofa kind of system which, if allowed to operate, would operate in a mannerindistinguishable from the operation of livingsystems, and that one shoulddo so using only neighborhood relations realized through the properties ofthe components of the system. Itwaswith such aim that I spoke for the firsttime in 1969 of livingsystems assystemsdefined asunities through the basiccircularity of their production of their components.Second question: What takes place in the phenomenon of perception?When Jerry Y. Lettvin and I wrote our severalarticles on frog vision [Lettvin,J. Y., H. R. Maturana, W. S. McCulloch and W.H. Pitts 1959; Maturana,H. R., J. Y. Lettvin, W.S. McCulloch and W.H. Pitts 1960], we did it withthe implicit assumption that we were handling a clearly defined cognitivesituation: there wasan objective (absolute) reality, external to the animal, andindependent of it (not determined by it), which it could perceive (cognize),and the animal could use the information obtained in its perception to corn-pute a behavior adequate to the perceived situation. Thisassumption of oursappeared clearly in our language. We described the various kinds of retinalganglion cells as feature detectors, and we spoke about the detection of preyand enemy. Weknew that was not the whole neurophysiological story, aswas apparent particularly in the discussion of the article called 'Anatomyand Physiology of Vision in the Frog (Rana pipiens)'. But even there theepistemology that guided our thinking and writing was that of an objectivereali ty independent of the observer. Thus, when Samy Frenk and I beganto work with pigeons in 1961, first studying form vision, we approachedthat study with the same fundamental v iew. No problem arose then andwithout any difficulty we could characterize many classesof retinal ganglioncells. Yet, when Gabriela Uribe joined usand wein fact began to study colorvision in 1964, i t soon became apparent to us that that approach leads todeep trouble. Neurophysiologically we did not see anything fundamentallydifferent from what other scholars had already seen. We found the classictypes of ganglion cells with separate, concentric or overlapping opponentspectral preferences. But we also found: (a) that although the geometryof the receptive fields of the ganglion cells with opponent spectral prefer-ences had nothing to do with the geometry of the visualobject, the geome-t ry of the visualobject had to do with the response of those cell s; and (b)that we could not account for the manifold chromatic experiences of theobserver by mapping the visible colorfu l world upon the act ivi ty of thenervous system, because the nervous system seemedto use geometric relations

INTRODUCTION xvto specify color distinctions. A different approach and a different epistem-ologywere necessary.There aremany visualconfigurations, with uniform andvariegated spectralcompositions, in simple and complex geometrical forms, that give rise toindistinguishable color experiences. How should one, then, look for theinvariances in the activity of the nervous system, if any, in relation to theperception of color? After we realizedthat the mapping of the external worldwas an inadequate approach, we found that the very formulation of the ques-tion gaveus the clue. What if, instead of attempting to correlate the activityin the retina with the physical stimuli external to the organism,we did other-wise, and tried to correlate the activity in the retina with the color experienceof the subject?Such a step entai led two difficulti es . On the one hand i t required thedefinit ion of a reference which would permit the characterization of theactivity of the retinal ganglion cells independently of the stimulus as such;on the other hand i t requi red us to close the nervous system and treat thereport of the color experience as if i t represented the state of the nervoussystem as a whole. In other words, the new approach required us to treatseriously the activity of the nervous system as determined by the nervoussystem itself, and not by the external world; thus the external world wouldonly have a triggering role in the releaseof the internally-determined activityof the nervous system. We did this r igorous ly, and showed that such anapproach did indeed permi t us to generate the whole color space of theobserver. That was a very fundamental result that we published in a veryunknown article [Maturana, H. R., G.Uribe,and S.Frenk, 1968].But what was sti ll more fundamental was the discovery that one had toclose off the nervous system to account for its operation, and that perceptionshould not be viewed as a graspingof an external reali ty, but rather asthespecification of one, because no distinction was possiblebetween perceptionand hallucination in the operation of the nervous system asa closednetwork.Although we arrived at this conclusion through the study of color vision,there are many earlier experimental studies (such as those of Stone on therotation of the eye of the salamander in the early 'forties) that could alsohave led to an understanding of the nervous system as a closednetwork ofinteract ing neurons. Whether they did or not, I do not know; but i f theydid it seems that the implications were not pursued to their ultimate conse-quences.

Whatever the case, for me this finding had great significanceand plungedme into the s tudy of cognit ion as a legi timate biological problem. Two


8/85

xvi INTRODUCTIONinmediate consequences arose from this: the first one was that inmy neuro-physiological studies I had to take seriously the indistinguishability in theoperation of the nervous system between perception and hallucination; thesecond one was that I needed a new languageto talk about the phenomena ofperception and cognition. The first consequence required that the question:'How does the organism obtain information about its environment?' bechanged to: 'How does it happen that the organism has the structure thatpermits it to operate adequately in the medium inwhich it exists?' A seman-tic question had to be changed into a structural question. The second ques-tion required the actual attempt to describethe phenomena that take placein the organism during the occurrence of the phenomena of perception andcognition in a languagethat retained them as phenomena proper to a closednervous system.A Congress in AnthropologyEarly in May of 1968 the University of Chile entered a state of revolution.The s tudents took over the Universi ty in an attempt to reformulate thephilosophy that had inspired its organization. I joined them. All standardacademic activities stopped and students and some members of the facultytried to say something new. It was not easy. Languagewas a trap, but thewhole experience was a wonderful school in which one could discover howmute, deaf and blind one was. Itwaseasyto be caught in one's own ego,butif one succeeded in attaining at least some degree of freedom from it, onebegan to listen and one's language began to change; and then, but only then,new things could be said.This lasted for severalmonths.In September of that year, I accepted aninvitation to go to the Universityof Ill inois at Urbana to the BiologicalComputer Laboratory of ProfessorHeinz von Foerster. Furthermore, Professor von Foerster invited me toparticipate in a symposium sponsored by the Wenner-Gren Foundation forAnthropological Research to be held duringMarch 1969 in Chicago,with thepurpose of considering the anthropology of cognition. The invitation wasto speak on 'The state of the art of the neurophysiology of cognition' . Iaccepted this invitation and decided not to speak about neuronal circuits,nerve impulses or synapses, but rather I decided to consider what should takeplace in the organismduring cognition by consideringcognition as a biologicalphenomenon. In doing this I found that my two apparently contradictoryacademic activit ies were not contradictory, and that they were, in fact, ad-dressed to the same phenomenon: cognition and the operation of the living

INTRODUCTION xvii

system - its nervous system included when present - were the samething.From this understanding the essay 'Biologyof Cognition' arose as anexpan-sion of my presentation in that symposium.

The Word 'Autopoiesis'The second essay included in this book waswritten in 1972, asan expansionof the section on 'Living Systems' in the 'Biology of Cognition'. The writingof this essaywas in fact triggeredby a conversation that FranciscoVarelaandI had in which he said: "If indeed the circular organizaton issufficient tocharacterize livingsystems as unities, then one should be able to put it inmore formal terms". I agreed, but said that a formalization could only comeafter a complete linguistic description, andwe immediately began to work onthe complete description. Yet we were unhappy with the expression 'circularorganization', and we wanted a word that would by itself conveythe centralfeature of the organization of the living,which is autonomy. Itwas in thesecircumstances that one day, while talking with a friend (Jose Bulnes) aboutan essay of his on Don Quixote de la Mancha, in which he analyzed DonQuixote's dilemma of whether to followthe path of arms (praxis, action) orthe path of letters (poiests, creation, production), and his eventual choiceof the path of praxis deferring any attempt at poiesis, I understood for thef irst t ime the power of the word 'poiesi s' and invented the word that weneeded: autopoiesis. This was a word without a history, a word that coulddirectly mean what takes place in the dynamicsof the autonomy proper toliving systems. Curiously, but not surprisingly, the invention of this wordproved of great value. It simplified enormously the task of talking about theorganization of the livingwithout falling into the alwaysgapingtrap of notsaying anything new because the languagedoes not permit it. Wecould notescape being immersed in a tradition, but with an adequate languagewe couldorient ourselves differently and, perhaps, from the new perspective generate anew tradition.

Let me now saysomething about the essaysthemselves.

B IOLOGY OF COGNITIONWhen I wrote this essay I did not yet havethe word 'autopoiesis' , nor had Icome to the more formal expression of the livingorganization givenin thenext essay. Yet, these shortcomings do not detract from what issaidbecausethe basic relations embodied in the notion of autopoiesis are fully implied,


9/85

xviii INTRODUCTIONalthough cumbersomely said,by expressionssuchas 'circular organization' and'self-referential systems'_ Furthermore, when I wrote the essay I decided notto make any concession to existing notions that I considered inadequate ormisleading, even if this seemed to make the text particularly obscure, How-ever, I made a concession which I havealways regretted, I submitted to thepressure of my friends and talked about causalrelations when speaking aboutthe circular organization of living systems, To do this was both inadequateand misleading. Itwas inadequate because the notion of causality isa notionthat pertains to the domain of descriptions, and as suchit isrelevantonly inthe metadomain in which the observer makes his commentaries and cannotbe deemed to be operative in the phenomenal domain , the object of thedescription. It was misleadingbecause it obscured the actual appreciation ofthe sufficiency of the notion of property asdefined by the distinctive opera-tion performed by the observer when specifyinga unity, for the description ofthe phenomenal domains generated by the specified unities. Itwas misleadingbecause it obscured the understanding of the dependency of the identity ofthe unity on the distinctive operation that specified it. It was misleadingbecause it obscured both the understanding of the phenomenal domains asdetermined by the propert ies of the uni ties that generate them, and thenon-intersection of the phenomenal domains generated by the operation ofa composite unity as a simple unity in a medium and bythe operation of itscomponents as components.There isnothing elsethat I wish to add as a commentary to this essay. Itis a cosmology and as such it iscomplete. Finally I wishto say that I find itpervading my views and understanding of everything. In a sense it has beenmy way to transcendental experience: to the discovery that matter, meta-phorically speaking, is the creation of the spirit (the mode of existence of theobserver in a domain of discourse), and that the spiri t is the creation of thematter i t creates, Thisis not a paradox, but it is the expression of our exist-ence in a domain of cognition in which the content of cognition iscognitionitself. Beyondthat nothing can be said.

AUTOPOIESISThis art ic le was wri tt en as an expans ion of two pages of the 'Biology ofCognition', in an attempt to show that autopoiesis isnecessaryand sufficientto characterize the organization of livingsystems, and that given the properhistorical contingencies one can deriveall the biological phenomenology fromthe characterization of livingsystems as autopoietic systems in the physical

INTRODUCTION xixspace. Notions of purpose, function or goal are unnecessary and misleading.This the essay does, and in this respect no commentary is needed; the essaystands by itself. Yet, when we wrote it we werejust beginning to realizethefundamental distinction between organization and structure, and we do notseparate the terms with complete rigor. Also, because it was not writtenunder the supposition that the reader had readthe 'Biologyof Cognition', theessay is not fully clear concerning the validity of the statement "Everythingsaid is said by an observer" in relation to the distinction between characteri-zation and description of a system. I shallmakesomecomments about thesethemes.Unity, Organization and StructureUnity. The basic cognitive operation that we perform as observers is theoperation of distinction. By means of this operation we specifya unity asanentity distinct from a background, characterize both unity and backgroundwith the properties with which this operation endowsthem, and specifytheirseparability. A unity thus specified is a simpleunity that defines through itsproperties the space in which it exists and the phenomenal domain which itmay generate in its interactions with other unities.

If we recursively apply the operation of distinction to a unity, sothat wedistinguish components in it, we respecify it as a composite unity that existsin the space that i ts components define because it isthrough the specifiedproperties of its components that we observers distinguish it. Yet we canalways treat a composite unity as a simpleunity that doesnot exist in thespace of its components, but which exists in a space that i t defines throughthe properties that characterize it as a simpleunity. Accordingly,if an auto-poietic system istreated as a composite unity, i t exists in the space definedby its components, but if i t is treated as a simpleunity the distinctions thatspecify it as a simple unity characterize its properties as a simple unity, anddefine the spacein which it exists assuch a simpleunity.Organization and Structure. The relations between components that definea composite unity (system) as a composite unity of a particular kind, con-stitute its organization. In this definition of organization the components areviewed only in relation to their participation in the constitution ofthe unity(whole) that they integrate. This iswhynothing issaid in it about the proper-t ies that the components of a par ticular uni ty may have other than thoserequired by the realization of the organization of the unity.


10/85

xx INTRODUCTIONThe actual components (all their properties included) and the actual rela-

tions holding between them that concretely realize a system as a particularmember of the class (kind) of composite unities to which it belongs by itsorganization, constitute its structure. Therefore, the organization of a systemas the set of relations between its components that defineit asa systemof aparticular class,is a subset of the relationsincluded in its structure. It followsthat any given organization may be realized through many different struc-tures, and that different subsets of relations included in the structure of agiven entity, may be abstracted by an observer(or its operational equivalent)as organizations that define different classesof composite unities.The organization of a system, then, specifiesthe classidentity of a system,and must remain invariant for the class ident ity of the system to remaininvariant: if the organization of a system changes, then its identity changesand it becomes a unity of another kind. Yet, since a particular organizationcan be realized by systems with otherwise different structures, the identity ofa system may stay invariant while its structure changes within limits deter-mined by it s organization . If these limi ts are overstepped, that i s, i f thestructure of the system changes so that its organization cannot any more berealized, the system loses its identity and the entity becomes something else,a unity defined by another organization.

It is apparent that only a composite unity has structure and organization,a simple unity does not. A simple unity only has the properties with whichit is endowed by the operations of distinction through which it becomesseparated from a background. It is also apparent that assoon asa compositeunity is treated as a simpleunity, any question about the originof its proper-t ies becomes inadequate because the properties of a simpleunity are giventhrough its distinction as a simple unity. It isalsoapparent that although theproperties of a composite unity as a simple unity arise from its organization,they are realizedthrough the properties of itscomponents. Accordingly,whiletwo simple unities interact through the simple interplay of their properties,two composite unities'interact in a manner determined by their organizationand structure through the interplay of the properties of their components.Structural CouplingIn the history of interactions of a composite unity in its medium, both unityand medium operate in each interaction as independent systems that, bytriggering in each other a structural change, select in each other a structuralchange. If the organization of a composite unity remains invariant while it

INTRODUCTION xxi

undergoes structural changes triggeredand selectedthrough its recurrent inter-actions in its medium, that is, its adaptation is conserved, then the outcomeof this history of interactions isthe selection,by the recurrent or changingstructural configuration of the medium, of a sequence of structural changesinthe composite unity, which results inthat the changingstructure of the organ-ism follows the changing structure ofthe medium through a continued struc-tural coupling to it. If organization or adaptation are not conserved, then theoutcome for the composite unity is disintegration. In other words, if a com-posite unity is structurally plastic its conservation of adaptation results in itsmaintained structural coupling to themediumthat selects itspath of structuralchange. In this process, the configuration of constitutive relationsthat remaininvariant in the adapted composite unity, determines the matrix of possibleperturbations that the composite unity admits at any instant, and, hence,operates as a reference for the selection of the path of structural changes thattake placein it in itshistory of interactions. Defined in this manner, structuralcoupling (conservation of adaptation) is not peculiar to livingsystems. It isaphenomenon that takes place whenever a plastic composite unity undergoesrecurrent interactions with structural changebut without loss of organization,which may follow any changing or recurrent structural configuration of itsdomain of interactions (medium). Therefore, all that is unique with respectto adaptation in living systems is that in them the autopoietic organizationconstitutes the invariant configuration of relations around which the selectionof their structural changes takes place during their history of interactions.

EpistemologyAs soon asa unity is specified, a phenomenal domain is defined. Accordingly,if a composite unity operates as a simple unity, it operates in a phenomenaldomain that i t defines as a simple unity that is necessarily different fromthe phenomenal domain in which its components operate. Therefore, theemergence of a phenomenal domain as the result of the operational distinc-tion of a composite unity asa simpleunity, makes phenomenal reductionism(and, hence, explanatory reductionism) impossible. Furthermore, the dy-namics of the establishment of unities through operational distinctions thatspecify their properties, have the result that all phenomenal domains arenecessarily realized through the operation (interplay) of the properties ofthe unities that generate them, that is, through relations of contiguity. If acomponent 'A' through its interaction with a component 'B' triggers aninteraction of 'B' with 'C' that triggersa reduction in the production of 'D',


11/85

xxii INTRODUCTIONan observer may say by considering the whole that 'A' controls the produc-t ion of 'D'. 'A', 'B', 'C' and 'D', interact through relations of contiguity inthe phenomenal domain that the component s def ine. Relations such asregulation, control or function, therefore, are not relations of contiguity;they are referential relations specified by the observer who puts himself in ametadornain of descriptions by using his view of the whole asa reference forhis description of the participation of the components that he describes inthe constitution of the composite unity."Everyth ing said is said by an observer." Th is I say in the 'Biology ofCognition'. The fundamental cognitive operation that an observer performsis the operation of distinction. By means ofthis operation the observerspec-ifies a unity as an entity distinct from a background and a background asthedomain in which an entity is distinguished. An operation of distinction, how-ever, is also a prescription of a procedure which, if carried out, seversa unityfrom a background, regardlessof the procedure of distinction and regardlessof whether the procedure is carried out by an observeror by another entity.Furthermore, the prescriptiveness of an operation of distinction implies auniversal phenomenalism of distinctions which, through the specification ofnew procedures of distinction or through their recursive application in thereordering of the distinguished entities, can, in principle, endlessly give riseto new simple and composi te uni ties , and, hence, to new non-intersect ingphenomenal domains. Thus, although a distinction performed by an observeris a cognitive distinction and, strictly, the unity thus specified exists in hiscognitive domain as a description, the observer in his discourse specifies ametadomain of descriptions from the perspective of which he establishes areference that allows him to speak as if a unity, simpleor composite, existedas a separate ent ity that he can characterize by denoting or connoting theoperations that must be performed to distinguish it.

In the perspective of a descriptive metadomain the distinction between thecharacterization of a unity and the observer's knowledge of it that permitshim to describe i t in a context , should be clear. In fact, knowledge alwaysimplies a concrete or conceptual action in somedomain, and the recognitionof knowledgealways implies an observer that beholds the action from a meta-domain. Therefore, when an observer claimsknowledge of a system, he claimsthat he can define a metadomain from the perspective of which he can simul-taneously behold the system as a simpleunity, describingits interactions andrelations as a simple unity, and its components as components, describingtheir interactions and relations as components. In these circumstances it islegitimate to distinguish between the characterization that an observer makes

INTRODUCTION xxiiiof a uni ty, either by point ing to i ts propert ies if i t is a simpleunity, or bypoint ing to i ts organizat ion if i t is a composi te one, from the knowledgeabout a uni ty that he reveals ,either by describingi ts operation asa simpleuni ty if i t is a simpleunity, or by describingboth i ts operation asa simpleunity and the operation of its components ifit isa composite entity. In eithercase, however, the knowledge that an observer claims of the unities that hedistinguishes consists in his handling of them in a metadomain of descrip-tions with respect to the domain inwhichhe characterizes them. Or, inotherwords, an observer characterizes a unity by stating the conditions inwhichitexists as a distinguishable entity, but he cognizesit only to the extent thathe def ines a metadomain in which he can operate with the enti ty tha t hecharacterized.

Thus, autopoiesis in the physicalspace characterizes livingsystemsbecauseit determines the distinctions that we can perform in our interactions whenwe specify them, but we know them only aslongas wecan both operate withtheir internal dynamics of states as composite unities and interact with themas simple unities in the environment in which we behold them. The factthatthe characterization of an entity isalsoa description made by the observer,and as such alsobelongs to his descriptive domain ('Biology of Cognition'),does not invalidate the operational effectiveness of the distinctions upondistinctions that constitutes the metadomain of descriptions in which thecognitive statements are made. The entity characterized isa cognitiveentity,but once it is characterized the characterization is also subject to cognitivedistinctions valid in the metadomain in which they are made by treating thecharacterization as an independent entity subject to contextual descriptions.Therefore, the complementarities system/environment, autonomy/control,totality/composition, etc., are complementarities only in our cognition of asystem that we observe in a context that allowsusto establish suchrelations,but they are not constitutive features of the referred system because they donot participate in its constitution through the interplay of the properties ofits components. Accordingly, that one should not be able to account for ordeduce all actual biological phenomena from the notion of autopoiesiswith-outresorting to historical contingencies, isnot a shortcoming of such a notion.On the contrary, it is to be expected because the notion of autopoiesis onlycharacterizes livingsystems as autonomous entities that can be distinguishedas composite unities realized through neighborhood relations.Finally, I would like to add some sociologicaland ethical comments thatI consider follow from the understanding of the autopoietic organizationof livingsystems. The essay on autopoiesis was supposed to have a second


12/85

xxiv INTRODUCTIONappendix on social and ethical implications. This appendix, howev~r , wasnever included because Francisco Varela and I never agreed about tts con-t en ts . Now I sha ll u se the pr ivi le ge tha t I have by wr it ing thi s in troduct ion ino rder t o presen t t he no ti ons tha t I would have inc luded in that a ppendix.

SOCIETY AND ETHI CSThe c en tr al f eature o f human exi st enc e i s i ts o ccu rr enc e in a l ingui st ic cogni -t ive domain. This domain is const itut ively soc ia l. Yet , what isa soc ia l system?how is a social s ys tem characterized? how do living systems in general, andhuman beings in particula r, participa te in the const itut ion of the soc ia l systemthat t hey int eg rat e? The answe rs t o these ques ti ons a re c en tr al for t he under -s tanding o f soci al dynamics and the p roce ss of soci al c hange . The fol low ingconsidera tions sta te the essence of my answers to these quest ions:

( I) I t i s a pparen t t ha t n atural soc ial sys tems a ssys tems cons ti tu ted by l ivingsys tems requi re t hese for t hei r a ctual r ea li za ti on. What i snot apparen t, how-ever , i s t he ex te nt t o which the coup ling o f l iv ing sys tems in the int eg ra ti onof a soc ia l system entai ls the rea liza tion of the ir autopoiesis. If the rea liza tionof a s ocial system wer e to entail the autopoiesis of its components, then therea li za ti on of t he au topoi es is o f t he component s o f a soci al sys tem wou ld beint ri ns ica ll y indi spensabl e and con st it ut iv e o f i ts operat ion a ssuch, a nd not amere contingency. Accordingly , any particula r fea ture , cul tura l or otherwise ,of the manner in which the autopoiesis of the components is realized andthei r i nd ividual it y and autonomy i s r es tr ict ed, would be a fe ature of t he pecul -i ar soci al sys tem (soci ety) i n ques ti on , a nd not i nt ri ns ic t o it asa soc ial sys tem .I f, however , t he autopoi es is o f t he component s o f a na tural soc ial sys tem werenot involved in its constitution becaus e the relations that defme a system assoc ial do no t e nt ai l t hem , then the autopoi es is o f t he componen ts ( and hencetheir autonomy and individuality) would be intrinsically dispensable.

(2) The ques ti on , 'What i s a soci al sys tem?' ca nnot be answe red by s imp lydescr ib ing a par ti cu lar one because we do not know the s igni fi can t r el at ionsthat we must abs tract when characterizing its or ganization. The questionmus t be an swered by p ropo sing a sys tem which , i f al lowed to operat e, wouldgenerate a phenomenal domain indis tinguishable from the phenomenaldomain proper to a natural social system. Accordingly, I propose that aco ll ect ion of autopoi et ic sys tems tha t, t hrough the rea li za ti on o f t he ir a uto-poiesis, interact with each other constituting and integrating a system thatoperates as the (or as a) medium in which they realize their autopoiesis , isi nd is ti ngu ishabl e f rom a natural soci al sys tem. O r, i n othe r words , I p ropo se

INTRODUCTION xxtha t the relat ions sta ted above character ize the organizat ion of a soc ia l syste ras a system, and that all the phenomena proper to s ocial systems aris e frort hi s o rgan iz at ion. I f one ac cept s t hi s propos it ion, a nd I shal l h encefor th p rec eed as i f i t we re a cc ep te d, t hen one has to a cc ep t t he fol low ing imp li cat ions

(i) The real iz at ion o f t he autopoi es is of t he component s o f a soc ial sy:t em i scon st it uti ve to the rea li za ti on of t he soc ial sys tem i tself . Thi s c anno t bignored in any cons id erat ion about t he ope ra ti on o f a soc ia l sys tem withounegating it.

(ii) A collect ion of l iv ing systems integra ting a composi te uni ty througlr el at ions that do no t i nvo lv e thei r a utopoi es is i s not a soci al sys tem , and thphenomena proper to its operation as such a compos ite unity are not sociaphenomena . Since the opera tional coupl ing of an organism integra ting a compos it e un it y does no t n ec essa ri ly i nvolve all i ts propert ies, an observer rna :see an organism as integra ting simultaneously several composi te uni ties thamayor may not al l b e soc ia l sys tems .

(iii) The str ucture of a society as a particular social system is determ ined both by the s tructure o f i ts a utopoi et ic componen ts a nd by the ac tuarel at ion s tha t hold between them whi le t hey int eg rat e it . Therefore, t he domain o f soci al phenomena, d ef in ed as the domain o f t he int er act ions and th:relations that an observer s ees taking place between the components ofs ociety, results from the autopoietic operation of the components of th:soc ie ty while they rea lize i t in the interplay of the ir propert ies.

( iv) In a s ociety, at any instance of observation, the structures of th.component s det ermine the proper ti es o f t he componen ts , t he p roper ti es 0the components realize the structure of the society, and the s tructure of ths ociety operates as a selector of the structure of its components by beingmedium in which they rea lize the ir ontogeny.

( v) An autopoietic s ys tem participates in the constitution of a sociasystem only to the extent that it par ticipates in it, that is , only as it r ealizethe relations proper to a component of the social system. Accor dingly, iJprinciple, an autopoietic system may enter or leave a social system at an:moment by just s atis fying or not satis fying the proper relations , and rna:participa te simultaneously or in succession in many diffe rent ones.

In what follows I shall pursue the consequences of these notions, an,whenever I speak of a social system or of a society as a social system of ;p ar ti cu lar kind , I speak of a sys tem def in ed a s a sys tem by the o rgani za ti orproposed above.

(3) A society defines the domain in which it is realized as a unity. Such.domain mayor may not include the components of the society itself, am


13/85

xxvi INTRODUCTIONmayor may not include other societies , but in any case i t const itu tes anoperationally independent medium that operates as: (a) a selector of the pathof structural change that the society follows in its individual history, and(b), if stable, a historical stabilizer of the structures that realize the selectedinvariant relations that define the society as a particular social system. Themore varied the medium of selection, the greater the domain of stabilizedrelations in the society and, hence, the more fixed the structure of its com-ponents. If the society i s a human society th is takes place in a language-centered culture, and the stabilization of the structure of the human com-ponents is realized through a cultural stabilization of the relations that theymust satisfy as socialentities.(4) To the extent that human beings are autopoietic systems, all theiractivities as social organisms must satisfy their autopoiesis. This they do inthe social domain through the fulfilment of the basic biological preferences(states of pleasure) and rejections (states of displeasure) that constitute theinmediate experiential domain in which they, as components of a society,necessarily realizetheir individual worlds and contribute to the determinationof the individual worlds of others. In man as a social being, therefore, allactions, however individual as expressions of preferences or rejections, con-sti tutively affect the lives of other human beings and, hence, have ethicalsignificance.(5) What determines the constitution of a social system arethe recurrentinteractions of the same autopoietic systems. In other words, any biologicalstabilization of the structures of the interacting organisms that results in therecurrence of their interactions, may generate a socialsystem. Amonghumanbeings the basic stabilizing factor in the constitution of a socialsystem isthephenomenon of love, the seeingof the other as a partner in some or all thedimensions of living. In these circumstances, when a human being makes thechoice of a particular way of living,apparent in his realization of a particularset of social relations, he makes a basic ethical choice in which he validates aworld for himself and for others that he has explicitly or implicitly acceptedas partners in living. Accordingly, the fundamental ethical problem that aman faces as an observer-member of a society isthe ethical justification ofthe particular relations of surrender of autonomy and individuality that hedemands from himself and from other members of the society that he gen-erates and validateswith hisconduct.(6) A social system is essentially a conservativesystem. Thisis sobecauseit is generated through the interactions of structure-determined autopoieticsystems andoperates asa medium that selects the path of ontogenic structural

INTRODUCTION xxviichange of its components, which, thus, become structurally coupled to it. Inour case, we as social beings generate, through our structure-determinedproperties, our societies as the cultural media that select ourindividual pathsof ontogenic change in a manner that leads each one of us to the structurethat makes us generate the particular societiesto which we belong. Asociety,therefore, operates as a homeostatic system that stabilizes the relations thatdefine it asa socialsystem of a particular kind.(7) In general, the domain of s tates of a sys tem asa composite uni ty i sdetermined by the properties of its components that realizeits organization.If some of these properties changebecause the structure of someof the com-ponents changes, then, while the systemeither changes itsproperties withoutchange of organization or disintegrates becoming somethingelse,the changedcomponents either integrate the system in a different manner or uncouplefrom it. This also applies to socialsystems, includinghuman societies, becauseit is the actual interplay of the properties of the components that constitutesa social system as an actual system in the spacein which these exist. In thesecircumstances, a change in the relations that define a society as a particularsocial system can only take place through a changein the properties of thecomponents that realize it. It follows that in a human society a socialchangecan only take place if the individual properties, and, hence, conduct, of itsmembers change.

(8) All that matters for the realization ofa society isthat the componentautopoietic systems should satisfy certain relations regardlessof the actualstructures (internal processes) through which they realizethem. Accordingly,hypocrisy plays an important role in the realization of human societies,per-mitting human beings under stress to feign having certain properties whichthey abandon assoon asthe stressis removed.This iswhy in a human societya social change takes place as a permanent phenomenon only to the extentthat i t is a cultural change: a revolution isa revolution onlyif i t isan ethicalrevolution.

(9) Interactions within a society are necessarily confirmatory of the rela-t ions that define it as a particular social system; if not, the organisms thatinteract do not interact as components of the society which they otherwiseintegrate. It is only through interactions operationally not defined withinthe society that a component organism can undergo interactions that leadto the selection, in its ontogeny, of a path of structural changenot confirma-tory of the society that i t integrates. This i swhy social creat ivi ty, as thegeneration of novel social relations, alwaysentails interactions operationallyoutside the society, and necessarily leads to the generation, by the creative


14/85

xxx INTRODUCTIONt ran si to ry and ci rcumstan ti al t o t he c re at ion o f r el at ion s tha t c on ti nuous lynega te t he ins ti tut ional iz at ion o f human abuse . Such a soci ety i s i n i ts e ssenc ean anarchist society, a society made for and by observers that would notsur re nder t hei r c ond it ion o f observe rs as t he ir only c laim to soc ial f ree domand mutual respect.

At this point ther e is either much more to say, or nothing - therefore, let usl et t he reader j udge. Thank you .

I. BIOLOGY OF COGNITION(1970)By

HUMBERTO R. MATURANA


15/85

xxviii INTRODUCTIONindividuals, of modes of conduct that either change the definingrel.ations~fthe society as a particular social system, or separate them fr~m It. Socialcreativity is necessarily antisocial in the socialdomain in which It t.akesplace.

(10) In general any organism, and in particular any human bel.ng,can besimultaneously a member of many social systems, such as a family, a club,an army, a polit ical party, a religion or a nation, and can operate in one.oranother without necessarily being in internal contradiction. A human beingoperating as an observer, however, can alway~de~ne a me~adomai~from theperspective of which he may seehis participation in the vanous socialsystemsthat he integrates, and find it contradictory. Conduct asobserverby a hum~nbeing implies that he stands operationally as if outside the various SOCialsystems that he otherwise integrates, and that hemay undergo in this mannerinteractions that do not confirm them. An observer always is potentiallyantisocial.(11) To grow as a member of a society consists in becoming structurallycoupled to it; to be structurally coupled to a society consists in havingthestructures that lead to the behavioral confirmation of the society. The spon-taneous course ofhistorical structural transformation of a society asa unity istoward its structural coupling to the medium in which it exists, and, there-fore toward the stabilization of the mechanisms that generate its definingrelations through the stabilization of the properties of its components. In thedomain of human societies this means the stabilization of human conduct.But the stabilization of human conduct always entails a restriction of crea-tivity through a restriction of the possible interactions of the individualhuman beings outside those prescribed by the society that they integrate.The extreme caseof this, of course, takes place in a totalitarian society of anykind. Or, in other words, the spontaneous course of the historical transforma-tion of a human society asa unity istowards totalitarianism; this isso becausethe relations that undergo historical stabilization are those that have to dowith the stabili ty of the society asa unity in a given medium, and not withthe well-beingof its component human beingsthat may operate asobservers.Any other course requires an ethical choice; it would not be spontaneous, itwould be a work of art, a product of human aesthetic design. If human beingswere not observers, or capable of being so,the stabilization of their propertieswould not appear to matter because they would not be able to desiresome-thing else.(12) We as human beings exist in a network of social systems and movefrom one to another in our daily activities. Yet, not all human beingscaughtin the mesh of relations generated in this network of social systems parti-

INTRODUCTION xxixcipate in it as socialbeings. A human being that through his interactions withother human beingsparticipates in interactions proper to their socialsystemina manner that does not involve his autopoiesis as a constitutive feature of it,isbeing usedby the socialsystem but isnot one of its members. If the humanbeing cannot escape from this situation because his life isat stake, he isundersocialabuse.(13) Al l kinds of societies are biologically legit imate. Yet not all areequally desirable as systems in which an observer human being may wishtolive. The capacity that man has asa language-centeredsocialbeingof becom-ing an observer, and thus of operating as if he were external to the situationin which he finds himself, allows him, if he has the proper experiences, tocontemplate the societies that he integrates and to like or dislike them. Ifthe observer human being defines a metadomain from the perspective ofwhich some of the defining relations of the society are undesirable he dislikesit, and if he acts accordinglyhe becomes antisocialand may cometo validateanother society with his conduct. A totalitarian society restricts this possi-bility either by specifying the experiences that its components may have,sothat they do not operate asobservers,or by uncouplingthe dissidentssothatthey may not seduce others to be observersasthemselves.However,there arecertain experiences that cannot be fully specifiedin a human society withoutdestroying the basic individual structural plasticity needed for the establish-ment of consensual domains and the generation of language and, hence, forhuman creativity in general. Loveis one of these experiences, and aslong asman has a language he can become an observer through the experienceoflove.

(14) When a human being 'A' encounters another human being 'B' andloveshim or her, he sees 'B' in a social context and becomes an observerof the society that 'B' integrates. 'A' may l ike or may not l ike what he seesin reference to 'B' and act accordingly, becoming antisocial if he does notlike what he sees. An absolute totalitarian society must negate love as anindividual experience because love, sooner or later, leads to an ethical evalua-tion of the society that the loved one integrates.

(IS) A human society in which to see all human beings as equivalent tooneself, and to love them, is operationally legit imate without demandingfrom them a largersurrender of individuality andautonomy than the measurethat one iswillingto accept for oneself while integ~atingit as anobserver,is aproduct of human art , that is, an artificial society that admits changeandaccepts every human being as not dispensable. Such a society isnecessarilyanon-hierarchical society for which all relations of order are constitutively


16/85

I dedicate this book to my wife, MariaMontanez Luna. I could not have writ teni t if I had not l ived asI havel ived ,and forthe most part of my existence the threadofmy being has been and is braided with hers.Therefore, strictly, although I wrote thisbook, she is, in every respect , asmuch itsauthor asI am. ThisI wish both to acknowl-edgeand to thank with deep love.

T A BLE OF CONTENTS

2DEDICATIONI. INTRODUCTION

II. THE PROBLEMCOGNITIVE FUNCTION IN GENERAL 8Ill.

8A. The Observer9B. The Li ving Sys tem11C . Evolut io n13D. The Cognitive Process

COGNITIVE FUNCTION IN PARTICULAR 15IV.Nerve CeUs 15A.

19B. Architecture21C. Functi on 22D . Repr es en t at io n26E. Description 29F. Thinking30G. Natural Language35H. Memory and Learning39I. The Observer

PROBLEMS IN THE NEUROPHYSIOLOGY OF COGNITION 41V.CONCLUSIONS 48VI.POST SCRIPTUM 57VII.


17/85

HUMBERTO R. MATURANA

BIOLOGY OF COGNITION

I. INTRODUCTIONMan knows and his capacity to know depends on his biol ogical int egrit y;fu rthe rmore , he knows tha t he knows. As a basic psycholog ica l and , hence ,biological function cognition guides h is hand ling of the unive rs e and know-ledge gives certainty to his acts; objective knowledge seems possible andthrough objective knowledge the universe appears systematic and predictable.Yet knowledge as an exper ience issomething personal and private that cannotbe transferred, and that which one believes to be transferable, objectiveknowledge, must always be cr ea ted by the l is tene r: the l ist ener understands,and objective knowledge appears transferred, only if he is prepared to under -stand. Thus cognition as a biological function is such that the answer to thequestion, 'What is cognition?' must arise f rom understanding knowledge andthe knower thr ough the lat te r' s capaci ty to know.

Such is my endeavor.

EpistemologyThe basic claim of science is objectivity: i t a ttempts, through the applicationof a well defined methodology, to make st atements about t he universe. Atthe very root of thi s claim, however , l ie s i t s weakness : the a priori assumptiontha t objec tive knowledge cons ti tu te s a desc ript ion of that which i s known.Such assumption begs the questions , 'What isi t to know?' and 'Howdo weknow?'.Biology(a ) The greates t h indrance in the unders tand ing of the l iv ing organ iza tion l ie sin the impos sibi li ty of account ing for i t by the enumera tion of i ts p roper ti es ;it must be understood as a unit y. But if the organism is a unity, in what sensear e i ts componen t propert ie s i t par ts? The organ ismic approach does no t an-swer thi s ques tion , i t mere ly re states i t by insi st ing that the re ar e elements oforgan iza tion tha t subordinate each part to the whole and make the organ ism

5


18/85

6 HUMBERTO R. MATURANA

a unity [Cf. Bertalanffy, 1960]. The questions 'How does this unity arise?'and 'To what extent must i t be considereda property of the organizat ionof the organism, asopposed to a property emergingfrom its mode of life' !'remain open. A similar difficulty exists for the understanding of the func-tional organization of the nervous system, particularly if one considers thehigher functions of man. Enumeration of the transfer functions of all nervecel ls would leave us with a l is t, but not with a system capable of abstractthinking, description, and self-description. Such an approach would beg thequestion, 'How does the livingorganization give rise to cognition in generaland to self-cognition in particular?'(b) Organisms are adapted to their environments, and i t has appeared ad-equate to say of them that their organization represents the 'environment'in which they l ive, and that through evolution they have accumulated in-formation about it, coded in their nervous systems. Similarlyit has been saidthat the sense organsgather information about the 'environment', and throughlearning this information is coded in the nervous system [Cf.Young, 1967].Yet this general view begs the quest ions, 'What does i t mean to "gatherinformation"?' and 'What is coded in the genetic and nervous systems?' .A successful theory of cognition would answer both the epistemologicaland the biological quest ions. This I propose to do, and the purpose of thisessay is to put forward a theory of cognition that should provide an epis-temological insight into the phenomenon of cognition, and an adequate viewof the functional organization of the cognizant organism that gives rise tosuch phenomena as conceptual thinking, language,and self-consciousness.

In what fol lows I shall not offer any formal definit ions for the variousterms used, such as 'cognit ion' , ' li fe ', or ' in teract ion' , but I shall let theirmeaning appear through their usage. This I shall do because I am confidentthat the internal consistency of the theory willshow that these terms indeedadequately refer to the phenomena I am trying to account for, and becauseI speak asan observer, and the val idity of what I say at any moment has i tsfoundation in the validity of the whole theory, which, I assert, explains whyI can say i t. Accordingly, I expect the complete work to givefoundat ion toeach of i ts parts , which thus appearjust ified only in the perspective of thewhole.Note: I sha ll be speaking of the organism as a unity , but when I wrote this essay Iwasnot aware that the word uni t did not always qui te mean uni ty. Since I cannot nowcorrect this. I beg the reader to bear this in mind.

II. THE PROBLEM

(1) Cognition is a biological phenomenon and can only be understood assuch; any epistemological insight into the domain of knowledgerequires thisunderstanding.(2) If such an insight isto beattained, two questions must be considered:

Whatis cognition asa function?What is cognition asa process?

Whatfollowsshould answer these two questions.

7


19/85

III. COGNITIVE FUNCTION IN GENERAL

THE OBSERVER

(1) Anything said is said by an observer. In his discourse the observer speaksto another observer, who could be himself; whatever applies to the oneapplies to the other as well. The observer is a human being, that is, a livingsystem, and whatever applies to living systems applies also to him. .

(2) The observer beholds simultaneously the entity that he considers(an organism, in our case) and the universe in which it lies (the organism'senvironment). This allows him to interact independently With both and tohave interac tions tha t ar e neces sa ri ly out side the domain of interact ions ofthe observed entity.(3) It i s an a tt ribu te of the obse rver to be able to interac t independen tlywith the observed entity and with its relations; for him both are units ofinteraction (entities).(4) For the observer an entity is an entity when he can describe it. Todesc ribe i s to enumera te the ac tual o r poten ti al interact ions and relat ions ofthe desc ribed en ti ty . Accord ingly, the observer can desc ribe anen ti ty only i fthere is at least one other entity from which he can distinguish it and withwhich he can observe it to interact or relate. This second entity that servesasa ref erence for the desc ript ion can be any ent ity, bu t the ul tima te ref erencefor any descr iption is the observer himself .(5) The set of all interactions into which an entity can enter is its domainof interact ions. The se t o f a ll re lat ions (interac tions thr ough the obser ve r) inwhich an ent ity can be observed i s i ts doma in of re lat ions. Th is lat te r domainl ies within the cognit ive domain of the observer. An entity is an entity if ithas a domain of interac tions , and i f thi s doma in inc ludes interact ions wi th theobse rver who can speci fy f or i t a domain of re lat ions. The obse rver can def inean entity by specifying its domain of interactions; thus part of an entity, agroup of entit ies, or their relations , can be made units of interactions (enti ties)by the observer.(6) The observer can define himself as an entity by specifying his owndomain of interac tions ; he can a lways r emain an obse rver of these interact -ions, which he can treat as independent entit ies.

8

.J

COGNITIVE FUNCTION IN GENERAL 9(7) The observer is a livingsystem and anunderstanding of cognition asabiological phenomenon must account for the observer and his role in it.

THE LIVING SYSTEM(1 ) Living sys tems ar e uni ts of interact ions; they ex ist in an ambience . From apurely b io logica l point o f v iew they cannot be unders tood independen tly ofthat part of the ambience with which they interact: the niche; nor can theniche be def ined independently of the living system that specifies it.

(2) Living systems as they exist on earth today are characterized byexergonic metabolism, growth and internal molecular replication, all organizedin a c losed causal c ircu la r p rocess tha t al lows for evolut iona ry change in theway the c irculari ty i s maintained, but not f or the loss of the c irculari ty i ts el f.Exergonic metabolism is required to provide energy for the endergonicsynthesis of specific polymers (proteins, nucleic acids, lipids, polysaccharides)f rom the corresponding monomers, that is, for growth and replication; specialreplication procedures secure that the polymers synthesized be specific, thatthey should have the monomeric sequence proper to their class; specificpolymers (enzymes) are required for the exergonic metabolism and thesynthesis of specific polymers (proteins, nucleic acids, lipids, polysaccharides)[Cf. Commoner, 19651.

This circular organization constitutes a homeostatic system whose functioni s to pr oduce and maintain thi s very same c ircu la r o rganizat ion by de te rmin-ing that the components that specify it be those whose synthesis orma in tenance i t s ecures . Furthe rmore, thi s ci rcular o rganiza tion def ines aliving system as a unit of interactions and is essential for its maintenanceas a unit; that which is not in it is external to it or does not exist. The circularorgan iza tion in which the components that specify it are those whose syn-thesis or maintenance it secures in a manner such that the product of theirfunct ioning i s the same funct ioning organ izat ion tha t p roduces them, i s theliving organization,

(3) It i s the ci rculari ty of i ts o rgan izat ion tha t makes a l iv ing system a uni tof interactions, and it is this circularity that it must maintain in order toremain a living system and to retain its identity through dif ferent interactions.All the peculiar aspects of the dif ferent kinds of organisms are super imposedon thi s bas ic c ircu la ri ty and a re subse rv ien t to i t, securing i ts cont inuancethrough success ive interactions in an always changing environment. A livingsystem def ines thr ough i ts o rgan izat ion the domain of al l interac tions intowhich it can possibly enter without losing its identity, and it maintainsits identity only as long as the basic circularity that defines it as a unit of


20/85

10 HUMBERTO R. MATURANAinteractions remains unbroken. Str ictly, the identity of a unit of interactionstha t o the rwise changes cont inuously i smaintained only with re spect to theobserver, for whom its character as a unit of interactions remains unchanged.

(4 ) Due to the ci rcular na tu re of i ts o rganizat ion a l iving system has a sel f-referring domain of interactions (it is a self-referring system), and its conditionof being a unit of interactions is maintained because its organization hasfunct ional sign ifi cance only in re lat ion to the maintenance of i ts c ircu la ri tyand defines its domain of interactions accordingly.(5 ) Living sys tems as uni ts o f interac tions speci fi ed by the ir cond it ion ofbeing l iv ing systems cannot en te r into interact ions that a re not speci fi ed bytheir organization. The circulari ty of their organization continuously bringsthem back to the same internal s tate ( same with re spect to the cycl ic process ).Each internal state requires that certain conditions (interactions with theenvironment) be satisfied in order to proceed to the next state. Thus, thecircular organization implies the prediction that an interaction that tookplace once will take place again. If this does not happen the system dis-integra tes ; i f the pr ed ic ted interact ion does take place, the system main tainsits integrity (identity with respect to the observer) and enters into a newpredict ion. In a cont inuously chang ing env ironment these pred ic tions canonly be successful if the environement does not change in that which ispredicted. Accordingly, the predictions implied in the organization of thel iv ing system are not predict ions of pa rt icular event s, but o f c la sse s of inter-act ions. Every inter ac tion i s a part icu la r inter ac tion, bu t ever y predict ion i s aprediction of a class of interactions that is defined by those features of itse lements tha t wi ll al low the l iv ing system to r etain i ts c ircu la r or ganiza tionafter the interaction , and thus, to interact aga in . Th is makes l iv ing sys temsinferential systems, and their domain of interactions a cognitive domain.

(6) The niche is defined by the classes of interactions into which anorganism can enter . The environment is def ined by the classes of interactionsinto which the observer can enter and which he treats as a context for hisinter ac tions with the obse rved organ ism. The obser ve r beholds or gani sm andenvironment simultaneously and he conside rs a s the n iche of the organi smthat part of the environment which he observes to lie in its domain of inter-actions. Accordingly, as for the observer the niche appear s as part of theenvironment, for the observed organism the niche constitutes its entiredomain of interactions, and as such it cannot be part of the environmentthat lies exclusively in the cognitive domain of the observer. Niche andenvi ronment, then, inter sec t only to the ex tent tha t the obse rver (inc lud inginst ruments) and the organ ism have comparable organ izat ions, bu t even then

COGNI TI VE FUNCTION I N GENERAL II

there are always parts of the environment that lie beyond any possibilityof intersection with the domain of interactions of the organism, and thereare parts of the niche that lie beyond any possibility of intersection withthe domain of interact ions of the obser ve r. Thus for every l iv ing sys tem i tso rgan izat ion implies a pr ed ict ion of a n iche , and the n iche thus predicted asa domain of class es of inter ac tions const itutes i ts en ti re cogn it ive rea li ty .If an organism interacts in a manner not prescribed by its organization, itdoes so as something different from the unit of interactions defined by itsbas ic c ircu la ri ty , and thi s interact ion rema ins out side i ts cogn it ive domain,although it may well l ie within the cognitive domain of the observer.

(7 ) Every un it o f interact ions can part icipa te in interact ions re levant toother, more encompassing units of interactions. If in doing this a livingsys tem does not lose i ts ident ity, i ts n iche may evolve to be con ta ined by thelarger unit of interactions and thus be subservient to it. If this larger unit ofinteractions is (or becomes) in turn also a self-referring system in whichi ts componen ts ( themse lves se lf-re fer ring systems ) are subserv ient to i tsmaintenance as a unit of interactions, then it must itself be (or become)subservient to the maintenance of the circular organization of its components.Thus, a part icular s el f- ref erring sys tem may have the ci rcu la r o rgan izat ionof a l iv ing system or par take funct ional ly of the c ircu la r o rganizat ion of i tscomponents, or both. The society of bees (the honey producing bees) is anexample of a thi rd order s el f- refe rring sys tem of thi s kind ; i t has a c ircu la ror ganiza tion superimposed on the second order s el f-r efe rring systems thatare the bees, which in turn have a circular organization superimposed onthe first order living systems that are the cells; all three systems with theirdomains of interac tions a re subord ina ted both to the ma intenance of them.selves and to the maintenance of the others.

EVOLUTION(1 ) Evolut ionar y change in l iv ing systems is the resul t of tha t a spec t o f the ircircular organization which secures the maintenance of their basic circulari ty,al lowing in each reproduct ive s tep for changes in the way thi s ci rcu la ri ty i smaintained. Reproduction and evolution are not essential for the livingorganization, but they have been essential for the his torical transformationof the cognitive domains of the living systems on ear th.

(2) For a change to occur in the domain of interactions of a unit ofinteract ions withou t i ts los ing i ts ident ity with re spec t to the obser ve r i t mustsuffe r an internal change . Converse ly , i f an inter na l change occurs in a uni t o f


21/85

12 HUMBERTO R. MATURANAinteractions, without itslosing its identity, its domain of interactions changes.A living system suffers an internal change without loss of identity if thepredictions brought forth by the internal changeare predictions which do notinterfere with its fundamental circular organization. A systemchanges only ifits domain of interactions changes.(3) After reproduction the new unit of interactions has the samedomainof in teractions as the parental one only if i t has the same organizat ion.Conversely,the new unit of interactions has adifferent domainof interactionsonly if its organizaticn is different, and hence, implies different predictionsabout the niche.(4) Predictions about the niche areinferences about classesof interactions.Consequently, particular interactions which are indistinguishable for anorganism may be different for an observer i f he has a different cognit ivedomain and can describe them as different elements of a classdefined by theconduct of the organism. The same applies to interactions that are identicalfor the organismbut different for (have different effects) its different internalparts. Such interactions may result in different modifications of the internalstates of the organism and, hence, determine different paths of changein itsdomain of interactions without loss of identity. These changes may bringabout the production of offspring having domains of interactions differentfrom the parental ones. If this is the caseand a new system thus producedpredicts a niche that cannot be actualized, it disintegrates; otherwise it main-tains its identity and a new cycle begins.(5) What changes from generation to generation in the evolution of livingsystems are those aspects of their organization which are subservient to themaintenance of their basic circulari ty but do not determine it, and whichallow them to retain their identity through interactions; that is,what changesis the way in which the basic circulari ty is maintained, and not this basiccircularity in itself. The manner in which a livingsystem iscompounded asaunit of interactions, whether by a singlebasic unit, or through the aggregationof numerous such units (themselves livingsystems) that together constitute alarger one (multicellular organisms), or still through the aggregation of thesecompound units that form self-referring systems of even higher order (insectsocieties, nations) is of no significance; what evolvesis alwaysa unit of inter-actions defined by the way in which it maintains its identity. The evolutionof the livingsystems isthe evolution of the niches of the units of interactionsdefined by their self-referring circular organization, hence, the evolution ofthe cognitive domains.

COGNITIVE FUNCTION IN GENERAL 13THE COGNITIVE PROCESS

(J) A cognitive system is a system whose organization defines a domain ofinteractions in which it can act with relevance to the maintenance of itself,and the process of cognition is the actual (inductive) acting or behavinginthis domain. Living systems are cognitive systems, and livingasa processis aprocess of cognition. This statement is valid for all organisms, with andwithout a nervous system.(2) If a livingsystem enters into a cognitiveinteraction, its internal state ischanged in a manner relevant to its maintenance, and it enters into a newinteraction without loss of its identity. In an organism without a nervoussystem (or its functional equivalent) i ts interactions are of a chemical orphysical nature (a molecule is absorbed and an enzymatic processis initiated;a photon is captured and a step inphotosynthesis iscarried out). For suchanorganism the relations holding between the physical events remain outsideits domain of interactions. The nervous system enlarges the domain of inter-actions of the organism by making its internal states also modifiable in arelevant manner by 'pure relations', not only by physicalevents; the observerseesthat the sensors of an animal (say, a cat) are modified bylight, and thatthe animal (the cat) is modified by a visibleentity (say, a bird). The sensorschange through physical interactions: the absorption of light quanta; theanimal is modified through its interactions with the relations that holdbetween the activated sensors that absorbed the light quanta at the sensorysurface. The nervous system expands the cogniti ve domain of the li vingsystem by making possible interactions with 'pure relations'; it does notcreate cognition.(3) Although the nervous system expands the domain of interactions ofthe organism by bringing into this domain interactions with 'pure relations',the function of the nervous system is subservientto the necessarycircularityof the livingorganization.(4) The nervous system, by expanding the domain of interactions of theorganism, has transformed the unit of interactions and has subjected actingand interacting in the domain of 'pure relations' to the process of evolution.As a consequence, there are organismsthat include as asubset of their possibleinteractions, interactions with their own internal states (as states resultingfrom external and internal interactions) as if these were independent entities,generating the apparent paradox of including their cognitive domain withintheir cognitive domain. In us this paradox isresolvedby what we call'abstractthinking', another expansion of the cognitivedomain.


22/85

14 HUMBERTO R. MATURANA(5 ) Fur thermore, the expansion of the cogni tive domain into the domain

of 'pure relations' by means of a nervous system allows for non-physicalinteract ions between organ isms such tha t the interact ing organ isms orien teach othe r toward interact ions wi thin the ir respect ive cogn it ive doma ins.Herein lies the basis for communication: the orienting behavior becomes arepresentation of the interactions toward which it orients, and a unit ofinteractions in its own terms. But this very process generates another apparentparadox: there are organisms that generate representations of their owninteractions by specifying enti ties with which they interact as if these be-longed t o an independent domain, while as representations they only mapthe ir own inter ac tions. I n us thi s paradox is re so lved simultaneously in twoways:

(a) Webecome observers through recursively generating representationsof our inter ac tions, and by interact ing with sever al repre sentat ions simul-taneously we generat e relations with the representations of which we canthen interact and repea t thi s p roces s r ecurs ively, thus rema in ing in a domainof interactions always larger than that of the representations .

(b) We become self-conscious through sel f-observat ion; by makingdescriptions of ourselves (representations), and by interacti ng wit h ourdesc ript ions we can descr ibe ourse lves desc ribing ours elves , in an endlessrecursive process.

IV. COGNITIVE FUNCTION IN PARTICULAR

NERVE CELLS

(i) The neuron is the anatomical unit of the nervous system because it is ace ll , and as such i t i s an independent integra ted sel f-re fe rring me tabol ic andgenetic unit (a living system indeed) .

(2 ) Ana tomica lly and func tional ly a neuron i s fo rmed by a col lector ar ea( dendri te s, and in some cases , a lso the ce ll body and part o f the axon) uni tedvia a dist ributive element (the axon, and in some cases, also the cell bodyand main dendri te s), capable of conduct ing propaga ted sp ikes to an e ff ec to rarea formed by the terminal branching of the axon. The functional state ofthe col lector a rea depends on both i ts internal state (re fe rence s ta te) and onthe state of act iv ity of the e ff ec to r a reas synapsing on i t. Corre sponding ly ,the s tate of ac tivi ty of the e ff ec to r a rea depends on both the t ra in of impulse sgenerated at t he corresponding collector area and on the pre-synaptic andnon-synaptic interactions with dis tr ibutive elements and other effector areasthat may take place in the neuropil and in the immediate vicinity of the nextcollector areas. This is true even in the case of amacrine cells, in which thecol lec to r and ef fector a reas may be interming led. The d ist ribu tive elementdetermines where the effector exerts i ts inf luence.

(3) Whether one or two branches of a bifurcating axon are invaded by anerve impulse propagating along it depends on thei r re la tive d iame ter and onthe state of polari zat ion of the ir membr anes a t thei r o rigin in the b ifu rcat ionzone. As a result, the pattern of effector activity, that is, the pattern ofbranch invasion whi~h a t ra in of impulse s determines in the branches of thedistributive element and effector area of a neuron, depends (i) on the spikeinterval d is tribu tion of the t rain of impulse s, which de te rmines the t ime thatthe axonal membrane at the branching zone has for recovery before thea rriva l o f the nex t spike, and (ii) on the non-synaptic inf luences which, in theform of local water and ion movements caused by the elect rical activit y ofneighboring elements, may produce d iame te r and polarizat ion changes a t thebranching zones, and thus modify t he invasibility of t he branches by thearriving spikes.

(4) At any moment the state of activity of a nerve cell, as represented by15


23/85

16 HUMBERTO R . MATURANAthe pattern of impulses travelling along its distributive element, isa functiono f the spa tio -tempora l con figurat ion o f i ts input , as dete rmined by therelative activity holding between the afferent neurons, that modulates thereference state proper of the coll

maturana and varela - autopoiesis

Documents