sitemap Whitaker (1995): ARCHIVE: Autopoietic Theory: Deeper Discussion
Spacer Spacer Spacer Spacer Spacer Spacer Spacer Spacer Spacer Spacer Spacer Spacer Spacer Spacer Spacer Spacer Spacer Spacer Spacer Spacer
ARCHIVE:   Self-Organization, Autopoiesis, and Enterprises (Whitaker, 1995) Enola Gaia
This is an archive edition of a web publication no longer accessible at its original location.

At the 1995 European CSCW Conference in Stockholm, I was asked to create an online essay to be hosted by ACM (Association for Computing Machinery) SIGOIS (Organizational Information Systems special interest group). The resultant set of webpages went online circa December 1995. Since then, the SIGOIS web resource that contained this material was discontinued.

Randall Whitaker

Autopoietic Theory: Deeper Discussion

The Observer

When a cognitive system makes distinctions which cleave its environment into 'object' and 'other', it is operating as an observer. In Maturana's own words:

'An observer is a ... living system who can make distinctions and specify that which he or she distinguishes as a unity, as an entity different from himself or herself that can be used for manipulations or descriptions in interactions with other observers.' (Maturana, 1978a, p. 31)

The observer is one of the key concepts in autopoietic theory, because:

'Observing is both the ultimate starting point and the most fundamental question in any attempt to understand reality and reason as phenomena of the human domain. Indeed, everything said is said by an observer to another observer that could be him- or herself.'

(Maturana, 1988, p. 27)

The fundamental operation in observing is that of distinction -- '...the pointing to a unity by performing an operation which defines its boundaries and separates it from a background.' (Maturana, 1975, p. 325) Through the recursive distinguishing of entities through action, the observer is ' to operate as if external to (distinct from) the circumstances in which he finds himself.' (Op.cit., p. 315) However, the observer is not actually standing apart from those circumstances. This is due to the fact that the entire and the only domain in which he/she operates is that of his/her closed (self-interconnected) nervous system. The nervous system's connectivity and closure permit interactions among its own states at time t1 to determine its states at time t2. This circular interaction allows for '... infinite recursion with continuous behavioural change.' (Op.cit., p. 324)

The notion of the observer circumscribes all enquiry and all discussion. The qualification of any observation with respect to the vantage point of a given observer makes autopoietic theory inherently relativistic with respect to the person of the observer. Second, the resulting qualification of any set of observations over time with respect to the vantage events of a given observer makes autopoietic theory inherently relativistic with respect to the history of the observer. Third, since shared or collectively negotiated descriptions of experience (e.g., recollections [past], consensus [present], plans [future]) are qualified with respect to the interactions among given observers, autopoietic theory is inherently relativistic with respect to the persons of interacting observers and the history of interactions among them.


Organization and Structure

'The relations that define a machine as a unity, and determine the dynamics of interactions and transformations which it may undergo as such a unity, constitute the organization of the machine.' (Maturana & Varela, 1980, p. 77)

Maturana (1975) notes 'organization' comes from the Greek and means 'instrument'. By using this word for the essential, defining character of a system he focuses attention on '...the instrumental participation of the components in the constitution of the unity.' (Op.cit., p. 315) It is the organization of a system which defines its identity, its properties as a unity, and the frame within which it must be addressed as a unary whole.

A systemic unity's organization is specifically realized through the presence and interplay of components in a given space. These comprise the unity's structure. Maturana (1975) points out the word 'structure' comes from the Latin meaning 'to build'. He employs this allusion in assigning to this label '...the actual components and ... the actual relations which these must satisfy in their participation in the constitution of a given unity.' (Op.cit., pp. 315-316) Structure does not determine the overall character of a unity; it determines only "...the space in which it exists and can be perturbed." (Ibid.)

A unity may change structure without loss of identity, so long as its organization is maintained. Maturana and Varela's distinction between organization and structure provides a basis for sorting out descriptions of systems into their abstract and concrete aspects. Formally speaking:

'The organization of a machine (or system) does not specify the properties of the components which realize the machine as a concrete system, it only specifies the relations which these must generate to constitute the machine or system as a unity. Therefore, the organization of a machine is independent of the properties of its components which can be any, and a given machine can be realized in many different manners by many different kinds of components. In other words, although a given machine can be realized by many different structures, for it to constitute a concrete entity in a given space its actual components must be defined in that space, and have the properties which allow them to generate the relations which define it.' (Maturana & Varela, 1980, p. 77)

A 'nitty-gritty' illustration of the distinction is given in the 1987 book The Tree of Knowledge:

' a toilet the organization of the system of water-level regulation consists in the relations between an apparatus capable of detecting the water level and another apparatus capable of stopping the inflow of water. The toilet unit embodies a mixed system of plastic and metal comprising a float and a bypass valve. This specific structure, however, could be modified by replacing the plastic with wood, without changing the fact that there would still be a toilet organization.' (Maturana & Varela, 1987, p. 47)

The organization / structure dichotomy is graphically illustrated in the work of the 16th Century Italian painter Giuseppe Arcimboldo, who devised remarkable portraits in which the faces are composed of (e.g.) fruits, vegetables, seafood, etc. His fanciful art realized a discernible facial 'organization' through a 'structure' of novel components.


Autopoiesis and Autonomy


Maturana and Varela's central concept is that of autopoiesis. According to Maturana (Maturana and Varela, 1980, p. xvii) the term was coined around 1972 by combining the Greek auto (self-) and poiesis (creation; production). The concept is defined formally as follows:

'An autopoietic system is organized (defined as a unity) as a network of processes of production (transformation and destruction) of components that produces the components that:

  1. through their interactions and transformations continuously regenerate and realize the network of processes (relations) that produced them; and
  2. constitute it (the machine) as a concrete unity in the space in which they [the components] exist by specifying the topological domain of its realization as such a network.'

(Varela, 1979, p. 13)

Any unity meeting these specifications is an autopoietic system, and any such autopoietic system realized in physical space is a living system. The particular configuration of a given unity -- its structure -- is not sufficient to define it as a unity. The key feature of a living system is maintenance of its organization, i.e, preservation of the relational network which defines it as a systemic unity. Phrased another way, '...autopoietic systems operate as homeostatic systems that have their own organization as the critical fundamental variable that they actively maintain constant.' (Maturana, 1975, p. 318)

Autopoietic theory is the primary (perhaps the only...) example of a definition for life which is framed purely with respect to a candidate system in and of itself. If you go back and check most definitions (e.g., in a biology text), you are likely to find nothing more coherent than a list of features and functional attributes (e.g., 'reproduction', 'metabolism') which describe what living systems do, but not what they are. For this reason, autopoiesis has become a topic of interest in the recent field of artificial life (Alife)


During the mid- to late 1970's, Varela expanded on autopoietic theory's original formalizations to delineate the systemic attribute of autonomy, of which autopoiesis is a subset. Autonomous systems are:

'...defined as a composite unity by a network of interactions of components that (i) through their interactions recursively regenerate the network of interactions that produced them, and (ii) realize the network as a unity in the space in which the components exist by constituting and specifying the unity's boundaries as a cleavage from the background...'

(Varela, 1981a, p. 15)

The difference between autonomy and autopoiesis is that autopoietic systems must produce their own components in addition to conserving their organization. As we shall see later, this difference has played a large role in the debates over the extent to which social systems can be characterized as autopoietic.

This more general class of autonomous systems are defined by their organizational closure:

'That is, their organization is characterized by processes such that

  1. the processes are related as a network, so that they recursively depend on each other in the generation and realization of the processes themselves, and
  2. they constitute the system as a unity recognizable in the space (domain) in which the processes exist.'

(Varela, 1979, p. 55)

It is important to note that this property of 'closure' does not make autonomous systems 'closed' in the classic cybernetic sense of 'isolated from the environment; impervious to environmental influence'. 'Closure' doesn't mean autonomous systems are unresponsive; it only means that their changes of state in response to changes in their medium are realized and propagated solely within the network of processes constituting them (as they are defined). The difference has more to do with the way a system is defined than how that system (once defined) operates. A fuller explanation of this point can be obtained in Varela (1979).


Domains and Spaces

A domain is a description for the "world brought forth" -- a circumscription of experiential flux via reference to current states and possible trajectories. Maturana and Varela define a number of domains in developing autopoietic theory's formal aspects into a phenomenological framework:

Domain of interactions
'...the set of all interactions into which an entity can enter...' (Maturana & Varela, 1980, p. 8).

Domain of relations
'...the set of all relations (interactions through the observer) in which an entity can be observed...' (Ibid.).

Phenomenological domain
That set of actions and interactions '...defined by the properties of the unity or unities that constitute it, either singly or collectively through their transformations or interactions.'(Varela, 1979, p. 46).

Cognitive domain
the set of '... all the interactions in which an autopoietic system can enter without loss of identity...' (Maturana & Varela, 1980, p. 136) An observer's cognitive domain circumscribes '...all the descriptions which it can possibly make.' (Op.cit., p. 119).

Consensual domain
'.. a domain of interlocked (intercalated and mutually triggering) sequences of states, established and determined through ontogenic interactions between structurally plastic state-determined systems.' (Maturana, 1975, p. 316)

Linguistic domain
'...a consensual domain of communicative interactions in which the behaviorally coupled organisms orient each other with modes of behavior whose internal determination has become specified during their coupled ontogenies.' (Maturana & Varela, 1980, p. 120)

The term space is used by Maturana and Varela to connote a static referential background within which systems are defined. The only specific 'space' included in basic autopoietic theory is the physical space -- i.e., the world of matter and energy addressed by (e.g.) physical sciences. Both Maturana (e.g, 1978b) and Varela (1979) make allowance for other spaces in which unities can be discerned, but neither has explicitly delineated examples of autopoiesis in other spaces.


Structural Determination

The basic thrust of the principle of structural determination is that the behavior of a system is constrained by its constitution. The set of potential system changes is circumscribed by:

  • the system's range of potential structural transformations

  • the set of potential perturbations impinging upon the system

Actual change is compensable behavior by the system's structure under perturbation by the environment and / or other systems in the course of its operation (cf. 'structural coupling', defined below). While a given perturbation may 'trigger' a change of system state, the particular change triggered is a function of the system's own organization and structure. Since 'structure' refers to any constitutive element of a discerned unity, structural determination concerns the manner in which observed (-able) phenomena are explained, not some formalized manner in which those phenomena objectively occur. As such, structural determination is an epistemological qualification, not a recourse to materialistic reductionism.

Structural determination should not be equated with strict causal determinism, in which all specific interactions are predetermined. It only means the space of all possible classes of interactions is determined. For example, in re-engineering an enterprise, the subject's structure does not uniquely predict its best new form. However, its structure circumscribes the range of new forms into which it can evolve without violating its organization (i.e., ceasing to exist as its current identity). Structural determination does not constrain the set of interactions in which a system can be observed to engage -- only the set in which that system can observe itself to be engaged:

'If the living system enters into an interaction not prescribed by its organization, it enters it not as the unit of interactions defined by this organization ... and this interaction remains outside its cognitive domain.'(Maturana, 1970a, p. 6)

This point is important to enterprise analysts and (re-)engineers. To the extent they proceed as observers 'external' to everyday operations, they engage enterprises at the intersection of the enterprise's domain of operations and their own domain of analytical activity. The behavior analysts observe in this 'intersection zone' may not be either representative of, or defined in accordance with, the domain of enterprise operations in which it is ordinarily realized.


Structural Coupling

Structural coupling is the term for structure-determined (and structure-determining) engagement of a given unity with either its environment or another unity. It is '...a historical process leading to the spatio-temporal coincidence between the changes of state..' (Maturana,1975, p. 321) in the participants. As such, structural coupling has connotations of both coordination and co-evolution.

Case 1: A System Coupling with its Environment

'If one of the plastic systems is an organism and the other its medium, the result is ontogenic adaptation of the organism to its medium: the changes of state of the organism correspond to the change of state of the medium.' (Maturana, 1975, p. 326)

'(T)he continued interactions of a structurally plastic system in an environment with recurrent perturbations will produce a continual selection of the system's structure. This structure will determine, on the one hand, the state of the system and its domain of allowable perturbations, and on the other hand will allow the system to operate in an environment without disintegration. ' (Varela, 1979, p. 33)

Case 2: A System Coupling with Another System

'If the two plastic systems are organisms, the result of the ontogenic structural coupling is a consensual domain.' (Maturana, 1975, p. 326)

A consensual domain is defined as '.. a domain of interlocked (intercalated and mutually triggering) sequences of states, established and determined through ontogenic interactions between structurally plastic state-determined systems.' (Maturana, 1975, p. 316). Because consensual domains are defined both by the structures of their participants and the history by which they came to exist, they are not reducible to descriptions framed only in terms of either:

'In each interaction the conduct of each organism is constitutively independent in its generation of the conduct of the other, because it is internally determined by the structure of the behaving organism only; but it is for the other organism, while the chain [of interactions] lasts, a source of compensable deformations that can be described as meaningful in the context of the coupled behavior.' (Varela, 1979, pp. 48 - 49)

Phrased in a slightly different way, the participating systems reciprocally serve as sources of compensable perturbations for each other. Such interactions are 'perturbations' in the sense of indirect effect or effectuation of change without having penetrated the boundary of the affected system. They are 'compensable' in the senses that (a) there is a range of 'compensation' bounded by the limit beyond which each system ceases to be a functional whole and (b) each iteration of the reciprocal interaction is affected by the one(s) before. The structurally-coupled systems ' will have an interlocked history of structural transformations, selecting each other's trajectories.' (Ibid.)



'A cognitive system is a system whose organization defines a domain of interactions in which it can act with relevance to the maintenance of itself, and the process of cognition is the actual (inductive) acting or behaving in this domain.' (Maturana & Varela, 1980, p. 13)

'Cognition' is the term conventionally used to denote the process by which a system discriminates among differences in its environment and potential states of that environment. The evidence for this 'cognition' is effectiveness of system behavior in response to the dynamics of its milieu. Today's dominant perspective on cognition is 'cognitivism' -- the idea that effective action is explainable in terms of manipulating abstract 'data', 'information', and 'knowledge'. This approach is best known from the 'Human Information Processing' (HIP) school of psychology, artificial intelligence (AI), and the 'cognitive sciences' lying at their intersection. During the last decade, there has been a growing realization that cognitivism is at best a limited way of analyzing humans and their interactivity (cf. Winograd & Flores, 1986).

Maturana and Varela attribute the capacity for discrimination to the organism's structure, but not as an internal manipulation of extrinsic 'information" or "signals', as the cognitivist viewpoint would have us believe:

'This would mean that such inputs or outputs are part of the definition of the system, as in the case of a computer or other machines that have been engineered. To do this is entirely reasonable when one has designed a machine whose central feature is the manner in which we interact with it. The nervous system (or the organism), however, has not been designed by anyone... (T)he nervous system does not 'pick up information' from the environment, as we often hear... The popular metaphor of calling the brain an 'information-processing device' is not only ambiguous but patently wrong.' (Maturana & Varela, 1987, p. 169)

A living system's organization circumscribes a domain of interactions within which activity relevant (and appropriate) to maintaining its autopoiesis is manifested. '[F]or every living system, its organization implies a prediction of a niche, and the niche thus predicted as a domain of classes of interaction constitutes its entire cognitive reality.' (Maturana & Varela, 1980, p. 11) In later writings, this circumscribed 'cognitive reality' is usually termed a cognitive domain -- '... all the interactions in which an autopoietic system can enter without loss of identity...' or, with regard to the system as an observer, '...the domain of all the descriptions which it can possibly make.' (Maturana & Varela, 1980, p. 136)



In colloquial discussions generally (and cognitivism specifically), interpersonal communication is typically treated as a 'piping' of 'information' among conversants. This view presumes 'information' is a quantum commodity, and it shifts the focus of observation from interactors to a presumed commerce in this commodity. Cognitivistic approaches conventionally treat this commerce as 'instructive interactions' (Maturana, 1978b) -- interactions in which the 'receiver' adopts a state determined by the state of the 'sender' as projected via the 'message'. This view of language concentrates on '...a denotative system of symbolic communication, consisting of words that denote entities regardless of the domain in which these entities may exist.'(Op.cit., p.50) Such an approach overlooks the fact that "Denotation ... requires agreement -- consensus for the specification of the denotant and the denoted." (Ibid.)

In analyzing actual communication, the prevailing approach is very problematical. Communication is of interest to the extent of what happens with or to the person 'receiving' it ('persons' in the case of reciprocal dialogue or one-to-many broadcasting). Because a quantum 'information' commodity is not defined with regard to the structure of the interactor(s), focusing on the 'message' blinds an observer (e.g., an enterprise analyst) to the actor and her activity during conversation. This leaves the analyst to wonder about cases in which apparently clear-cut 'messages' were not 'instructive' -- i.e., didn't induce the effect of their content.

Maturana views language as the archetypal illustration of a human consensual domain. Linguistic interaction is a venue for action, coupling the cognitive domains of two or more actors. This is reflected in Maturana's preference for discussing languaging (the act) as opposed to 'language' (a symbolic schema).

The primary function of linguistic interaction is therefore not conveyance of 'information quanta', but the mutual orientation of the conversants within the consensual domain realized by their interactivity. 'Communication' becomes a matter of mutual orientation -- primarily with respect to each other's behavior, and secondarily (only via the primary orientation) with respect to some subject. This is extremely important for delimiting the constraints on an observer's (e.g., an enterprise analysts') analysis of communicative interactions. In today's conventional (e.g., cognitivistic) approaches, such interaction is described as a semantic coupling -- a process by which each of the observed interactors computes the appropriate response state from some informative input from the other. Maturana warns that this is not warranted:

'(a) because the notion of information is valid only in the descriptive domain as an expression of the cognitive uncertainty of the observer, and does not represent any component actually operant ... and (b) because the changes of state of a [structurally] determined system, be it autopoietic or not, are determined by its structure, regardless of whether these changes of state are adequate or not for some purpose that the observer may consider applicable.' (Maturana, 1975, p. 322)

This moves linguistic interaction to a conceptual base whose elements apply to a much broader range of actors and interactions than symbolic data. The structural coupling of the participating organisms is the only operative element -- all other items treated in descriptions of linguistic behavior are secondary. How, then, can one account for the seemingly secure framework within which we ordinarily consider conversation to occur -- shared lexicons, objective meanings, and syntactic conventions? Maturana claims: (1) such a question is biased in its presumption that such a framework objectively exists, and (2) such regularities are imposed by an observer:

'If recursion is possible in a particular kind of behavior ... a closed generative domain of behavior is produced. ... What is peculiar about a language, however, is that this recursion takes place through the behavior of organisms in a consensual domain. In this context, the superficial syntactic structure or grammar of a given natural language can only be a description of the regularities in the concatenation of the elements of the consensual behavior. ...This superficial syntax can be any, because its determination is contingent on the history of consensual coupling ... (T)he 'universal grammar' of which linguists speak as the necessary set of underlying rules common to all human natural languages can refer only to the universality of the process of recursive structural coupling.' (Maturana, 1978b, p. 52)

The reclassification of communicational behavior from conceptual commerce to mutual orientation expands the range of behaviors we may consider as 'communicative'. The autopoietic view of language is not constrained to coded symbols for the manner in which interactors couple. 'The richness attained by a language ... depends necessarily both on the diversity of behaviors that can be generated and distinguished by the organisms that participate in the consensual domain.' (Op.cit., p. 51) By disengaging interaction from lexical reference and grammatical performance, the autopoietic model implicitly allows for all manner of non-verbal or extra-verbal signalling -- a scope more akin to semiotics than mainstream linguistics.

Copyright 1995 Randall Whitaker. This material may be freely copied and reused, provided the author and source are cited
Spacer Spacer Spacer Spacer Spacer Spacer Spacer Spacer Spacer Spacer Spacer Spacer Spacer Spacer Spacer Spacer Spacer Spacer Spacer Spacer