Professional Documents
Culture Documents
Pub - Autopoiesis in Organization Theory and Practice Ad PDF
Pub - Autopoiesis in Organization Theory and Practice Ad PDF
Previous Volumes:
RODRIGO MAGALHÃES
Kuwait-Maastricht Business School, Kuwait and
Technical University of Lisbon, Portugal
RON SANCHEZ
Copenhagen Business School, Denmark and
National University of Singapore, Singapore
No part of this book may be reproduced, stored in a retrieval system, transmitted in any
form or by any means electronic, mechanical, photocopying, recording or otherwise
without either the prior written permission of the publisher or a licence permitting
restricted copying issued in the UK by The Copyright Licensing Agency and in the USA
by The Copyright Clearance Center. No responsibility is accepted for the accuracy of
information contained in the text, illustrations or advertisements. The opinions expressed
in these chapters are not necessarily those of the Editor or the publisher.
ISBN: 978-1-84855-832-8
ISSN: 1877-6361 (Series)
Contents
PART I INTRODUCTION
INTRODUCTION
Chapter 1
We human beings are not rational animals; we are emotional, languaging animals that use the
operational coherences of language, through the constitution of rational systems, to explain and
justify our actions, while in the process and without realizing it, we blind ourselves about the
emotional grounding of all the rational domains that we bring forth. (Humberto Maturana,
1988, p. 787)
1 Introduction
This introductory chapter elaborates some of the key ideas which shaped the concept
of this book. The overriding idea is that autopoiesis theory has the potential to
provide a unifying framework for the study of organizational phenomena in the 21st
century. Although organization studies have recently had no shortage of new
paradigms and approaches — such as postmodernism, phenomenology, ethnometho-
dology, reflexivity, and critical theory — the field seems to be expanding in ways that
make it increasingly difficult to comprehend, especially for the uninitiated.
In the 1950s and 1960s, open systems theory, together with sociological systems
theory, was enormously influential in providing a coherent framework for the study
of organizations and their environments. These approaches were in important
respects motivated by ideals of order, stability, and predictability. So influential were
they that the paradigm they defined is still prevalent today. Although today’s
organizations and their environments are often characterized by transformation,
emergence, much unpredictability, and a strong emphasis on people, the systems
approach to understanding organizations is still not being conveyed in a coherent
manner, especially to students and managers. The reason for this, in our view, is the
For many people, the adoption and use of concepts from autopoiesis theory in
organizational analysis is uncontroversial. However, some organization scholars are
reluctant to adopt autopoiesis concepts or theory because of concerns as to whether
organizations really are autopoietic systems (Mingers, 2002, 2004). Thus, while some
authors choose to apply autopoiesis theory to organization studies following strict
ontological principles, others are less convinced theoretically and treat autopoiesis as
a metaphorical perspective. The controversy between these two approaches is
somewhat surprising given the freedom with which metaphors are used in much
organizational theorizing (Morgan, 1997).
The reluctance that many authors show in applying autopoiesis to social settings
may be partly due to the fact that both Maturana and Varela stated in their writings
that autopoiesis is not a social theory. From his personal contacts with Humberto
Maturana, Zeleny (2007) has added a further perspective, suggesting that the creators
of autopoiesis were so careful in pointing out that the theory should be left out of the
Autopoiesis Theory and Organization: An Overview 5
social domain because of the political climate of Chile in the 1970s. Zeleny suggests
that Maturana and Varela may have been apprehensive about the misuse by the
prevailing political powers of the mechanistic nature of their theoretical propositions
in the biological sciences in the social sphere.
Nevertheless, in the following, we provide an overview of the principal applications
of autopoiesis theory in organizational settings.
In the organizational world, there are forces which are informal, enduring, and hard
to change (e.g., cultural norms), and others which are formal, often ephemeral, and
more amenable to adoption (e.g., processes, procedures, and tasks). The latter are
inevitably influenced and shaped by the former. In organizational theory and
research, these two kinds of forces are usually treated separately, because it is often
very difficult to reconcile them, although from the point of the practitioner, this is
always disappointing. Autopoiesis theory, however, offers organizational theorists
and researchers new possibilities to address such disparate organizational phenomena
in a much more integrated fashion. Take the concepts of organization and structure,
for example. Within the autopoietic perspective, organization means necessary
relationships or network of rules that govern relations between system components
and that thereby define the system conceptually. Structure means the actual relations
between the components that integrate the system in practice and that satisfy the
constraints placed by the organization.
Using the tenets of autopoietic theory, Zeleny (2005) interprets organizations as
networks of interactions, reactions, and processes identified by their organization
(network of rules of coordination) and differentiated by their structure (specific
spatio-temporal manifestations of applying the rules of coordination under specific
conditions or contexts). Following these definitions, Zeleny argues that the only way
to make organizational change effective is to change the rules of behavior (i.e., the
organization) first, and then change processes, routines, and procedures (i.e., the
structure). He explains that it is the system of the rules of coordination, rather than
the processes themselves, that defines the nature of recurrent execution of coordinated
action (recurrence being the necessary condition for learning to occur). He states:
‘‘Organization drives the structure, structure follows organization, and the observer
imputes function’’ (ibid, p. 197).
Espejo, Schumann, Schwaninger, and Bilello (1996) adopt similar terminology, but
instead of organization they refer to an organization’s identity as the element that
defines any organization, explaining that it is the relationships between the
participants that create the distinct identity for the network or the group.
Organization is then defined as ‘‘a closed network of relationships with an identity
of its own’’ (ibid, p. 75). Like Zeleny, Espejo et al. (1996) also see the organization’s
structure as being the differentiating factor. While organizations may share the same
kind of identity, they are distinguished by their structures. People’s relationships form
routines, involving roles, procedures, and uses of resources that constitute stable
6 Rodrigo Magalhães and Ron Sanchez
forms of interaction. These allow the integrated use and operation of the
organization’s resources. The emergent routines and mechanisms of interaction then
constitute the organization’s structure. Hence, just like any autopoietic entity,
organizations as social phenomena are characterized by both an organization (or
identity) and a structure. The rules of interaction established by the organization and
the execution of the rules exhibited by the structure form a recursive bond.
The adoption of autopoietic notions of organization and structure by conventional
organization theory may create exciting new opportunities to establish theoretical and
practical links between the structurally determined or ‘‘engineered’’ parts of an
organization, such as its business processes, and the emergent properties arising from
the actions and interactions of human actors that jointly shape the organization’s
identity. Our understanding of the heterogeneous engineering (Law, 1987) of the
multitude of soft and hard aspects of social organizations can greatly benefit from an
elaboration of this dichotomy and the ways in which the two dichotomous parts
interact and influence each other. We return to this point below.
Goldspink and Kay (2004) suggest that autopoiesis also provides the basic concepts
for understanding the mechanics of sociality and therefore of organizations. They
state:
Humans exist in and through domains, which are the product of their structural coupling with an
environment. This environment is the world around them, including other humans, and exists
both physically and causally. As humans enter into reciprocal interaction over time there
emerges, as a consequence of structural coupling, a certain alignment of their behaviors,
including their linguistic behaviors. Hence we can refer to the resulting domain as a consensual
domain. This domain now forms the basic unit of social analysis, and exists in a causal sense but
not in a physical one. (ibid, p. 605)
Human beings are autopoietic, which means that as individuals we are all
operationally closed. To illustrate, we have all experienced occasions when no matter
what we say and explain to our dialoguing counterpart, he or she is unable to
comprehend our point of view. This situation can last for a few minutes, or hours, or
may endure for years and even lifetimes. Operational closure can be observed in our
daily interactions at work, in the shopping mall, and in the family. The only way to
overcome autopoietic closure is by building structural couplings. The nature and
degree of structural coupling that emerges when two or more individuals interact is a
8 Rodrigo Magalhães and Ron Sanchez
defining feature of the macro system of invisible rules and procedures that
characterize social institutions.
Organizational closure, however, should not be confused with the notions of
‘‘closed’’ and ‘‘open’’ systems from traditional systems theory. Maula (2006) argues
that openness and closure are not only simultaneous phenomena, but they also
necessitate each other. In other words, there are no environmentally ‘‘closed’’ systems.
An organizational closed system cannot be completely closed to its environment,
because it cannot be completely unresponsive to environmental signals and
perturbations. Organizational closed systems are therefore closed with respect to
their own organization and structure, but they may nevertheless maintain intense
interactions with the environment. Through recurrent environmental signals,
perturbations, and triggers, a system becomes coupled to its environment. Such
coupling is achieved through changes in the system’s structure, even while the
organization remains autonomous and closed (Zeleny, 2003).
Structural coupling is an essential concept in understanding the decentralized
networked organization. The concept of coupling provides the basis for under-
standing how an organization may be fragmented in time and space while retaining its
unity as a system. This autopoietic concept has a direct counterpart in the
organizational terminology created by Weick (1976) when he refers to ‘‘loose
coupling.’’ He asserts that organizations are loosely coupled systems and defines loose
coupling as ‘‘a situation in which elements are responsive but retain evidence of
separateness and identity’’ (Orton & Weick, 1990, p. 203). In autopoietic terminology,
‘‘responsiveness’’ refers to compensating actions in response to perturbations from
the environment. ‘‘Separateness and identity’’ refer to the maintenance of the network
of interactions that defines the organization of the system. Orton and Weick (1990)
further propose loose coupling in organizational design requires more modularity in
organizational design (Sanchez & Mahoney, 1996), a broader range of requisite
variety (Ashby, 1956), and greater behavioral discretion. They also suggest that an
organization’s compensatory mechanisms for loose coupling may include enhanced
leadership, focused attention, and shared values.
The implications of the notions of structural coupling and loose coupling for
organization theory are very significant. They underpin the representation of
organizations as composites for which words such as bricolage or assemblage are
increasingly being used as descriptors (Ciborra and Associates, 2000).
Von Krogh and Roos (1995b) made one of the most significant contributions to
integrate autopoiesis into management theory and research. In so doing, they
advanced an anticognitivist position in the organizational knowledge debate. They
reject the notion that knowledge is a given and that the task of organizational systems
is to represent it as accurately as possible. Instead, they argue that knowledge is
embodied in human beings and that representations of the world in the human mind
Autopoiesis Theory and Organization: An Overview 9
Imagine that you are about to enter an office that is new to you. Your experience (knowledge)
tells you to take an initial sweeping look in order to locate the reception desk, your assumed
point of entry into the inner circles of the office. Having located what you believe is the reception
desk (world) you take the first steps towards the desk. In doing this you get a glimpse of a
corridor on your right-hand side in which you see a door and on which you locate a name plate
(world). You recognize the name on the door to be the person you are supposed to visit
(knowledge).
The ideas that the world is brought forth in knowledge and that knowledge is not
abstract but is embodied in human action frame the discussion about individual
versus organizational knowledge (Sanchez, 2001; Sanchez & Heene, 2006). Von
Krogh and Roos (1995b) argue that the bridge between socialized and individualized
knowledge is achieved by means of language. Language is what allows action to be
coordinated in the organization, and such coordination is achieved through
organizational members making useful distinctions about the organization (an
important form of organizational learning). The first and broadest distinction is the
concept of ‘‘organization’’ itself. Linguistically, the organization has to be
distinguished from its environment. The emergence in social interactions of a new
entity, in this case the organization, presupposes a languaging capability. Organiza-
tional members conceive of the organization they are working for through language,
and from this very broad distinction (i.e., the organization from the environment),
finer distinctions can start to be made. For example, there will be linguistic
distinctions associated with the concept of ‘‘product’’ in a given organization. In this
way, an organization develops its own languaging process and resulting language that
conveys its own system of meaning. An organization’s language-enabled system of
meaning, in turn, develops its own autopoiesis.
‘‘Languaging’’ is the expression used by Maturana and Varela (1980, 1992) to
denote the act of using language. Given its dynamic nature, languaging fulfils a dual
but conflicting function. On one hand, because languaging contributes to creating a
unique identity for an organization (e.g., language is integral to its culture),
languaging can be instrumental in bringing about change. On the other hand,
language is important in maintaining the status quo and may thereby be a source of
resistance to change, given the self-referential nature of autopoietic systems. Hence,
‘‘to allow for rules and languaging that give way for effective action’’ (von Krogh &
Roos, 1995b, p. 101) is one of the main goals for and functions of socialized
organizational knowledge. Von Krogh and Roos (1995a) suggest that knowledge
development in organizations comes about through the innovative use of old and new
words and concepts — for example, through managerial efforts to shape language
development in an organization.
10 Rodrigo Magalhães and Ron Sanchez
One form of communication in organizations is the conversations that can take place
between two or more persons. When conversations happen and become recurrent
among the same group of people, a social network, group, or community is formed.
Conversations allow a structuration process (Giddens, 1984) to evolve, and once the
structure of the network is formed, conversations become organizationally
closed and self-referential. Metaphorically speaking, conversations have embedded
in them the genetic code of a social network, through the three elements of
structure — signification, domination, and legitimation (Giddens, 1984). The internal
dynamics, roles, and values of networks, groups, and communities develop through
conversations. Hence, for a newcomer to become part of a group — a behavioral
domain — he/she has to learn, through participation, the group’s genetic code and
his/her role that is implicit in it. In this way, the social individual becomes structurally
coupled to the social network.
Social membership means accepting the unwritten rules of a group and (thereby)
being accepted by the group.1 Without mutual acceptance on some basis, cooperation
and social action are not possible. Social boundaries, social norms, and emerging social
practices transcend the individual and remain even after individuals have departed.
Particular members may join or leave, but the social organization carries on. Moreover,
organizations are based on self-transcendence — the reaching out beyond one’s own
existence in order to create shared understandings with others. In empathizing with
colleagues or customers in the process of socialization, the boundaries between
individuals are diminished. In the process of committing to a group and becoming part
of the group, the individual transcends the boundaries of the self. In the process of
internalizing organizational knowledge, individuals cross the boundaries and enter the
domain of the group or an organization (Nonaka et al., 2001).
The notion of boundaries of social systems implies a complementary notion of
organizational contexts. Context can be understood as a situation in which
individuals, work teams, or an organizational unit exerts a significant influence on
internal and external interpersonal relationships. Kakabadse and Kakabadse (1999,
p. 7) assert that ‘‘the power of context is substantial, for context helps form the
attitudes and perspectives individuals hold about life, work, people, and organiza-
tion.’’ Viewed through the lens of autopoiesis theory, the notion of organizational
context can be seen in a new light. Maturana (1988) argues that emotions form the
background for the embodiment of all our knowledge and thus cannot be separated
from logical thought in everyday action. For Maturana, emotions are the ingredient
1
Maturana (1988) argues that decisions about acceptance and rejection by groups are likely to be emotional
rather than rational – i.e., emotions are the ingredient that makes social phenomena possible, through
mutual acceptance (love, in his terminology). Sanchez suggests (here), however, that this view may also
reflect Maturana’s Latin cultural context in Chile, and that other cultural contexts may emphasize more
rational bases for acceptance or rejection of members by social groups (e.g., trust among managers in a
network of frequently transacting firms, commitment to adhere to the norms of a professional group, and a
recognized common interest in a coordinated community response to an opportunity or threat).
Autopoiesis Theory and Organization: An Overview 11
that makes all social phenomena possible, through mutual acceptance. However, in
our western-style management we have evolved a paradigm that encourages the
separation of logic and emotion. One of the earlier voices to denounce this state of
affairs was Selznick (1957, p. 80): ‘‘The importance of values is affirmed but the
choice of goals and of character-defining methods is banished from the science of
administration.’’ However, this situation may be changing — for example, through
the emergence of the idea of karma capitalism (which we will revisit further on) as
exemplified by the notions of soft power and smart power put forward by Nye (2008).
Such movements suggest that there is a renewed perception of the importance of
intangible elements like attitudes, emotions, and values in the workplace. The
merging of the economic and emotional contexts of firms, for example, is at the heart
of the holistic representation of firms in new strategic management theory (Sanchez &
Heene, 2004).
James March (2007) suggests that the field of organization studies may be entering a
fourth ‘‘invasion’’ era characterized by the growing influence of information
technologies and biological advancements on social life and by the earth’s declining
ability to sustain the current conduct of the rapidly growing human species.2 We agree
with this broad assessment and suggest that four trends will be decisive in shaping the
organization of the future:
The earth’s declining capacity to sustain the current practices of the human species;
New kinds of capitalism, leading to the individualized corporation;
Technical and social networking as the basis for decentralized, autonomous
organizational forms;
A world fuelled by ubiquitous, real-time data and information.
2
The previous critical landmarks according to March (2007) were (1) the Second World War; (2) the social
and political protests of the 1960s and early 1970s; and (3) the collapse of the Soviet Empire and the
triumph of the markets.
12 Rodrigo Magalhães and Ron Sanchez
Organizational trends
New Kinds of Capitalism
Networking
New
epistemological
approaches
inspired on non-
linearity and
complexity
The networking The integration of the
nature of social and the
organizing and technological
organizations architectures
Ubiquitous, real-time
information
Figure 1: Major external pressures and key challenges of the organization of the
future.
form of capitalism for the very poor also reflects these trends. However, capitalism is
also changing from within the firm, often leading to dramatic changes in the
relationship between the firm and the individual. Such changes are captured in the
notion of The individualized corporation as proposed by Ghoshal and Bartlett (1998).
More than ever before in some firms, the individual worker is becoming the center of
management concerns. This trend is due to the ongoing shift in the economy from
traditional industries based on manual workers to new enterprises based on
knowledge workers who are now the crucial asset in many businesses. At the same
time, while many corporations can no longer guarantee employment, growing
numbers of knowledge workers no longer need or are even concerned about
guaranteed employment. As a result, the nature of the bond between the organization
and its employees is changing radically, and the notion of a ‘‘moral contract’’ between
a firm and its employees is beginning to replace legal contracts as the basis for
employment (Ghoshal & Bartlett, 1998).
The organizational world is also being transformed by phenomena that run
counter to the traditional command and control model of organization. Although the
idea of the ‘‘networked organization’’ has been a topic of discussion for a couple of
decades, we are now entering an era in which we can observe real decentralized,
autonomous, networked organizations on a global scale. What is important in this
development is that not only organizations as institutions are able to network with
other organizations, but people are now able to network person-to-person as never
before. Internet and mobile telecom technologies are enabling people to meet and to
coordinate their activities in ways that are profoundly affecting their lives, both
professional and private. Perhaps the best example of the positive potential of a
global, decentralized, autonomous, networked organization is the World Wide Web.
However, other examples such as Al-Qaeda, where individual networking capabilities
seem to play a predominant role in organizing, show that such developments are not
limited to corporate or high-tech domains. Both examples not only follow a
decentralized, networked form, but also lack any kind of conventional management
structure. In a similar manner, the Linux phenomenon has no formal structure,
employees, or budgets, and its product is free. Yet Linux is already posing a serious
threat to the largest software firms in the world (Hernes & Bakken, 2003).
Networking (individual and institutional) is of course intimately related to the
proliferation of information technology (IT) in human society. Ever faster enterprise
LANs, telephony over IP data networks (VOIP), mobile telephony, home networks,
and Internet access in automobiles, planes, and trains are all having a major impact
on an organization’s and an individual’s capability to transmit information. As
citizens of the world, we are increasingly surrounded by real-time or near real-time
data and information. When an ice sheet breaks loose in the Arctic, citizens of Africa
learn about it a few minutes later. If a car breaks down in Outer Mongolia, the
manufacturer’s assistance services in Europe will be alerted instantaneously. Online
DNA data is being used by international police forces to solve crimes in a fraction of
the time it took to solve similar crimes in the past.
An even greater effect of IT than the ubiquity of information is its ability to
represent large chunks of organizational life as information. Balanced scorecards,
14 Rodrigo Magalhães and Ron Sanchez
Autopoiesis provides a model of how phenomena (which we may now call social phenomena)
emerge from the complex (and non-linear) interplay between the heterogeneous (in having unique
ontogenies) agents (people) which make it up. Complexity then allows us to explain the resulting
dynamics by describing the generative processes that link empirical observation and causal
actuality. Social systems can be seen as a specific class of complex systems and it is autopoiesis
which clarifies the distinguishing characteristics of this class, in particular the linguistic/reflexive
character of social agents. (ibid, p. 615)
3.2.2 The Search for a New Organizational Paradigm. The situation in organiza-
tion science/studies today is that the overwhelming majority of textbooks used in
graduate and postgraduate courses convey linear, reductionist representations of
organizations that are typical of what we might call the old paradigm. An example is
the book by Scott and Davis (2007) which undertakes to bring together much of the
accumulated wisdom in rational, natural, and open systems views of organizations.
Although warning about the dangers of clinging too much to the past when trying to
16 Rodrigo Magalhães and Ron Sanchez
17
18 Rodrigo Magalhães and Ron Sanchez
One of the key characteristics of the networking perspective is the interactions that
organizational members design and develop in seeking to communicate with other
organizational members. Through interactions, organizational members perform
socially embedded (i.e., role-based) actions and build relationships at a variety of
levels (local, group, intergroup, organization, and interorganization). The relational
nature of organizational life and the conception of an organizational member as a
social actor are also features of actor-network theory (ANT), an important landmark
in contemporary organization theory. As tool for understanding crucial organiza-
tional concepts, the network concept can be explained as follows:
Autopoiesis Theory and Organization: An Overview 19
[the network] has the same relationship with the topic at hand as a perspective grid to a
traditional single point perspective painting: drawn first, the lines might allow one to project a
three-dimensional object onto a flat piece of linen; but they are not what is to be painted, only
what has allowed the painter to give the impression of depth before they are erased. (Latour,
2005, p. 131)
ANT can be seen as a systematic way to bring out the network infrastructure that is
usually omitted in ‘‘heroic’’ accounts of scientific and technological achievements
(Ryder, 2008). ANT views social change as an emergent process that is initiated and
guided by actors with specific interests and strategies and describes the progressive
constitution of a network in which both human and nonhuman actors assume
identities according to prevailing strategies of interaction. In this sense, organiza-
tional life is heterogeneously engineered, after an expression coined by Law (1987).
Instead of characterizing the technological world as a neat set of homogenously
engineered, cause-and-effect relationships, Law describes it as the result of the activity
of myriad dynamic networks, comprised of multiple actors possessing many different
attributes, interests, and goals.
Besides ANT, other approaches have been put forward that are consistent with the
networked view of organizational life. Another relevant example is social capital
(Burt, 1997; Nahapiet & Ghoshal, 1998; Adler & Kwon, 2002) which holds that the
source of the organizational capacity to create value is the interactions between
individuals or between firms, rather than the individuals themselves. According to this
view, knowledge about what products systems produce is becoming less important
than knowledge about how systems produce themselves — i.e., how systems renew
their own ability and capacity to produce (Zeleny, 2003). Given the considerable
degree of convergence between the networked approaches to organization and the
tenets of autopoietic theory, can autopoiesis become a more encompassing
organizational theory of networks?
a complex social object which results from the embedding of computer systems into an
organization (y) where it is not possible to separate the technical from the social factors given
the variety of human judgments and actions, influenced by cultural values, political interests and
participants’ particular definitions of their situations intervening in the implementation of such a
system.
20 Rodrigo Magalhães and Ron Sanchez
(1) Organizations are complex adaptive systems where efficient, effective, and
sustainable growth and development depends upon the constant production of
new internal knowledge.
(2) As in complex adaptive systems, the transformation of organizations often starts
from small innovations found at the fringes of the system’s central core of activity.
(3) Innovations and new knowledge are partly associated with the implementation
and management of IS/IT.
(4) IS/IT-related innovations and new knowledge creation are often found at the
fringes of the organization’s central core of activity.
(5) Organizations need to adopt new managerial theories and practices in order to
discover and benefit from the knowledge assets found at their fringes, including
new IS/IT-related knowledge (ibid, p. 225).
Autopoiesis Theory and Organization: An Overview 21
4 Conclusion
In spite of the growing interest in autopoiesis and autopoietic systems in
organizations over the last 10 or 15 years, such interest has not made its way into
the textbook domain. In the 1960s, open systems theory, a breakthrough in the
biological sciences (Von Bertalanffy, 1950), made its way into the organization
sciences through the seminal work of Katz & Kahn (1966) and has held a dominant
position ever since. Many of the tenets of open systems now need to be revisited, but
so far there has not been an alternative perspective as powerful or influential in
organization theory. Although autopoiesis has been heralded by many as a new
systems theory, it has not yet achieved the same kind of impact as open systems
thinking, in large part because there is no clear-cut agreement among organization
scholars regarding the role or the place of autopoiesis in organization science.
Nevertheless, there has been a considerable amount of literature on autopoiesis in
organization studies, a selective summary of which has been presented in this chapter.
While some authors have adhered to the qualified approach of Luhmann, others have
taken autopoiesis straight from the realm of the biological sciences to organization
studies, and have even combined it with other approaches. The result has been a
number of proposals, some cautioning observation and interpretation, others
supporting analysis and direct intervention, but none attempting to elaborate an
integrated or comprehensive approach. In this volume, we have set ourselves the
challenge of moving the possibilities for theoretical integration forward and of placing
autopoiesis alongside other mainstream approaches in organizational thinking.
To this end, five questions encapsulating the trends at the forefront of
contemporary organizational thinking were posed to the authors included in this
volume:
(1) Can autopoiesis provide the backdrop for a new organizational paradigm?
(2) Framed within the complexity paradigm, can autopoiesis provide the metalan-
guage for a new theory of organization and management?
22 Rodrigo Magalhães and Ron Sanchez
(3) What might be the role of autopoiesis in the turn toward a focus on practice and
transdisciplinarity in organizational thinking?
(4) How might autopoiesis theory lend further support to the views supporting the
networked nature of organizations and organizing?
(5) Given its holistic nature, can autopoiesis provide a suitable framework for the
integration of IT/IS into social organizations?
The invited authors responded in a variety of ways to these questions, and the
result is the present volume of original and peer-reviewed papers. The editors hope
that the ideas presented here will provide the basis for establishing autopoiesis as an
innovative intellectual lever for the study of organizations.
References
Adler, P. S., & Kwon, S. W. (2002). Social capital: Prospects for a new concept. Academy of
Management Review, 27(1), 17–41.
Argyris, C. (1977). Double loop learning in organizations. Harvard Business Review, Sept.–Oct.,
115–124.
Ashby, W. R. (1956). An introduction to cybernetics. London: Chapman and Hall.
Axelrod, R., & Cohen, M. D. (2000). Harnessing complexity: Organizational implications of a
scientific frontier. New York, NY: Basic Books.
Barnard, C. I. (68). The functions of the executive. Cambridge, MA: Harvard University Press.
Barney, J. B. (1991). Firm resources and sustained competitive advantage. Journal of
Management, 17, 99–120.
Blau, P. M. (1970). A formal thcory of differentiation in organizations. American Sociological
Review, 35, 201–218.
Boulding, K. E. (1956). General systems theory: The skeleton of science. Management Science,
2, 197–208.
Burt, R. (1997). The contingent value of social capital. Administrative Science Quarterly, 42(2),
339–365.
Ciborra and Associates. (2000). From control to drift. Oxford: Oxford University Press.
Conner, K. (1991). A historical comparison of resource-based theory and five schools of
thought within industrial organization economics. Journal of Management, 17, 121–154.
Conner, K., & Prahalad, C. K. (1996). A resource-based theory of the firm: Knowledge versus
opportunism. Organization Science, 7(5), 477–501.
Dimaggio, P., & Powell, W. W. (1983). The iron cage revisited: Institutional isomorphism and
collective rationality in organizational fields. American Sociological Review, 48, 147–160.
Espejo, R., Schumann, W., Schwaninger, M., & Bilello, U. (1996). Organizational transforma-
tion and learning: A cybernetic approach to management. Chichester: Wiley.
Fayol, H. (1949). General and industrial management. London: Pitman.
Ghoshal, S., & Bartlett, C. A. (1998). The individualized corporation: A fundamentally new
approach to management. London: Heinemann.
Gibbons, M., Limoges, C., Nowotny, H., Schwartzman, S., Scott, P., & Trow, M. (Eds).
(1994). The new production of knowledge: The dynamics of science and research in
contemporary societies. London: Sage Publications.
Autopoiesis Theory and Organization: An Overview 23
Senge, P. (1990). The fifth discipline: The art and practice of the learning organization. London:
Century Business.
Simon, H. (1997). Administrative behavior. New York: The Free Press.
Symons, V. J. (1991). Impacts of information systems: Four perspectives. Information and
Software Technology, 33(3), 181–190.
Taylor, F. W. (1911). The principles of scientific management. New York: Harper & Brothers.
Trisoglio, A. (1995). Managing complexity. Unpublished working paper presented at the
Strategy and Complexity Seminar, London School of Economics, London.
Varela, F. J. (1984). Two principles of self-organization. In: H. Ulrich & G. J. B. Probst (Eds),
Self organization and management of social systems. New York, NY: Springer Verlag.
Varela, F. J., Thompson, E., & Rosch, E. (1991). The embodied mind. Cambridge, MA: MIT
Press.
Von Bertalanffy, L. (1950). The theory of open systems in physics and biology. In: F. E. Emery
(Ed.), Systems thinking. Harmondsworth: Penguin.
Von Krogh, G., & Roos, J. (1995a). Organizational epistemology. Basingstoke: Macmillan.
Von Krogh, G., & Roos, J. (1995b). Conversation management. European Management
Journal, 13(4), 390–394.
Weber, M. (1947). The theory of social economic organizations. London: Oxford University
Press.
Weick, K. E. (1976). Educational organizations as loosely coupled systems. Administrative
Science Quarterly, 21, 1–19.
Weick, K. E. (1979). The social psychology of organizing. Reading, MA: Addison-Wesley.
Wernerfelt, B. (1984). A resource-based view of the firm. Strategic Management Journal, 5,
171–180.
Wernerfelt, B. (1995). The resource-based view of the firm: Ten years after. Strategic
Management Journal, 16, 171–174.
Whittington, R. (2006). Completing the practice turn in strategy research. Organization Studies,
27(5), 613–634.
Whyte, W. F. (1959). Man and organization. Homewood, IL: Richard D. Irwin.
Williamson, O. E. (1975). Markets and hierarchies: Analysis and antitrust implications. New
York: Free Press.
Woodward, J. (1965). Industrial organization: Theory and practice. New York: Oxford
University Press.
Zeleny, M. (2003). Knowledge and self-production processes in social systems. In: F. Parra-
Luna (Ed.), Systems sciences and cybernetics, encyclopedia of life support systems (EOLSS),
UNESCO. Oxford: Eolss Publishers.
Zeleny, M. (2005). Human systems management: Integrating knowledge, management and
systems. Singapore: World Scientific Publishing.
Zeleny, M. (2007). Personal conversation during the 9th International Conference on
Enterprise Information Systems, June 12–16, Funchal, Madeira.
PART II
1 Introduction
‘‘y hmmm, let’s have a quick look y he opened the suitcase y holy Jesus!, he said y legs!’’
– W.S. Burroughs from Spare Ass Annie and Other Tales
Maturana and Varela’s distinctive writing style compounds the problem. As John
Mingers (1995, p. 2) notes, common words such as ‘‘organization,’’ ‘‘structure,’’ and
‘‘conversation’’ are employed in very uncommon ways, and there is a predilection to
communicate ‘‘a density of ideas y expressed with almost mathematical sparseness.’’
Once one becomes accustomed to this style of writing it can be highly effective.
Initially though, it makes for hard going, and this does not help in getting the message
across.
Of the various secondary accounts of autopoiesis that are available, Mingers’
(1995) text has almost certainly had the greatest impact in terms of removing much of
the mystique surrounding the area. However, even this work has its limitations.
Mingers acknowledges that the breadth and coherence of autopoietic theory is one
of its great strengths. Yet his main project was to delve into various controversies
surrounding specific aspects of the theory, such as the question of applying
autopoiesis to social systems and organizations, and to evaluate applications within
various areas of professional practice such as the law, therapeutic practice,
information systems, and artificial intelligence. Mingers did not specifically set out
to show how the various elements ‘‘fit together’’ in a coherent whole. Hence, gaining a
sense of this coherence from this particular source is by no means easy; it requires a
degree of cross-referencing and synthesizing that some readers will not have the time
or perhaps the inclination to do.
Against this background, the following chapter presents an account of what I
currently take to be ‘‘autopoietic theory.’’ It is a ‘‘high level’’ view, one that combines
the various biological aspects with the derivative human and phenomenological
aspects. Arguably the latter present the greatest opportunities for connecting
autopoietic theory with issues that are of interest to organization scholars. In order
to draw attention to key concepts, these are highlighted, italicized, and presented in
uppercase. Thereafter, the chapter presents just enough detail to enable organiza-
tional scholars to gain a sense of what the area is all about and, when read in
conjunction with the other contributions to this volume, to think about how
autopoiesis and autopoietic theory does and might further contribute.
Extrapolating from the basic proposition that biology and language generate the
phenomenon of observing, Maturana (see, e.g., Maturana, 1988, p. 63) claims that
there are two ‘‘PHENOMENAL DOMAINS’’ involved. Propositions relating to the
first of these originate in experimental work on color vision carried out in the early
1960s (see Maturana, Lettvin, McCulloch, & Pitts, 1960; Maturana, Uribe, & Frenk,
1968; Maturana, 1970). Subsequently, this led Maturana and Varela to reject the
prevailing idea that the nervous system is open to information from the environment,
and that human cognition mimics the symbol-based operation of computers.
In these experiments (and as an example), it was found to be possible to identify
direct correlations between the configuration of external wavelengths of light and
activity on the retina of the eye, and between activity in the optic nerve and subjects’
verbal descriptions of color. It was not possible, however, to establish such a link
between cellular activity on the retina and activity in the optic nerve. This leads to the
proposition that no external stimulus acting on the nervous system can determine an
organism’s experience of it. Experience corresponds to neuronal activity; it does not
correspond to external perturbations. Thus, whereas popular opinion takes the view
that the nervous system is open to environmental inputs, Maturana claims that it is
closed, autonomous, and circular. He further claims that the sensory and effector
surfaces of the nervous system have a dual character. They operate as elements of the
organism and as elements of the nervous system. Acting as components of the
organism they operate in the interactions of the organism in its medium, but acting as
components of the nervous system they operate in its closed dynamics. This means
that the organism as a systemic totality interacts with the medium, but the nervous
system as a component part does not.
An important consequence of this is the understanding that there is no ‘‘outside’’
to the internal components of the nervous system; there are only internal correlations
of neuronal activity. Notions such as ‘‘inside’’ and ‘‘outside’’ require the existence of
an observer who can see both and explain changes in one in terms of the other.
The nervous system, says Maturana, is an example of a ‘‘COMPLEX UNITY,’’
i.e., an ‘‘entity’’ or ‘‘system’’ that an observer decomposes to focus not just on it, but
also on its component parts. All such entities, he suggests, are subject to the principle
of ‘‘STRUCTURE DETERMINISM.’’ That is, ‘‘y everything that happens in
them happens as a structural change determined y either in the course of their
own internal dynamics or triggered but not specified by the circumstances of their
interactions’’ (Maturana, 1990, p. 13).
32 John Brocklesby
Logically, the circularity and closure of the nervous system means that it cannot
work with representations. Cognition, therefore, cannot be taken to involve the brain
processing symbols that stand for external elements, manipulating representations
and then computing a response that is adequate in the light of prevailing
circumstances. On this account, although we frequently speak as if we know about
the ‘‘real world’’ and can accurately perceive the ‘‘things’’ in it, this is a biological
impossibility. Instead of characterizing an object (e.g., ‘‘that tree is green’’) according
to what are thought to be its intrinsic characteristics, we do so on the basis of ‘‘what
happens to us,’’ through experiences, as the object is ‘‘brought forth.’’ This is the basis
for Maturana’s claim that human beings do not exist in the physical space; instead
that the physical space exists through human beings.
Extending the key notion of circularity, Maturana and Varela sought to relitigate
the age-old question of what is meant by the term ‘‘living’’ (Maturana & Varela,
1980). Eventually this led to the claim that all living systems are autopoietic or self-
producing, i.e., they are: ‘‘networks of molecular production such that the molecules
produced, through their interaction, generate the network that produced them and
specify its extension’’ (Maturana, 1993).
Life then is constituted through a dynamic, through a particular closed and circular
‘‘manner of relating’’ of molecules. The term ‘‘AUTOPOIESIS’’ describes what
Maturana and Varela describe as the ‘‘ORGANIZATION’’ of living systems. In other
words it describes that which is invariant and common to all living systems. When
used in this context, ‘‘organization’’ refers to the nature of relationships between
components of an entity. As long as these relationships are conserved, the entity
maintains its class identity. The ‘‘STRUCTURE’’ of the entity refers to the tangible
manifestation of these relationships, i.e., it describes how they appear in phenomena
that are perceived to belong to that class identity. In addition whereas class identity
is conserved, the concrete manifestation of this can change over time. Thus, when the
term autopoiesis is used in the molecular domain, to refer to the ‘‘organization of the
living,’’ its structural embodiment across the full range of biological phenomena that we
refer to as ‘‘living’’ is subject to infinite diversity and constant change over time.
At first sight, the idea that living systems are structure-determined, that they have
an autonomous autopoietic organization, and, that the nervous system — if there is
one — is closed to information from the outside world presents serious difficulties in
accounting for the manifest adaptability of most living systems.
Conventionally, the environment is seen as the primary source of adaptation and
change. Thus, in human beings, adaptation is taken to arise as a result of the nervous
system’s ability to construct and manipulate environmental representations and,
on the basis of these, to act with intent. On this ‘‘COGNITIVIST’’ view, adaptation is
very easy to explain. In contrast, as we have already said, autopoietic theory claims
that the nervous system is not open to external information; indeed it regards this as
biologically impossible. So, if adaptation is not about building representations of the
‘‘outside world’’ and then acting on these, what mechanism is involved?
In answering this question, it is helpful to look at what autopoietic theory has to
say about the main sources of change in living systems. Essentially there are three of
these. Firstly, there is change that arises conditional upon the flow of matter through
Outlining the Terrain of Autopoietic Theory 33
the system (e.g., food, water, air quality, etc.). Secondly, there is change that results
from internal dynamics such as metabolic processes and aging. Thirdly — and in the
context of this discussion — the most important source of change occurs through a
process known as ‘‘STRUCTURAL COUPLING.’’ Maturana and Varela employ this
term to refer to situations in which it is possible to observe some sort of ‘‘fit’’ between
the living system and the medium. When there are recurrent interactions between an
organism and what an observer would regard as its environment, or between one
organism and another, structure-determined changes occur in both. In other words,
the two structures change congruently, each one according to its own structure
determinism. Through this process, the structure of the organism, at any point in
time, contains a record of previous interactions.
As long as the system survives, i.e., as long as there is conservation of autopoiesis
and adaptation, external interactions trigger structural change. Structural coupling,
then, is a key mechanism for change in a living system during its lifetime. In human
beings, it is also the basis of learning.
Whereas cognitivism tends to emphasize the response of the system to its
environment, and posits a primarily one-way information flow, the idea of structural
coupling implies a much more complex interdependent relationship. It says that when
one system perturbs the other, the structure of the perturbed system can change.
However, this is not adaptation as cognitivism construes it. Where a change does
occur in response to an environmental perturbation, it is because the particular
change that occurs is a pre-existing feature, or one possible state, of the system’s
structure. The perturbation only triggers it.
Turning now to cognition, structural coupling generated by the demands of
autopoiesis plays the role that conventional wisdom attributes to having a representa-
tion of the world. Traditionally, as I have said, we have come to regard cognition as
a process that has the brain manipulating representations of the external world.
For organisms that possess a nervous system, this sounds eminently plausible.
But what about organisms that are not endowed with a nervous system? How is it that
they are often just as well adapted to their medium? And does the absence of a
nervous system mean that these organisms are not ‘‘cognitive’’ beings?
Of course it is possible to answer the first of these questions using cognitivist logic.
Thus, an observer might regard physical movement in a single cell unity such as an
amoeba — as it surrounds a source of nourishment in its medium — in terms of the
amoeba having somehow ‘‘perceived’’ its environment and having ‘‘computed’’ an
appropriate response (see Maturana & Varela, 1987, p. 147). Yet this is unsatisfactory
since these functions imply the presence of a nervous system and the ability to develop
representations which the amoeba clearly cannot do.
Despite this, under normal circumstances, the amoeba’s behavior is adequate in
its domain of existence; it is adapted to its medium. One could even say that it is
a ‘‘cognitive being’’ that has the capacity ‘‘to know’’ albeit not in a conventional
humanistic sense.
On this account, cognition is inextricably linked to structural coupling which in
turn is linked to biology. The physical movement of the amoeba does not involve
perception. Instead, there is a process in which changes in the chemical composition
34 John Brocklesby
of the medium trigger changes at the sensory surface of the amoeba. This sets up an
internal dynamic that results in the amoeba altering its position relative to whatever
in the medium is triggering change.
It is this logic that leads to the conclusion that, in its most basic form,
cognition is not a mentalistic phenomenon. Rather, it is ‘‘effective action’’
(‘‘COGNITION ¼ EFFECTIVE ACTION’’) in a defined domain of existence
(see Maturana & Varela, 1987, p. 29). This process involves the whole organism
and is not limited to what might happen in the nervous system or brain if these exist.
If an observer sees a unity behaving adequately in some defined domain then we can
surmise that it ‘‘knows’’ relative to whatever criteria the observer applies in making
the necessary judgment. In this sense cognition equates to the whole process of living.
The abstract thinking that goes on in human beings is but a special case of what is
a much broader phenomenon (see Varela, Thompson, & Rosch, 1991 for a detailed
extension of this argument).
In human beings, this linking of cognition with structural coupling provides a
distinctive take on the increasingly popular idea that cognition is an integral part
of our normal everyday mindful and unmindful activity. According to this line of
thought, action that looks like adaptation to an observer, and which the representa-
tionalist perspective would explain in mentalistic terms, is merely the system operating
in a relationship of structural coupling with a medium.
The distinctive contribution of autopoietic theory then is to allege that cognition
has to do with the process of living, which, in turn, is inextricably linked to structural
coupling. This turns on its head reductionist conceptions of cognition that equate,
for example, the mind with the brain and/or limit it to conscious thought. Instead,
it asserts that the brain is merely one of many structures through which the mind
operates.
Such then are the basic distinctions that pertain to the first of the two domains that
are central to Maturana’s systemic account of observing. To summarize, we can say
that living systems are self-producing systems of molecular production which operate
in a dynamic structural coupling with a medium. In the case of human beings,
the basic biological character of the nervous system is critically important since it
makes observing possible, and it has implications for the what and the how of
observing in specific instances. However, it is in the second phenomenological
domain, that of relations and interactions with a medium, where observing, and
behavior more generally, takes place.
these structurally coupled systems that the key mechanism that allows observing to
take place occurs. This process is referred to as ‘‘LANGUAGING.’’
The idea of languaging describes a process in which the fundamental aspect is
behavior. It is not, as conventional wisdom would have it, the speech act itself.
In order to explain this, it is necessary to revisit the concept of structural coupling.
According to Maturana when we observe two or more structurally coupled entities
‘‘in language’’ with each other, what we see is a behavioral process. Using his
terminology we see an initial ‘‘coordination of action’’ between the two entities,
followed by a further coordination of the two actions. Consider the following: a
parent in the company of strangers looks disapprovingly at his/her misbehaving child;
the child immediately ceases to misbehave. Alternatively, consider the owner of a dog
whose whistle attracts the attention of the dog that brings it ‘‘to heel.’’ In both cases,
there is an initial coordination of action: the child’s gaze falls upon the parent and
the dog’s falls upon the owner. Next, both the child and the dog do something on the
consequences of what is done. In both cases, there is a further coordination of the
now already-coordinated actions.
This leads to the basic contention that the minimum operation that is involved
in language is a ‘‘COORDINATION OF A COORDINATION OF ACTION.’’
On this definition it follows that a rudimentary form of languaging is possible
even in organisms that do not possess a nervous system, and/or in those with
a nervous system, but which are incapable of abstract thought in the manner of
human beings. Much learned animal behavior such as the courtship, nest building,
and chick rearing behavior of birds, or the ‘‘dance’’ of bees, is of this basic type
(see Mingers, 1995, p. 78). However, to the extent that there is no abstract thought
involved in these activities, and in order to differentiate it from more complex
forms of languaging, Maturana refers to this as ‘‘LINGUISTIC CONSENSUAL
BEHAVIOR.’’
Where there is an advanced nervous system, the languaging potential of the
organism is increased dramatically. In simple terms, the intrusion of a nervous system
severs the direct link between sensory and motor surfaces of the organism. That
having occurred, although the sensory surface can and does perturb the nervous
system, the nervous system begins to interact with itself. Activity within the nervous
system thus becomes the object of further activity, which becomes — ad infinitum —
the object of further activity and so on. This process (see Mingers, 1995, pp. 73–79,
for a fuller description) provides the basis for a massive expansion in the cognitive
capability of the nervous system that, in human beings, culminates in abstract
thought.
Because we human beings have such a capacity, and because we live our lives
completely immersed in the full complexity of language, it is hard for us to see that, in
its minimum form, languaging involves such a simple act as a coordination of an
already-coordinated set of actions. Thus, we tend to describe language in semantic
terms, i.e., involving a transmission of information from one organism to another
that embodies some meaning. However, when — in a relationship of structural
coupling — one organism coordinates its actions with another and there is a
further coordination, then we have the basic dynamic that culminates in languaging.
36 John Brocklesby
The precise nature of how these two domains interact is not often acknowledged, or
I suspect, fully understood. Key to this is Maturana’s insistence that phenomena must
be understood in terms that pertain to it and not in terms that pertain to some other
component or domain. This is critically important when it comes to understanding
observing and in understanding human behavior more generally.
Observing, as we have seen, happens through the mechanism of language.
Logically then, since language is a relational phenomenon it cannot be taken to be
a process through which the brain manipulates symbols, nor is it reducible to the
operation of the nervous system. The same can be said about observing activities that
take place ‘‘in language,’’ for example thinking and explaining. These need to be
explained in relational terms and not biologically.
From this we can further deduce that since ‘‘objects’’ arise in language, notions
such as ‘‘self,’’ ‘‘the mind,’’ ‘‘personality,’’ etc. are also relational. As such they arise
in, and vary according to the relational circumstances of the moment. They are not
‘‘inscribed’’ in the bodyhood of the individual concerned. Remove the relational
context and to all intents and purposes they do not exist. All that is left is a
dehumanized biological entity.
On this logic ‘‘behavior’’ too is something that arises out of the interaction between
the biological entity and the medium. Consider, for example, the nature of the
‘‘behavior’’ or ‘‘act’’ that results from an anatomically generated movement in a
particular direction of someone’s arm. This depends wholly upon the medium with
which the arm interacts and the prevailing relational circumstances. Identical
anatomical movements can generate very different ‘‘behaviors’’ depending upon the
physical and social context of its interaction with the medium. Just as a nervous
system is necessary for speech acts to take place but language is not explainable
simply through reference to the nervous system, so too anatomy and physiology is
necessary for ‘‘behavior’’ to occur, but it is the interaction with the medium that
makes a behavioral act ‘‘what it is.’’
This illustrates one of the key features of autopoietic theory. Although the
biological and relational domains are separate and independent phenomenal
domains, they are nonetheless inextricably intertwined as components of a larger
systemic whole.
38 John Brocklesby
It follows then that when someone has any sort of relational encounter — when
they reflect on an issue (in isolation or with others), when they participate in a
conversation, they do so with a built-in predisposition to act in a particular way.
The biological structure of the system contains a record of past interactions, and the
physiological and emotional predispositions of the moment delimit the range of
possibilities. However, what happens during that relational encounter depends upon
what happens in the relational context. As a conversation proceeds, as its flow of
languaging and emotioning unfolds, people see and do things differently; as someone
reflects upon an issue their biological processes and mood impact upon what they do.
Biology then impacts upon but does not determine what happens in the relational
domain. At the same time relational encounters impact upon but do not directly
imprint themselves upon biology. At one extreme, an acrimonious exchange between
two people may increase the blood pressure of the people concerned. Longer term,
if the exchanges are repeated it could trigger cardiovascular disease; one of the
participants might even suffer a heart attack and die. At the other extreme, using
various meditation techniques and through processes of ‘‘self-talk,’’ someone might
learn how to reduce muscular and nervous tension to good long-term effect. In both
there is a structural transformation of the system in line with changes in relational
circumstances and in the medium itself. Indeed, this is the basis for evolutionary
development and learning. However, such changes are always subject to the system’s
structure determinism.
2.4 From Biology and Social Relations to Phenomenology and the Derivative
Epistemology
The issue here is how the epistemology of autopoietic theory, which again is often
handled as a separate discrete topic or aspect, arises out of, and relates, the biological
and social components of the work as a whole. Fundamentally, for both Maturana
and Varela, ‘‘knowing’’ is a biological issue since, as we have seen, cognition is
inextricably tied to the process of living and structural coupling between the
organism and the medium. Varela et al.’s (1991) concept of ‘‘ENACTION’’ or
‘‘EMBODIED COGNITION’’ further develops this basic idea. Knowledge,
they submit, is constituted in our actions. As an individual confronts new situations
various experiences are gained through thinking, sensing, and moving. This means
that the way we experience (and ‘‘bring forth’’) the world is very much an active
construction involving the whole body. Effective action depends upon having a body
with various sensorimotor and orienting capacities that allow an agent to act,
perceive, and sense in distinctive ways.
Outlining the Terrain of Autopoietic Theory 39
The first and most fundamental point then is that cognition is an integral part of
normal everyday activity and extends beyond the brain to encompass the whole body.
Cognition, as Mingers (1995) rightly puts it, ‘‘is not detached cogitation, but situated
practical action.’’
However, epistemology refers to the grounding and underpinnings of knowledge
that, in human beings, emanate from more conventional language-based processes of
understanding and explaining the world. For Maturana, what human beings ‘‘know’’
about their worlds depends, moment by moment, upon the precise nature of whatever
‘‘EXPLANATORY DOMAIN’’ is operational for that observer. An explanatory
domain is an explicit or, more commonly perhaps, an implicit frame of reference that
is defined at the very moment in which an observer distinguishes a particular area of
experience, and then explains it. Such domains are constituted through the particular
criteria that the observer applies in his/her listening in accepting a generative
statement relevant to the area of experience as valid.
Because there are as many domains of explanation as there are generative processes
and validity criteria, it follows that there is a ‘‘MULTIVERSE’’ of equally legitimate
realities. Because the nervous system is closed and circular, and because we are
subject to the principle of structure determinism we human beings cannot know in a
transcendental sense, and there is therefore no single ‘‘universe.’’
Maturana describes action undertaken in the awareness that an observer only
explains experiences and not some independent reality, as the observer placing
‘‘OBJECTIVITY IN PARENTHESES.’’ This ‘‘explanatory path’’ invokes a
particular dynamic of interpersonal relations. Statements made in the name of
objectivity and ‘‘truth,’’ i.e., in the explanatory path ‘‘OBJECTIVITY WITHOUT
PARENTHESES,’’ can easily imply a demand for obedience. Such cognitive state-
ments, says Maturana, ‘‘y are implicit claims of a privileged access to an objective
independent reality and are, hence, demands for obedience’’ (Maturana, 1988, p. 61).
In contrast, if a speaker gives up on the idea of an independent reality, he or she
must take responsibility for statements and present these ‘‘y as invitations to enter in
the same domain of reality as the speaker’’ (Maturana, 1988, p. 62), where there is full
disclosure of the basic premises and coherences that give the statement validity.
Finally, Maturana’s notion of ‘‘CONVERSATION’’ is central in further under-
standing the linkage between biological and social processes, and people’s
explanations and actions. When used in this context, ‘‘conversation’’ refers to a
dynamic braiding of ‘‘languaging,’’ which we have already discussed, and what he
refers to as ‘‘EMOTIONING.’’ Maturana describes the latter as ‘‘bodily dispositions
for actions’’ (Maturana, 1988, p. 42).
Importantly, as conversations flow through an interweaving of distinctions and
emotions, experiences and explanations alter. While the structure of the observer
imposes limits on what is and what is not possible, it is the nature of the conversation
and its flow that determines the outcome.
As people shift from one emotional state to another, changes take place in
what they will and will not do. People behave differently, they see differently, and,
importantly, they describe and interpret things differently according to the prevailing
emotion. Moreover, languaging and emotioning are braided, each process affecting
40 John Brocklesby
the other. This can be seen in everyday conversations where the specific distinctions
that people use invoke an emotional response and where distinctions used reflect the
emotion of the moment.
Maturana uses the term ‘‘CONSENSUAL DOMAIN’’ to describe these networks
of structural coupling that are the site for conversations. In these contexts people
learn their emotioning and their languaging with people to whom they are structurally
coupled, and, through recurrent interactions, structural patterns become conserved.
The picture is further complicated on account of the bodyhood of the observer
being at the intersection of not one, but many different conversations within and
across different structural couplings. Each one of these has its own braided flow of
distinctions and emotions that have been learned through recurrent interactions over
time and which alter subject to the dynamic flow of the conversation. This means that
just as thoughts and descriptions within a single conversation are subject to change
depending upon the flow of the conversation, they can also alter as the observer shifts
from one conversation to another.
3 Conclusion
In this chapter, I have attempted to provide a broad account of the developed form
of autopoietic theory as I currently see it. This account has sought to describe in
relatively straightforward terms the various components and aspects of autopoietic
theory and demonstrate how these synthesize into a highly distinctive way of thinking
that has much to offer scholars from other fields of inquiry. As I noted at the outset,
autopoietic theory has already made some in-roads within the broad territory of
organization studies. The aim now is to take this work to another level, to generate
more interest and clarity in anticipation that autopoietic theory can play a much
stronger role in organization studies than it has to date.
References
Maturana, H. (1970). Biology of cognition. In: H. Maturana & F. Varela (Eds), Autopoiesis and
cognition. Dordrecht: Reidel.
Maturana, H. (1988). Reality: The search for objectivity or the quest for a compelling
argument. Irish Journal of Psychology, 9, 25–82.
Maturana, H. (1990). Science and daily life: The ontology of scientific explanations. In:
W. Krohn, G. Kuppers & H. Nowotny (Eds), Selforganization: Portrait of a scientific
revolution (pp. 12–35). Dordrecht: Kluwer.
Maturana, H. (1991). Response to Jim Birch. Journal of Family Therapy, 13, 375–393.
Maturana, H. (1993). Videotaped seminar on language and cognition. Melbourne: St. Kilda.
Maturana, H. (1997). Metadesign. Available at: http://www.inteco.cl/articulos/metadesign.htm
Maturana, H. (2003). Autopoiesis, structural coupling and cognition: A history of these and
other notions in the biology of cognition. Available at: http://www.matriztica.org/1290/
article-28335.html
Outlining the Terrain of Autopoietic Theory 41
Maturana, H., Lettvin, J., McCulloch, S., & Pitts, W. (1960). Anatomy and physiology of
vision in the frog. The Journal of General Physiology, 43, 129–175.
Maturana, H., Uribe, G., & Frenk, S. (1968). A biological theory of relativistic color coding in
the primate retina. Archives of Biological and Medical Experiments, 1, 1–30.
Maturana, H., & Varela, F. (1980). Autopoiesis and cognition: The realization of the living.
Dordrecht: Reidel.
Maturana, H., & Varela, F. (1987). The tree of knowledge — The biological roots of human
understanding. Boston, MA: Shambala.
Mingers, J. (1995). Self-producing systems — Implications and applications of autopoiesis.
New York, NY, and London: Plenum Press.
Varela, F., Thompson, E., & Rosch, E. (1991). The embodied mind — Cognitive science and
human experience. Cambridge, MA: MIT Press.
Whitaker, R. (1996). The Observer Web: The Internet Nexus for Autopoiesis and Enaction,
available at: http://www.enolagaia.com/AT.html
Chapter 3
1 Introduction
Modern organic metaphors for society have run parallel to the very idea of sociology
as a science, starting with Comte and Spencer’s use of the term ‘‘social organism’’
(Comte, 1830–42; Spencer, 1897). These metaphors provide a self-renewing source of
debate, analogies, and disanalogies. Processes of social regulation, conservation,
growth, and reproduction provoke an irresistible epistemic resonance and make us
lose little time in offering explanations resembling those of biological regulation,
conservation, growth, and reproduction. The phenomenon has not been restricted to
metaphor-hungry social scientists: the final chapter of W. B. Cannon’s The wisdom
of the body (1932) is called ‘‘Relations of biological and social homeostasis.’’
Attempts to apply a modern theory of living organisms — the theory of autopoiesis
(Maturana & Varela, 1980) — to social systems are but the latest installment in this
saga. Despite the appeal of the organic metaphor, there are good reasons to remain
skeptical of these parallels. ‘‘Because every man is a biped, fifty men are not a
centipede,’’ says G. K. Chesterton (1910) ironically in his essay against the medical
fallacy. Doctors may disagree on the diagnosis of an illness, he says, but they know
what is the state they are trying to restore: that of a healthy organism (implying,
admittedly, a rather unproblematic concept of health). In social systems, a ‘‘social
illness’’ confronts us with precisely the opposite situation: the disagreement is about
what the healthy state should be.
In asking about the health of an organism, we ask about its norms, about the logic
of integration by which its components are reciprocally means and purposes of a
unity, and about how this unity enters into relations with its milieu. The question is:
Is such logic applicable to social systems? Georges Canguilhem (who already in 1951
used the term autopoetique to define the character of living organisms, see
Canguilhem, 1965) contrasts biological history with the history of societies:
The biological evolution of organisms has proceeded by means of stricter integration of organs
and functions for contact with the environment and by means of a more autonomous
internalization of the conditions of existence of the organism’s components and of what Claude
Bernard calls the ‘‘internal environment.’’ Whereas the historical evolution of human societies
has consisted in the fact that collectivities less extensive than the species have multiplied and,
as it were, spread their means of action in spatial externality and their institutions in
administrative externality, adding machines to tools, stocks to reserves, archives to traditions.
In society the solution to each new problem of information or regulation is sought in, if not
obtained by, the creation of organisms or institutions ‘‘parallel’’ to those whose inadequacy,
because of sclerosis and routine, shows up at a given moment. Society must always solve a
problem without a solution, that of the convergence of parallel solutions. Faced with this, the
living organism establishes itself precisely as the simple realization – if not in all simplicity – of
such a convergence. (Canguilhem, 1991/1966, pp. 254–255)
but changing with time, societal norms and underlying injunctions that we do not
question.
Contrary to what it might seem at this point, I do believe that it is possible to
advance on an understanding of human social systems by starting from the theory of
autopoiesis. However, it requires a detour, not a direct mapping; this paper merely
starts along this path. It requires understanding cognition and its relation to life. In
recent years, the effort has been increased in attempts to explore the continuity between
life and mind by taking seriously the autonomy of the organism and the experience of
the cognizer. This approach began in the work of Varela himself and his colleagues
(Varela, Thompson, & Rosch, 1991) and was complemented by efforts in different
disciplines including neuroscience, biochemistry, philosophy of biology, cognitive
science, and artificial life, leading to what now may be loosely called an enactive
approach to cognition (Thompson, 2007). Attempts to ground mind in life have
revealed a need to develop the theory of autopoiesis by disclosing important
distinctions that remain unthematized in the original literature (the version to which
most of the work on social autopoiesis reverts). If nothing else, these distinctions will
simply add to the toolbox of systemic concepts that may be deployed to approach social
systems. But, more likely, they may introduce changes in the very questions that drive
this research. My purpose here will be to expound the potential of developments in
embodied and biologically grounded approaches to cognition to drive these changes.
It will be an unsatisfactory paper in many respects. Risking, and almost certainly
achieving, a degree of unfairness, I will not go into any detailed exegesis of the rich
literature on social autopoiesis. This is a sin that my list of references will verify. The
reason for this is solidarious with a second shortcoming: I will not, strictly speaking,
develop in full the passage from life to human mind and from human minds to the
social. To do this with the care it deserves is beyond the scope of this paper (and of
this single author). Moreover, the very direction of this ‘‘passage’’ will soon come into
question. I will instead present some themes that become apparent (and sometimes
recur) when attempting to reach, from the departure point of autopoiesis, relatively
more modest but still quite tall targets such as the concepts of normativity, agency,
and autonomy. In this way I seek to make a good out of two wrongs. By remaining at
the ‘‘lower ends’’ of the history of transitions in life and mind (i.e., by not quite
reaching the complex stages of human beings and even less those of human social
institutions), I intend to focus on those aspects of a systemic and biologically
grounded approach to cognition that move beyond the theory of autopoiesis and yet
remain (slightly) neutral with respect to the social autopoiesis debates. The point
being that, in my opinion, many such discussions will simply dissolve as we begin to
explore how autopoiesis is extended into an enactive theory of cognition.
2 Bare Autopoiesis
grounding. This is not to say that cognitive science has lacked a theory. The
computational/representational view of the mind (or cognitivism, according to which
cognition is essentially information processing that occurs in the brain) has
dominated the field since its beginning and has developed to a very sophisticated
degree. However, this view has been strongly criticized over the last two decades from
different fronts (robotics, phenomenology, cognitive linguistics, and situated artificial
intelligence, to name a few). The general gist of these criticisms is that observing
cognitive systems ‘‘in the wild’’ often throws a very different picture to that of
cognitivism, one where complex, causally spread processes encompassing the brain,
the body, and the environment self-organize in opportunistic ways to produce
appropriate performance under tight temporal constraints. These observations do
not fit well the computational/representational picture, as they demand a deeper
understanding of the autonomy and identity of a cognitive system. However,
criticisms of cognitivism have so far remained like an asteroid belt of negativity
around a computational gravity pit into which cognitive science keeps falling again
and again. This situation has sometimes resulted in the paradoxical adoption of
critical terminology, words like embodiment or dynamics, as a makeover for essentially
quite classical views of cognition. What is still lacking is a theoretical core that is rich
enough (though not necessarily simple or easy to sell) so as to nucleate these criticisms
and at the same time offer a novel positive alternative by thematizing the blind-spots
of cognitivism into genuine research questions. Hence, what some people have called
the enactive approach has, at its most radical (Varela et al., 1991; Thompson, 2007),
turned to the theory of autopoiesis as its conceptual nucleus.
It soon became clear that planet enaction would not hold together and provide a
hospitable surface to migrate into unless a proper look was cast at its autopoietic
core. It is a mistake to take the theory of autopoiesis as originally formulated as a
finished theory (a trap that is easy to fall into because of the rather decisive and
muscular language with which the theory is presented in the primary literature — it
reads as if ‘‘love it or leave it’’ is the only choice). And yet, several researchers have
recently come to the realization that, as a theory of life and as theory of cognition,
autopoiesis leaves many important questions unanswered. In particular, several
essential issues that could serve as a bridge between life and mind (like a proper
grounding of teleology and agency) are given scant or null treatment in the primary
literature, and questions about important biological phenomena such as health are
not even raised.
We must break free from the binary choice in order to make progress, and must do
so in more than one sense. We must be able to criticize autopoiesis constructively and
soften up its edges so as to create a more fertile theory that holds dear to the most
radical and richest ideas of the original formulation and at the same time allows for
the fact that these ideas do not explain everything and must be elaborated and/or
complemented in order to be useful for the study of cognition (and social systems).
We must also go beyond binary choices in the very recognition that the phenomena
that autopoiesis has problems with are precisely the biological, cognitive, and social
phenomena that are best understood not in binary but in graded and comparative
terms: values, meaning, norms, pathologies, and temporality.
Overcoming Autopoiesis 47
easily reconciled with the primary literature on autopoiesis. To make it work, Varela
must introduce notions such as ‘‘breakdowns’’ in autopoiesis, which may be major or
minor (Varela, 1991), thus running against the conservation doctrine. Intuitive as
such a notion sounds, it makes no sense since autopoiesis is an all-or-nothing
property. Autopoiesis as such does not come in degrees (Maturana & Varela, 1980,
p. 94). Otherwise, all the talk about conservation simply evaporates (can a system
conserve only a part of its autopoiesis?). We simply cannot derive from the axioms of
autopoiesis that an autopoietic system will attempt to improve a situation that leads
to the future loss of autopoiesis. This would break the assumption of structural
determinism by linking knowledge about past events, or about what has occurred
to others, in conjunction with the system’s current state with a reference to a future
condition — a heresy.
And yet, this is precisely what being a cognizer is all about. How, then, to reconcile
autopoiesis and cognition?
3 Autopoiesis Plus
It would seem as if the conservation doctrine of autopoietic theory is at the root of the
problem. However, the set-theoretic analysis based on organizational properties
(as opposed to only structural ones) is one of the strongest contributions of the
theory. To get rid of it by softening the concept of autopoiesis (making it something
relative and capable of partial breakdowns) amounts to reverting to a hazy view of
living systems as being defined by a list of properties (growth, reproduction,
responsiveness, etc.), the very view that autopoietic theory is trying to overcome.
Moreover, such a move would do a biologically grounded theory of cognitive systems
no favors since differences in cognitive performance or cognitive capability would be
too easily married to essentially metabolic differences. The cognitive domain would
not be grounded in autopoiesis but reduced to it. This reduction would be unable to
provide explanations of how some cognitive engagements could ever find or produce
meaning in situations that do not immediately affect metabolism and yet may have
future consequences for it (such as a predator spotting the snow prints recently left
by a prey).
Within the terms of the problem we must attempt a solution that will provide the
required property for a grounding of cognition that complements bare autopoiesis:
a property that should (1) come in degrees, (2) respond differentially to different
situations according to their consequences for the organism, (3) sometimes malfunc-
tion, (4) obey the axiom of structural determinism, and yet (5) allow the living system
to alter its present operations with respect to nonactualized situations.
As external observers we can recognize and evaluate structural differences in
autopoietic systems that bear on their future continuity. We can make distinctions
that result in measurable events and have predictable consequences. We know when
a sick organism has a few hours or days to live. We can indeed point to structural
breakdowns in its realization of autopoiesis. We also know it is still alive during this
Overcoming Autopoiesis 49
a system’s capacity, in some circumstances, to regulate its states and its relation to the
environment with the result that, if the states are sufficiently close to the limits of its viability,
1. tendencies are distinguished and acted upon depending on whether the states will approach or
recede from these proximal limits and, as a consequence,
2. tendencies that approach these limits are moved closer to or transformed into tendencies that
do not approach them and so future states are prevented from reaching these limits with an
outward velocity.
One further important aspect of this emerging picture is that only thanks to
adaptivity can we speak of organismic dysfunction, stress, fatigue, maladaptation,
and pathology. Autopoiesis in the conservation view is blind to such phenomena since
they all occur while the system is still autopoietic, but adaptivity provides a measure
for them. Indeed, it is possible to define these phenomena in terms of failures
of adaptivity such as the exhaustion of adaptive resources, malfunction of regulation,
loss of adaptive buffering provoking the activation of extreme regulation,
disharmonious activation of conflicting adaptive mechanisms, and so on. Thus, by
re-establishing an adapted state, possibly through the simultaneous repair of adaptive
processes and change in the range and kind of acceptable relations with the
environment, a successful cure may well redefine rather than simply restore the
organism’s own normativity. Health, from this perspective, is very different from a
statistical species-specific correlation of normality, and there are consequently many
ways of being healthy (Canguilhem, 1991/1966; Goldstein, 1995/1934).
4 Norms of Life
determine what sort of access it has (if any) to the norms that describe its different
modes of viability. This access may be less or more mediated (the difference,
say, between reacting with aversion to contact with hot surface vs. planning our
movements so as to avoid touching it, both of which are examples of adaptive
regulation). Jonas’ contention is that in the history of life, novel forms of increasingly
mediated engagements have appeared allowing for more freedom at the cost of more
precariousness.
A good example, but not the only one, is provided by animality, where a new order
of values is found with the arrival of motility and the coemergence of perception,
action, and emotion. By putting a distance and a time lapse between the tensions
of need and the consummation of satisfaction, the temporality of adaptivity is
‘‘spatialized.’’ Animals can appreciate right now the danger that is impinging on them
from a distance. This is the origin of a special relation with the world, that of
perception and action, which is charged with internal significance, and hence with the
development of an emotional dimension (what might have been an inner life of need
and satisfaction now becomes rich in possibilities such as fear, desire, apprehension,
distension, tiredness, curiosity, etc.). But this comes at a cost of more severe energetic
demands (allowing the necessary fast and continuing movement across varying
environmental conditions without replenishment for long periods) and novel forms of
risk. Jonas identifies other such transitions, for instance, those afforded by a complex
visual system or the capacity to make images. It is clear that no intrinsic gain is
implied at the metabolic level by expanding the realm of freedom at the cost of
increased precariousness. However, he says, ‘‘the survival standard is inadequate for
an evaluation of life’’ (Jonas, 1966, p. 106). He goes on:
It is one of the paradoxes of life that it employs means which modify the end and themselves
become part of it. The feeling animal strives to preserve itself as a feeling, not just a metabolizing
entity, i.e., it strives to continue the very activity of feeling: the perceiving animal strives to
preserve itself as a perceiving entity – and so on. Without these faculties there would be much
less to preserve, and this less of what is to be preserved is the same as the less wherewith it is
preserved. (ibid)
Effectively, such transitions inaugurate a domain that feeds back on itself; they
imply a new form of life, not just in a metaphorical sense, but in the strict sense of a
novel process of identity generation underdetermined by metabolism.
But how is this possible? Can we make sense of this in terms of bare autopoiesis?
The problem of how to connect the constructive and the interactive aspects of living
organization is already inherent in the phrasing of autopoietic theory. This difficulty
is hard to appreciate (let alone resolve) from within the terms of theory. The problem
is the impossibility of crossing the operational and the relational domains. The first
pertains to the functioning of the autopoietic network so that it constitutes a unity
(a composite system), the second to the relations that such a unity enters into in its
structural coupling with the environment (e.g., see Maturana, 2002). For all the
logical accountancy that this separation into so-called ‘‘non-intersecting domains’’
affords, the authors have not dwelled on the problem that this separation brings,
i.e., a systemic analogue to mind–body dualism. In effect, the theory of autopoiesis
Overcoming Autopoiesis 53
says nothing about how relational interactions and internal compensations are
coordinated. They just happen to be or there is no autopoietic system. So, in this
specific interpretation (irreducibility as nonintersection), autopoietic theory is strictly
Cartesian in a way that Jonas is trying to avoid.1 The Malebranchean solution, —i.e.,
a divine intervention to ensure that body and mind, being nonintersecting substances,
remain in coordination — is today represented by appeals to evolution, an appeal
pregnant in the phrase ‘‘otherwise it disintegrates.’’ Evolution takes care of sieving
out those unhappy organisms for which the two domains are uncoordinated.
The major problem that is apparent with this solution is that the relation between
the two domains is purely contingent. It happens to be like this because it has helped
the system survive. This falls short of grounding the cognitive in the systemic. For,
without denial of the role of evolution, it is clear that grounding mind in life requires
establishing the necessary links between phenomena in these two domains. What an
organism is and what it does are not properties external to each other. By contrast,
the Jonasian solution is that a transition to a sustained new form of value-making
(such as in animality or image-making) modifies the very organizational conditions
that made transition possible. It either changes the form of identity generation
that sustains the new interactive domain, or indeed it establishes a new form of
autonomous identity (the feeling animal, the perceiving animal, etc.).
Despite the problem just highlighted with autopoietic theory, the idea of other
forms of autonomy in terms of operationally closed dynamics apart from autopoiesis
is indeed an acceptable possibility. The theory highlights the operational closure
of the nervous system (Maturana & Varela, 1980, p. 127), and Varela has suggested
that other domains may possess similar forms of autonomy, albeit not in terms of
relations of material production. Such could be the case of conversations and social
interactions (Varela, 1979, 1991, 1997). However, what is left unsaid is in what ways
can such identities relate to one another. The transformative relations between
constructive and interactive aspects of autonomy leading in themselves to a novel
form of identity cannot be directly addressed by autopoietic theory. This is simply
because a logical barrier is put between the two domains and because an emphasis
on conservation of autopoiesis obscures the possibility of a structural becoming of
novel forms of organization encompassing both constructive and interactive aspects
of the living.
1
The intention behind the distinction between domains is clear: to prevent any attempt at reducing
phenomena across domains. Given that one is a domain that is established by the presence of a whole unity
and its relations to its environment, there are good systemic reasons to distinguish those relations from
constitutive processes that give rise to the unity. To reduce phenomena across these domains is to confuse
things, to search for the speed of the car inside its engine. This is a strong point that should be preserved.
The problem, however, is introduced by the term ‘‘non-intersecting.’’ This implies strict separability
whereas in fact, non-reducibility does not imply isolating the phenomena between domains. In this way, it is
indeed possible for explanations in domain A to depend on phenomena in domain B, but not exclusively so;
a powerful engine helps us make sense of the speed of the car even if we cannot deduce the latter exclusively
from the former. Where we must be careful is in the form that such a dependence takes since any relation
across domains will always be a relation of modulation or constraint, and not of determination.
54 Ezequiel A. Di Paolo
We can now understand why the transitions in mediacy described by Jonas (and
several others not made explicit by him) have an irrevocable character. They are
authentic births of new lifeforms. These new lifeforms may relate to the metabolic
substrate and other lifeforms in a variety of ways, calling for veritable topology of
processes of identity generation (intersecting, embedded, hierarchical, shared, etc.).
It is also an open possibility that the dependence on a form of life so much modifies
the basic autonomy of metabolism that the higher identity essentially intervenes
in the very condition of organizational closure of autopoiesis (at least temporarily
in the case of vivipary or indefinitely as in cases of permanent medical intervention).
We shall return to the last possibility later. However, a proper treatment of these
problems is beyond the scope of this paper.
It will be important to remark that, from a systemic perspective, the relation
between different self-sustaining processes enabled by a substrate of autopoiesis need
not be one of perfect harmony and that, on the contrary, the inherent regulative
tendencies of sophisticated processes of identity generation are likely to enter into
conflict even with basic metabolic values. I have proposed that habits should be seen
as such autonomous structures (encompassing partial aspects of the nervous systems,
physiological and structural systems of the body, and patterns of behavior and
processes in the environment) (Di Paolo, 2003, 2005). And habits, as we know, can
be ‘‘bad.’’ The question is that as self-sustaining structures, they are never bad
for themselves, but for some other identity (typically, in the case of humans, a
combination of the metabolic and socio-linguistic self).
I have dwelled on the complex issues that emerge from attempting to answer
questions about cognition in connection with autopoietic theory partly in order to
demonstrate the difficulties inherent in such a task. The warning should be clear:
exporting and expanding the concept of autopoiesis is never an easy ride. I will now
sketch more specifically how the enactive approach has been applied to questions in
social cognition.
For the enactive view, cognition is an ongoing and situated activity shaped by life
processes, self-organization dynamics, and the experience of the animate body. This
approach is based on the mutually supporting concepts of autonomy, sense-making,
embodiment, emergence, and experience (Di Paolo, Rohde, & De Jaegher, forth-
coming; Thompson, 2005, 2007). In this perspective, the properties of living and
cognitive systems are part of a continuum and relate to each other in mutually
constraining ways. What provisionally could be designated as an ontological ordering
(from life to mind) ends up overcoming itself into more complex inner relationships
through which mind may have a life of its own and constrains the domain of
metabolism that gives rise to it (a point that we shall elaborate later). We could
expect, in principle, a similar situation if we consider the relations between life, mind,
and society. Only recently has an enactive view been proposed to account for general
Overcoming Autopoiesis 55
aspects of micro-level social interactions. And indeed the general picture partly
repeats itself.
Of the five core ideas, the concepts of embodiment and experience have received
much attention. I will not discuss them in detail here (see Di Paolo et al., forthcoming
and references therein). We define sense-making as the engagement of a cognitive
system with its world in terms of significance or value. Action and perception as well
as affective processes are forms of sense-making. It is an activity binding affect and
cognition together at the very origins of mental life. This is in contrast to the more
traditional view of organisms receiving information from their environment in a more
or less passive manner and then processing it in the form of internal representations,
which are invested with significant value only after such processing. Natural cognitive
systems do not build ‘‘pictures’’ of their world (accurate or not). They engage in the
generation of meaning in what matters to them according to the logic laid by their
self-sustaining identity. They enact a world. The notion of sense-making grounds in
biological organization a relational and affect-laden process of regulated exchanges
between an organism and its environment.
Sense-making is connected with the regulatory capacities of an organism, but more
generally with the presence of a process of identity generation. This is the idea
of autonomy that we adapt from (Varela, 1979) to include a requirement of
precariousness (see also Di Paolo, 2009). Accordingly, an autonomous system is
defined as a system composed of several processes that actively generate and sustain
an identity under precarious circumstances. In this context, to generate an identity is
to possess the property of operational closure. This is the property that among the
conditions affecting the operation of any constituent process in the system there will
always be one or more processes that also belong to the system. And, in addition,
every process in the system is a condition for at least one other constituent process,
thus forming a network. In other words, there are no processes that are not
conditioned by other processes in the network, which does not mean, of course, that
external processes cannot also influence the constituent processes, only that such
processes are not part of the operationally closed network as they do not depend on
the constituent processes. Similarly, there may be processes that are influenced by
constituent processes but do not themselves condition any of them and are therefore
not part of the operationally closed network. In their mutual dependence, the network
of processes closes upon itself and defines a unity that regenerates itself (in the space
where these processes occur). Precarious circumstances are those in which isolated
constituent processes will tend to run down or extinguish in the absence of the
organization of the system in an otherwise equivalent physical situation. In other
words, individual constituent processes are not simply conditioned (e.g., modulated,
adjusted, modified, or coupled to other processes), but they also depend for their
continuation on the organizational network they sustain; they are enabled by it and
would not be able to run isolated.
We shall return to a discussion of the concept of autonomy. As we have seen, similar
constitutive and interactive properties have been proposed to emerge at different
levels of identity-generation, including sensorimotor and neuro-dynamical forms of
autonomy (Thompson, 2007; Di Paolo et al., forthcoming; Varela, 1979, 1997).
56 Ezequiel A. Di Paolo
The notion of emergence has had a revival over the last three decades with the
advent of the sciences of complexity. Beyond the debates about the possibility
of ontological emergence (Kim, 1999; Silberstein & McGeever, 1999), there is a
pragmatic application of the term that stems from the well-understood phenomenon
of self-organization, which has served to remove the air of mystery around emergence
in order to bring it back in line with a naturalistic project. Emergence is used to
describe the formation of a novel property or process out of the interaction of
different existing processes or events (Thompson, 2007; Thompson & Varela, 2001).
In enaction, a relatively strong sense of emergence is often implied (in our case, the
sense is slightly stronger than Thompson’s). Accordingly, in order to distinguish an
emergent process from simply an aggregate of dynamical elements, two things must
hold: (1) the emergent process must have its own autonomous identity, and (2) the
sustaining of this identity and the interaction between the emergent process and its
context must lead to constraints and modulation to the operation of the underlying
levels. The first property indicates the identifiability of the emergent process
whose characteristics are enabled but not fully determined by the properties of the
component processes. The second property refers to the mutual constraining
between emerging and enabling levels (sometimes described as circular or downward
causation).
Based on these core ideas, an enactive theory of social cognition would be
concerned with defining the social in terms of the embodiment of interaction, in terms
of shifting and emerging levels of autonomous identity, and in terms of joint sense-
making and its experience. This is in contrast to defining the problem space of the
social as the expansion of a very narrow, but dominant, perspective that focuses only
on a problem that might be caricaturized as that of figuring out someone else’s
intentions; because of the detached manner in which this is supposed to happen, we
have called this a Rear Window approach to the social (De Jaegher & Di Paolo, 2007).
Many embodied criticisms of cognitivist theories of social cognition still sometimes
fall into some version of this individualism (cf., Gallagher, 2001, 2005; Hutto, 2004;
Klin, Jones, Schultz, & Volkmar, 2003). This removed cognitive problem belongs
indeed to a theory of social understanding, but it has unfairly defined the flavor of
most of the field at the expense of downplaying the role of more engaged forms of
interaction. The ‘‘social,’’ in today’s social cognition, is defined as a matter of degree
(it is nothing but a cognitively more complex domain).
The enactive perspective approaches the question of social understanding by
means of two nontraditional starting moves: first, by providing the tools that allow us
to recognize the interaction process as establishing in itself an emerging autonomous
domain, and second, by specifying how the activity of sense-making is shaped by
interaction to the point that its very nature may change to become a joint activity.
By these two moves, the door is open for the autonomy of the micro-social, a bridge
between social cognition and macro-level social structures.
For the first move, we borrow the concept of coordination from dynamical systems
theory. Coordination is the nonaccidental correlation between the behaviors of two or
more systems that are in sustained coupling, or have been coupled in the past, or
have been coupled to another, common, system. A correlation is a coherence in the
Overcoming Autopoiesis 57
behavior of two or more systems over and above what is expected, given what those
systems are capable of doing. For instance, observing a lot of people in a main city
square, some standing and some walking, is a coherence of behavior. However, it is
hardly surprising since we expect people to walk or stand in such situations
(as opposed to hover above the ground, which is not possible, or crawl, which is not
usual). However, if we found that they are all facing the same direction, this would
be a correlation. It is unlikely, though not impossible, that this would be accidental,
however, if we were to discover a common source (a giant TV screen) for this
correlation, then this is a case of coordination.
Of course, coupling itself is often a source of coordination, a well-known fact in
physical and biological systems (Kelso, 1995; Kuramoto, 1984; Winfree, 2001).
Coordination is to be expected under a variety of circumstances and does not
generally require the postulation of a dedicated mechanism underlying it. It is, on the
contrary, often quite hard to avoid. For instance, when asking two people to avoid
synchronous oscillations while swinging a pendulum with their arms, Schmidt and
O’Brien (1997) found that their oscillations were independent (uncoordinated) when
not looking at each other, but presented strong tendency to synchronize when
they were allowed to look at each other. Such synchrony is a form of absolute
coordination: two series of events are perfectly entrained. Relative coordination,
in contrast, has a much wider range of possibilities (Kelso, 1995), as there are no
such transitions from one strictly coherent state to another. Systems in relative
coordination do not entrain perfectly. Instead they show phase attraction, which
means that they tend to go near perfect synchrony, and move into and out of the zone
that surrounds it. This is a common phenomenon in biology (Haken & Köpchen,
1991). Of course, coordination may be more than entrainment. Many cases of
appropriately patterned behavior, such as mirroring, anticipation, imitation, etc., are
general forms of coordination according to our definition.
Several researchers in social science have recognized the importance of different
forms of coordination for understanding social interaction, e.g., the tradition
championed by figures such as Erving Goffman, Harvey Sacks, and others (see, e.g.,
Goffman, 1972, 1983; Sacks, 1992; Sacks, Schegloff, & Jefferson, 1974). A whole field
of study is dedicated to uncovering behavioral coordination in interaction going
under different labels such as interaction studies, conversation analysis, and gesture
analysis (see Schiffrin, 1994). Similarly, the coregulation of different kinds of social
spaces during interaction has been the interest of social science since at least the
work of Edward T. Hall (1966) and Adam Kendon (1990). From the enactive
perspective, the concept of coordination helps us to understand social interaction as
an ongoing process with a space–time structure and organizational properties.
In most approaches that care to define it, social interaction is simply the spatio-
temporal coincidence of two agents that influence each other (e.g., Goffman, 1972,
p. 1; Schutz, 1964, p. 23). We must move from this conception toward an
understanding of how a history of coordination demarcates the interaction as an
identifiable pattern with its own role to play in the process of social understanding.
In the social domain, patterns of coordination can directly influence the continuing
disposition of the individuals involved to sustain or modify their encounter.
58 Ezequiel A. Di Paolo
In this way, what arises in the process of coordination (e.g., gestures, utterances,
changes in intonation, etc.) can steer the encounter and facilitate its continuation.
The unraveling of these dynamics itself influences what kinds of coordination are
likely to happen. This is due to the fact that the interactors are highly plastic systems
that are susceptible to being affected by the history of coordination. When this
mutual influence is in place (from the coordination onto the unfolding of the
encounter and from the dynamics of the encounter onto the likelihood to coordinate),
we say that we are in the presence of a social interaction. This emergent level is
sustained and identifiable.
In accordance with the core ideas of enaction, the above description is nothing less
than that of an emergent and autonomous process. It is, however, typically a fleeting
one. Even though normal social encounters, for instance conversations, may only last
a few minutes, our point is that during that period they may organize themselves
according to the two avenues of influence just described: the agents sustain the
encounter, and the encounter itself influences the agents and invests them with the
role of interactors. The interaction process emerges as an entity when social
encounters acquire this operationally closed, precarious organization. It constitutes a
level of analysis not reducible to individual behaviors. This perspective bypasses the
circularity that arises from preconceiving individuals as ready-made interactors.
Individuals coemerge as social agents with the social process. This brings us to
the second requirement for calling an interaction properly social. Not only must the
process itself enjoy a temporary form of autonomy, but the autonomy of the
individuals as interactors must also remain unbroken (even though the interaction
may enhance or diminish the scope of individual autonomy). If this were not so, if the
autonomy of one of the interactors were destroyed, the process would reduce to the
cognitive engagement of the remaining agent with his nonsocial world. The ‘‘other’’
would simply become a tool, an object, or a problem for his individual cognition.2
In (De Jaegher & Di Paolo, 2007), we propose the following definition of social
interaction:
Social interaction is the regulated coupling between at least two autonomous agents, where
the regulation is aimed at aspects of the coupling itself so that it constitutes an emergent
autonomous organization in the domain of relational dynamics, without destroying in the
process the autonomy of the agents involved (though the latter’s scope can be augmented or
reduced). (p. 493)
2
There may be a social motivation and social consequences to the act of destroying someone else’s
autonomy, but the act itself does not in itself constitute an case of inter-action. The definition of social
interaction is aimed at capturing the fact that the other is not fully constituted as a cognitive object by my
own actions, but sometimes plays along with them, and sometimes not and this is the crux of the problem of
social cognition, the shift backwards and forth between these conditions and the unobjectifiable character of
the other.
Overcoming Autopoiesis 59
along a narrow corridor in opposite directions. In order to get past each other, they
must adopt complementary positions by shifting to the left or to the right. Sometimes
they happen to move into mirroring positions at the same time creating a symmetrical
coordinated relation. Due to the spatial constraints of the situation, such symmetry
favors an ensuing shift into another mirroring position (there are not so many more
moves available). Coordinated shifts in position, then, sustain a property of the
relational dynamics (symmetry) that all but compels the interactors to keep facing
one another, thus remaining in interaction (despite, or rather thanks to, their efforts
to escape from the situation). In addition, the interaction promotes individual actions
that tend to maintain the symmetrical relation. Coordinated sideways movements
conserve symmetry, and symmetry promotes coordinated sideways movements. While
it lasts, the interaction shows the organization described above in terms of the mutual
influence between individual actions and relational dynamics. It becomes clear that
interaction is not reducible to individual actions or intentions but installs a relational
domain with its own properties that constrains and modulates individual behavior.
(Anyone who has reluctantly participated in a self-fuelling argument will immediately
appreciate the parallels.)
An immediate consequence of this perspective is that if the regulation that sustains
a social interaction happens through coordination patterns, and if those patterns
affect the movements — including utterances — that are the tools of individual sense-
making, then social agents can coordinate their sense-making during interaction,
resulting in participatory sense-making: the interactive coordination of intentional
activity, whereby new domains of sense-making may appear that were unavailable to
each solitary individual.
Participatory sense-making describes in fact a qualitative spectrum of involvement,
going from the mere modulation of meaning by physical aspects of the interaction
(e.g., delays on a video-conferencing line that affect the fluidity of a conversation and
might be sometimes interpreted meaningfully) to intentional regulation of activity
between interactors (orientation, teaching) and to cases where the proper act of sense-
making is only completed by joint action (leading potentially to the creation of new
meaning).
We have seen that individual sense-making possesses the structure of a regulative
act. There is an intention in this regulation and, if successful, the conditions that gave
rise to this act are extinguished. Consider in contrast a simple social act such as the act
of giving. It already has a different structure. A single person cannot complete it
because it requires acceptance from another. In a study of mother–infant interaction,
Fogel (1993) describes a filmed session between a 1-year-old baby and his mother
that captures possibly the baby’s first act of giving. The baby extends his arms and
holds it relatively stationary only to gently release the object as the mother’s hand
approaches. The object is released only as the mother gently pulls it.
Assuming for a moment that the infant is the initiator of the act, we realize that he
must create an opening by his action that may only be completed by the action of the
mother. The giving involves more than orientation of the mother’s sense-making;
it involves a request for her not only to orient toward the new situation, but also to
create an activity that will bring the act to completion — in other words, to take up
60 Ezequiel A. Di Paolo
the invitation for an intention to be shared. This invitation may go unperceived and
the act frustrated. But this is not the same as the situation in which the invitation is
perceived and declined. The two situations are different from the perspective of the
mother, and this difference confirms that an invitation to participate is experienced
as a request to create an appropriate closure of a sense-making activity that was
not originally hers. To accept this request is to produce the ‘‘other half of the act,’’
bringing it to a successful completion. When we remove the simplifying assumption
that the infant intentionally originated the act, we open up the possibility for even
richer degrees of participation. The act may then indeed result from a ‘‘coregulation’’
that emanates from previous aspects of the interaction, as Fogel proposes. A certain
movement extending the object in the direction of the mother, without yet intending
to give it, may now be opportunistically invested with a novel meaning through joint
sense-making. Latent intentions become crystallized through the joint activity so that
not only the completion of the act is achieved together, but also its initiation.
This sketch hardly does justice to the richness of social interaction, but it highlights
the novel aspects of the enactive focus. There is no unified account that can
encompass the whole range of social capacities from primary intersubjectivity to the
highest reaches of human language and social cognition. The enactive approach has
potential to advance on some of these problems. What this approach does ensure, in
contrast to noninteractive proposals, is an explicit two-way link between individual
and social processes, leaving open the possibility for individual cognitive skills to have
dual or even purely social developmental origins. This is a strictly closed avenue for
approaches that are not properly interactive. Social skills, under the enactive view, are
by definition relational. Although agents can have different individual potentials
for entering into an interaction, this potential is modulated and transformed by actual
interactions. This is an implication of having established the autonomy of the
interactional domain. At the same time, the social domain remains social as long as
individual autonomy is not lost. This already offers a sharp contrast with some
attempts to apply the idea of autopoiesis to social systems. This dialogue, the mutual
modulation and potential ongoing conflict between autonomies, is not typically
discussed in such attempts. But it is precisely a focus concern that comes from
defining the social domain in enactive terms. Under this view, the individual and
social autonomies are not presented as mutually exclusive starting points from a
methodological standpoint. An enactive social science is concerned with what goes on
at the interface between these different forms of autonomy.
The enactive approach to life and mind, as presented above, is only now starting to
turn to the multiplicity of social phenomena. Consequently, at this point it can say
very little that is of direct relevance to specific problems in social science and
organization theory. What is possible at this stage, however, is to make explicit some
general systemic implications of this approach in the hope that they will expand
Overcoming Autopoiesis 61
and/or complement the dialogue between systemic approaches to life and social
science beyond the bare bones of autopoietic theory. This should help prevent sterile
debate around what might turn out to be false problems. Interestingly, as we shall see,
such reflections turn back into a richer understanding of closure and temporality at
the cognitive and metabolic levels.
It could be said that our approach has proceeded by disclosing some hidden
potentialities that remained unexpressed in the theory of autopoiesis: the possibility of
adaptive autopoiesis leading to active homeostasis as opposed to passive conserva-
tion; the possibility of adaptive regulation of structural coupling leading to agency
and a cognitive relation to the world; the possibility of spatializing the intentional
temporality of sense-making into action, perception, and emotion; the possibility of
emergent and overlapping processes of identity generation; and the possibility of
autonomous social interaction and participatory sense-making.
Our interpretation of these developments has so far been positive. They are
previously unthematized potentialities that do not break with the basic tenets of
autopoietic theory. There is, however, another route for theoretical development that
moves from these potentialities back to the core of autopoiesis. This route produces a
shift of perspective such that autopoietic theory moves from being a theory of (all)
the living to being a moment that allows us to grasp the phenomena of life and mind.
By this very development, this moment is overcome (aufgehoben in the Hegelian
sense) leading to a more encompassing perspective. It is my firm belief that social
systems thinking will benefit much more directly from this sublated view of life than
from bare autopoietic theory. The element that is added to our theoretical toolbox
by this turn from the higher forms of identity back into the enabling layers of
metabolism is the same kind of tool that may be used to dispel much of the discomfort
that work on social autopoiesis still provokes due to its lack of a proper analysis of
the relations between the social, the cognitive, and the metabolic levels of autonomy.
The ‘‘negative’’ development is best examined as a possibility pregnant in the
concepts of operational closure, precariousness, and interactive autonomy. This is
shown at the simplest level of a single autopoietic organism, but a similar analysis
may be repeated in its central points for all the other possibilities described above
where we have seen new forms of autonomy emerging, especially for the case of the
autonomy of social interactions.
6.1 Reduplication
under which the system is inviable in the long term but not necessarily immediately
destroyed — in other words, conditions under which the system is in a ‘‘dangerous’’
transient moving toward its boundary of viability. Agency allows the system to cope
with a portion of such dangerous conditions (conditions, as we have seen, that are
also cognitively evaluated as dangerous by the system itself by this very coping).
It does so by regulating structural coupling in such a way that a dangerous condition
is not allowed to subsist long enough to lead the system to destruction. A temporal
dimension is introduced in the set of environmental conditions. The severity of these
conditions becomes a matter of degree, which is ‘‘measured’’ by of the adaptive
regulation deployed to cope with them.
Sometimes the regulation at the level of agency can be so reliable as to allow quite a
durable persistence in dangerous conditions. As an example, consider the water
boatman, one of several species of insects able to breathe underwater by trapping air
bubbles (plastrons) in the tiny hairs of the abdomen. The bubbles refill with oxygen due
to the differences in partial pressure provoked by respiration and potentially can work
indefinitely (see, e.g., Thorpe, 1950). They allow the insect to spend time underwater for
longer periods thanks to a mediated regulation of environmental coupling (which is
nevertheless riskier than normal breathing). The regulation of coupling (agency) takes
the form of maintaining an external structure that directly supplements a vital function
in an environmental condition that belongs to the unviable subset.
6.2 Life-Support
and interactive domains do intersect. Precariousness acquires a higher order; not only
are the constitutive metabolic processes unable to continue in the absence of the
closed network of relations, but the network itself is unable to remain closed in the
absence of the interactive regulation that it originally gave rise to. But, following
Jonas, so does freedom acquire a higher order. The set of conditions under which life
thrives is now extended so that the transformed metabolism/agent is now able to
survive situations in which all possible conditions would lead to its destruction if it
had remained only metabolism.
And, again following Jonas, as the means that modify the end themselves become
part of it, interactive life may acquire a closure of its own. It becomes autonomous by
self-organizing plasticity and behavior into habits. It also becomes normative, not only
by (a) filling in with its own norms a ‘‘metabolically neutral’’ space of values and
conserving a higher form of life that is enabled by metabolism, as we have already
suggested (see also Di Paolo, 2005), but also by (b) potentially driving metabolism
to depend on this new form of life. For while the possibility of option (a) is always
present, the normativity it introduces is only constrained by that of metabolism — this
normativity should not contravene the latter’s viability. New norms, in this case, relate
to autopoiesis in a contingent way (like a strong preference in terms of taste between
two kinds of food of similar nutritional value). In other words, such norms are
metabolically indifferent. This case alone would not be sufficient to understand why the
Jonasian transitions would be irrevocable. However, in case (b) metabolism itself
changes fundamentally due to the possibilities afforded by autonomous agency.
Normativity in the interactive domain is now not entirely contingent and will bear
an inner relation to the normativity of metabolism (e.g., a preference for a certain taste
in food changing metabolism so that it actually becomes more nutritious; you are what
you prefer to eat). Agency in this case does more than downward-regulate metabolism;
it ‘‘downward-constitutes’’ it. (In the imaginary case of the permanently underwater
insect, this could take the form of the development of a metabolic accommodation to
other gases dissolved in the water apart from oxygen followed by a subsequent
specialization in diet enabled only because of this new metabolic adaptation).
So we can appreciate that it is within the potentialities of agency to alter the
domain of viability of metabolism so as to allow it to subsist in conditions that would
otherwise be inviable. However, this very alteration can potentially allow metabolism
to drift into previously inaccessible situations in which all conditions are inviable,
as long as it remains under the self-scaffolded life-support of agency.
In strict terms, such a system would be alive but not operationally closed in and of
itself, i.e., as a composite network of processes whose operation regenerates the
network and defines it as a unity. It is alive inasmuch as these very same conditions
still verify. But these conditions obtain thanks to the system’s actions in the relational
domain in which the unity as a whole enters. This relational domain becomes not
a contextual, enabling (and essentially contingent and external) condition for the
conservation of life, but a necessary, active, and operational process flowing from the
relations subtended by the whole unity into the constitution of the composite system.
In other words, the system is alive and not sensu stricto autopoietic. It functions as
a life/mind unity — the self-sustaining structures of the interactive domain (habits)
Overcoming Autopoiesis 65
become mutual renditions (not just external coordinations) between the psychic and
the somatic.
6.3 Life/Mind
3
Isn’t the inevitability of death related to the situation whereby all environmental conditions are inviable? If
so, perhaps death comes as a consequence of the sublation of autopoiesis in life/mind. Notice the difference
between any autopoietic system that ceases to exist because it faces an inviable condition, but for which the
set of viable condition remains not empty (a prokaryote cell might in principle reproduce ad infinitum if the
external conditions are maintained stable, but can of course die as soon as a negative intervention is made in
its medium or directly on it) and those life/mind systems for which all conditions are inviable and yet subsist
thanks to life-support. Maybe the organizational precariousness of such systems puts them in a situation
where death is unavoidable since they have no ‘‘stable home’’ except death.
66 Ezequiel A. Di Paolo
7 Conclusion
The logic of the return from the higher to the lower — the life-support of agency that
turns into the agency of life-support; the scaffold that modifies the terrain upon which
it is assembled — is generally lacking in systemic approaches to social systems (and up
to this point cognitive science has been in no better position). If we open up a channel
through which behavior plastically inscribes itself in the body, so then might social
interactions and so could social institutions.
Once this possibility is understood in systemic terms, some concomitant questions
that arise in debates about whether social systems are or not autopoietic (where
are the boundaries?, what is a component?) stop making sense. Such questions form
part of an ineffectual discourse and inevitably lead back to the dichotomies of
structuralist vs. individual actor perspectives on society. Overcoming a parallel
dichotomy in the relation between metabolic and agential normativity should
help us surmount the opposition between social structures and individual subjectivity.
The notion of a transition in identity generation is the key operative concept
introduced by the enactive approach. It overlays autopoietic ideas with an inherently
dynamical dimension and redefines it as a biology of transformation, not just
conservation.
The substrate of the social is not just the space of meaning (communications,
exchange) but also its inscription in living agencies, artifacts, and oblique structuring
of habits, which can only be uncovered through genealogical as well as systemic
analysis. We can envision a systemic approach to social science whose mission would
be to reveal an ecology of different social lifeforms and their transitions, conflicts, and
transformations. Some social lifeforms might be similar to the extracellular matrix
in multicellular organisms or bacterial biofilms, an active medium that structures
and is structured by the activity of social actors, something resembling Bourdieu’s
habitus (Bourdieu, 1990). Such a form is a soft machine with a receding horizon.
It makes little sense to characterize it in terms of boundaries, distinctions, and
components. The nature of this kind of social lifeform is such that any attempt at a
distinction immediately brings forth a wider background of significance and processes
in relations never closed upon themselves.
Other forms are less diffuse. Self-managed businesses (a phenomenon with many
manifestations throughout history but that that has peaked in countries like
Argentina and Venezuela over the last 10 years, born out of desperate need, strong
communal support, and institutional instability or reform) can in many cases be seen
as single exemplars of several transitions in emergent identities (with the attendant
expansion of both freedom and precariousness). In their history from bankruptcy to
profit-making self-management, they undergo a transformation from an externally
supported, almost ascriptional identity which at best is able to sustain a relation
of self-distinction — an in-itself — through a process of ‘‘de-grammaticalization’’ of
labor, a re-signification of individual and collective activity, and a devolution of
responsibility to workers and the community at large, so as to become a for-itself — a
cause. Such entities might indeed be a closer social analogue not to ruthless bacteria,
but to some higher form of life/mind.
Overcoming Autopoiesis 67
What the enactive approach invites us to do is to see in life, mind, and society, not
single unifying notions but multiplicities of events, entities, relations, and processes.
These are organized, they are not chaotic, hence the value of systems thinking, but
they induce a break from the mythology of stability, boundaries, and conservation.
By contrast, the enactive approach foregrounds a discourse of transformations,
freedom, precariousness, identities, norms, negativity, temporality, sense-making,
and re-inscriptions of meaning in matter and bodies — all notions largely absent in
autopoietic theory but not incompatible with it.
References
Bourdieu, P. (1990). The logic of practice. Cambridge: Polity Press.
Canguilhem, G. (1965). La connaissace de la vie. Paris: Vrin.
Canguilhem, G. (1966). The normal and the pathological. New York, NY: Zone Books.
Chesterton, G. K. (1910). What’s wrong with the world. London: Cassell.
Comte, A. (1830–42). Cours de philosophie positive (4th ed.). Paris: J.-B. Bailliere.
De Jaegher, H., & Di Paolo, E. A. (2007). Participatory sense-making: An enactive approach to
social cognition. Phenomenology and the Cognitive Sciences, 6(4), 485–507.
Di Paolo, E. A. (2003). Organismically-inspired robots: Homeostatic adaptation and natural
teleology beyond the closed sensorimotor loop. In: K. Murase & T. Asakura (Eds),
Dynamical systems approach to embodiment and sociality (pp. 19–42). Adelaide, Australia:
Advanced Knowledge International.
Di Paolo, E. A. (2005). Autopoiesis, adaptivity, teleology, agency. Phenomenology and the
Cognitive Sciences, 4, 97–125.
Di Paolo, E. A. (2009). Extended life. Topoi, 28, 9–21.
Di Paolo, E., Rohde, M., & De Jaegher, H. (forthcoming). Horizons for the
enactive mind: Values, social interaction, and play. In: J. Stewart, O. Gapenne &
E. Di Paolo (Eds), Enaction: Towards a new paradigm for cognitive science. Cambridge,
MA: MIT Press.
Fogel, A. (1993). Developing through relationships: Origins of communication, self and culture.
London: Harvester Wheatsheaf.
Gallagher, S. (2001). The practice of mind: Theory, simulation or primary interaction? Journal
of Consciousness Studies, 8, 83–108.
Gallagher, S. (2005). How the body shapes the mind. Oxford: Oxford University Press.
Goffman, E. (1972). Interaction ritual: Essays on face-to-face behavior. London: Allen Lane.
Goffman, E. (1983). The interaction order. American Sociological Review, 48, 1–17.
Goldstein, K. (1934). The organism. New York, NY: Zone Books.
Haken, H., & Köpchen, H. P. (1991). Rhythms in physiological systems. Berlin: Springer.
Hall, E. T. (1966). The hidden dimension. New York, NY: Doubleday.
Harrington, A. (1996). Reenchanted science: Holism in German culture from Wilhelm II to
Hitler. Princeton, NJ: Princeton University Press.
Hutto, D. D. (2004). The limits of spectatorial folk psychology. Mind and Language, 19, 548–573.
Jonas, H. (1966). The phenomenon of life: Towards a philosophical biology. Evanston, IL:
Northwestern University Press.
Kelso, J. A. S. (1995). Dynamic patterns: The self-organization of brain and behavior.
Cambridge, MA: MIT Press.
68 Ezequiel A. Di Paolo
1 Introduction
Few words in modern society have become as positively charged as the word
innovation. Of course, premodern societies were also innovative in their way. Still,
technology, ideas, and organizational forms have changed over time, and it is only in
modern society that innovation has become almost mandatory; that is to say, ranked
uppermost in society’s value system. ‘‘Be innovative!’’ has become an imperative in
modern society.
Niklas Luhmann viewed innovative processes (understood as social change or
renewal) not from an action-theoretical perspective, i.e., as the result of an
intervention into a social system with the structural changes that go with it, but
rather from the perspective of self-referential processes of systems where change in
structures are interpreted as changes in communicative events (Ereignisse). Innovation
can thus be understood as structural changes where systems react to events in the
environment with a changed connectivity between communications (Luhmann, 1984,
470ff.). In action-theoretical studies, however (Joas, 1992; Giddens, 1990), it has
been argued that Luhmann’s theory of autopoietic systems prevents a satisfactory
understanding of the phenomenon of innovation because it operates with a subjectless
action concept that cannot explain learning and innovation processes. The acting
subjects that provide the occasion for changed expectations are not co-thematized,
which leaves the theory indefinite apropos the question why new structures appear.
Autopoietic systems are cognitively oriented systems where ‘‘events’’ appear in the
light of a historically arising structure and its state (what Maturana calls ‘‘structure
states’’). Experience is therefore always an important concept in autopoietic systems
and becomes in a sense a link between the system’s history and its possible future.
With the organizationally required conservatism with which a cognitive system
operates, an increasing number of possibilities become actualized, even if in the
light of complementary experiences they cannot be satisfied. This means that the
production of the radically new cannot be ruled out, but depends on the self-
referential, autopoietic system’s mode of operation. This two-sidedness is expressed in
Varelas’s (1979) definition of autopoietic systems:
According to such a definition, the new becomes a construction that must be seen
as a communicative and socially produced relationship; and not as a ‘‘Ding an Sich.’’
A posited ‘‘mechanics’’ that can explain the new in this sense is more closely
connected with the self-referential character of autopoietic systems. It is in this sense
that such systems are cognitive systems. Different reality constructions lead to
different (new) action and descriptions, which in turn lead to a differentiation process
in the treatment of similar objects. Examples are differentiation between the scientific
disciplines but also the transformation of techniques and technologies between
different subsystems — e.g., that continuous change in rocket technology and space
research depends upon other subsystems developing new materials that can tolerate
heat, which then in turn can be transformed back into space research. Such processes
can be described as innovation processes.
Innovation and Organization 71
In the following discussion, we will attempt to show how social systems, as a result
of their organization,1 are conservative systems, while they at the same time produce
social changes, e.g., in the form of innovations due to the fact that they are operative
constructions. In order to understand the phenomenon of innovation one needs the
concept of ‘‘irritations.’’2 Autopoietic systems allow themselves to be ‘‘irritated’’ by
their environment, and this produces new combinatory modes, i.e., innovations. It is
in this context, as we will see later, that decisions become so important, as they can
serve to accelerate the process.
Maturana, Varela, and Uribe (1982) have developed a simulation model for
cellular autopoiesis that helps explain how an autopoietic system works. In this
simulation model (based on biological autopoiesis), they show how the cell is
constantly renewed and changed over time. It is the spontaneous production of an
autopoietic unit, in other words, that this formalized model will illustrate. Framed in
the terminology of Wittgenstein, one might perhaps say that autopoiesis represents a
new language play that can serve to distance the problem from ontological questions
connected to what nature’s essence is. If we then introduce the predicate ‘‘produce
oneself,’’ then it becomes a part of the language play’s grammar (Wittgenstein, 1967)
that determines which organisms can be described. Through the lens of autopoiesis,
the concept of the new can thus be seen as first and foremost a new language: a new
language play establishes a new grammar, and thus new ways of seeing things that
can open the way for new combinations, i.e., prepare us to think innovatively.3
This enables a new access to reality: a new language play establishes an area
for descriptions that are different in relationship from the old; it demarcates a new
context for observation so that new phenomena become clear, which under the old
way of looking at things were not. This is how the language play can be made into the
logical primacy, and not the ontology that stands in the shadow of the language play
or is constituted by it. The grammar of the language play thus defines the space for
possible experiences that are described in the language play, such that the language
play creates a basis for distinctions, i.e., a basis for that which is described as
self-organizing, autopoietic, and so on. In this way, one can say that the limits of
1
According to Maturana and Varela, the notion of ‘‘organization’’ indicates the identity of a unity, that is to
say, what kind of class/group it belongs to. The identity of the unity tells us something about the relations
between the components/parts which define the composite unity of that class/group. The notion refers to
the relations between the components which have to be invariant in order to maintain the identity of the
unity. Luhmann did not apply Maturana/Varela’s biological notions of organization and structure, as he
felt they had little relevance for sociology.
2
In Maturana’s language it is called perturbations.
3
In systems theoretical constructivism ‘‘the new’’ is a kind of emergent order: ‘‘Emergence, in its ‘classical’
meaning, refers to the rise of new levels of being (life as opposed to inanimate nature or of spirit as opposed
to life), which in no way may be derived or predicted or explained by turning to the properties of an
underlying level. This is why they are perceived as ‘unexpected,’ ‘surprising,’ etc. A modern version of
emergence is the rise of, through microscopic interactions at a microscopic level, new qualities that cannot
be derived (causally explained, formally derived) from the properties of the components, although they
consist solely of the interaction between these components’’ (Krohn & Krüppers, 1992, p. 389 own
translation).
72 Tore Bakken et al.
experience lie in the grammar of our experience concepts, i.e., in the language play
with which we describe our experience.
For Luhmann, there is never any question of a direct analogy between social
systems and organisms. Instead he takes the roundabout way of ‘‘generalization and
respecification’’ (Luhmann, 1995). There are thus differences with the autopoiesis
concept of Maturana/Varelas in two respects: (1) Luhmann’s non-organic autopoiesis
considers the reproductive processes as ‘‘temporalized.’’ While organisms’ autopoiesis
operates with elements (e.g., cells’ organization and operation) which time-
wise remain relatively constant, social autopoiesis as well as the autopoiesis of
consciousness are based on units (communication and thoughts) that have the
character of ‘‘events’’: as soon as they take place they vanish.4 A problem thus arises
with the connectivity of an event, which is solved with the help of meaning. This leads
to a further specification of the autopoiesis concept. (2) For in structures and
processes that are organized according to meaning, external sizes such as system limits
and environment also become included in the analysis in a more cogent way (as we
shall see, this will have significance for the understanding of innovation, which
requires precisely that one disturb autopoiesis’s network). Limits and environment
presume meaning for the system, which means that such systems can operate
internally with the difference between system and environment. Meaning thus makes
it possible for co-interpretation of references not only to the system itself, but also to
the system’s environment. The observer can now, to a greater degree, choose what
should be attached importance to, the system or the environment. Meaning thus gives
support to an evolutionary gain that lies in the capacity to make new combinatory
possibilities of closure and environment openness in the system. In this sense,
Luhmann allows for operationally closed systems to be open in a new way: closure
and openness are no longer seen as opposites.
The difference between system and environment is applied system internally as a
principle for obtaining information. The operation that promotes this identity-
constitutive acknowledgement is observation, and within Luhmann’s system theory,
this is understood as marking a difference. In their difference-form systems observe
themselves as environment. This relationship between simultaneous environment
dependence and independence in systems can be described by the terms mode and
event. The systems are independent of their environment as regards their special mode
of self-government (this is how they acquire their identity), but dependent when it
comes to events in the environment perceived as information. But this opening is
always relativized: in its observations, the observing system constructs the observed
object from its own perspective. Information is therefore always a self-production of
the system, and not a fact in the environment that stand independently of observers.
This has consequences for the study of innovation processes. Instead of, as
for Maturana/Varela, the observer being ‘‘captured’’ by the system (i.e., constituted
4
This distinction is emphasized by Dirk Baecker (2000): ‘‘But Maturana and Varela use the relative static
notion of ‘component.’ Luhmann’s notion of event sheds light over a moment’s disappearance, like a
reproductive moment to the creation of the situation — and in this way we can leave the sociology’s use of
the notions of intention and action.’’ (p. 126)
Innovation and Organization 73
within a separate domain isolated from the autopoietic system that can thus exist only
in an ad hoc relation to it), the observer generates the system by drawing a distinction.
By placing the observer at the starting point, Luhmann opens the system to
alternative constructions. The activity itself that produces the system formation can
now be seen more openly. The mechanism for closure can now be shifted from how
perceptions function (Maturana et al., 1982) to how the systems’ codes function.
The autopoiesis concept in Luhmann’s version thus becomes interesting because
it provides a means for understanding how the new is produced (Bakken & Hernes,
2003, 2008). At the same time, the concept of autopoiesis makes an interesting
contribution to a better understanding of the interplay between the old and the new in
innovation processes. Starting with the autopoiesis concept, we can now bring in a
more ‘‘playful attitude’’ to the phenomenon of the new. It is no longer necessary to
accept a value system that states that everything new is better than that which existed
before.5
So how does one explain this in sociological terms? We gain little by explaining
phenomena in the modern organizational world from the standpoint that there exist
inner tendencies in organizations that lead toward more rational, improved states or
innovations, and that management’s task is to nurture these tendencies. Sociology,
on the other hand, is better off inquiring into the relationship between organization
and society.
But why the preference for the new in society? We need to go all the way back to
the early modern society of the 1600–1700s, or the period after the discovery of the
printing press, to find a decisive break (cf. Luhmann, 1995). It was then that the old
5
Luhmann (1997) shows that a shift from what to how questions gives a better description of the new. ‘‘The
beautiful world is no longer an object of religious admiration, accompanied by the practical problems of
finding ones path. The ways in which the world manifests itself gives rise to the question of how the world
came into being and how similar effects can be obtained [erzeugt] y If we know how to produce something,
it may be possible, on the basis of this, to vary [our] goals and decide to create unknown phenomena. The
‘neuzeitliche’ science formulated its understanding of nature based on references to methods and
experiments; but even the art of governing started with the question of how to obtain dominance [Herrshaft]
and remain in positions of power’’ (p. 520). The German cybernetics and philosopher Gotthard Günther
(1980) is also highlighting this, though in a little bit more Hegelian tone: ‘‘In the old classical conception of
the world [Weltbild], which, to be sure, was rich in content, but had a total contextural simplicity, nothing
truly and really new could come into existence y Nothing (totally) new newness y cannot manifest itself
within a given contexture. And since Hegel, right or wrong, interpreted nature as a closed contexture,
nature cannot, according to him, generate anything really new within this contexture. The trivial passage
from one content to the next will at the most produce subaltern ‘news,’ like the ones we find in the changing
world of fashion y The genuinely new, which holds up to the philosophical scrutiny, requires a change in
content as well as a switch of contexture. Thus, not only the first negation is involved, but the second as
well’’ (pp. 197–198, vol. 3, own translation). That is to say, the really new demands not only changes in
content but also a change in perspective.
74 Tore Bakken et al.
Latin term novus — which basically meant deviation — was temporalized. And thus
the new became a time concept in the sense of receiving attention if it deviates from
what came before. This marked a break with the old aristocratic society where age
and origin was decisive, also for all contemporary orientation. In principle one was
expected to know one’s place and all innovation was regarded as deviation. With the
growing functional differentiation of the 1600–1700s came a radical change that swept
through all the functional subsystems. Within art originality became prized; genius
and the ability to create new works were preferred over the cultivation of memory as
it applied to known stories and known picture configurations. Within the sciences
the real meaning of research turned toward innovation, and the same was true of
all technical development. Education became subject to comprehensive reforms,
particularly in Protestant schools, and within the economy all was now geared to new
markets where one demonstrated one’s success with new and improved products.
While the various functional areas may have different degrees of preference for
newness, there is nevertheless a general tendency to prefer the new over the old. To be
sure, there still exist reservations toward novelty; throughout the entire 1700s, there
was no shortage of warnings against unrest, especially among weak government
officials. Even so, and often described by sociologists, with an increasing functional
differentiation came a shift in orientation from deviation to novelty. The problem of
deviation was set up against the normative area in society, especially in the law courts.
And by virtue of its positive shift it may change, but until such time as it does, one
cannot offend against it without incurring sanctions.
The mode of thinking sketched here has long had the support of sociological
theory, but mostly in the light of the theory of differentiation understood as the
division of labor as Adam Smith prescribes, whereby differentiation operates on
the premise that society develops toward constantly increasing rationality, as well as
increased production within all sectors, not only in the form of better resource
exploitation within the economy, but also in politics in the form of increased
democracy, etc.
There is a one-sidedness in this way of thinking, and it is to this that Luhmann
directs his criticism: the hope of increased rationality as a result of the social division
of labor is today no longer tenable (Luhmann, 1995). We need a new conception of
functional differentiation, claims Luhmann, and it must be extended to include
function-systems’ autonomy in the sense of operative closure and self-regulation.
There no longer exists an external authority or an Archimedean point that stands
independent of the systems. This is well known from Luhmann’s commentaries and
will not be pursued further here.
Our point is that when Luhmann speaks of operative closure, this in no way means
causal closure. Indeed quite the opposite. The causal interdependencies increase
because each functional system is dependent upon the others functioning. In addition,
the respective systems’ problems can be exported to other systems. Politics can have
problems that it cannot solve politically, whereupon a legal clause is made with the
expectation that the constitutional law courts will solve the problem; and often
the courts then send the problem back to politics and so on. This is somewhat how the
economic system thinks when it comes to ecological regulations: my company cannot
Innovation and Organization 75
6
Cf. Luhmann (1997): ‘‘Accordingly, irritation is a systems state that stimulates continuation of the
autopoietic operations of a system. But as a simple irritation it leaves open the question of whether
structures must be changed as well; i.e. whether new learning processes will be initiated by additional
irritations or whether the system is counting on that the irritations, in time, will disappear by themselves,
since they are only a one-time event’’ (p. 790, own translation).
76 Tore Bakken et al.
We can now adopt a more reflexive attitude to innovation. In this way Luhmann
can dismiss old European descriptions of novelty and innovation as an unambiguous
rationalization tool. This is a way of thinking that expresses not only the relationship
between the old and the new, but also its kindred problem, namely the relationship
between the real and the possible.
When Luhmann was awarded the Hegel prize in 1989, Robert Spaemann made
the following characterization in his tribute: Luhmann’s ideal type is not homo
oeconomicus, politicus or sociologicus, but rather ‘‘the man without qualities’’ in the
way of the great novel by Robert Musil. In his book Musil depicts the consequences
a plural social world has upon the individual who becomes a ‘‘plural I,’’ as expressed
through the main character Ulrich. Such a social world demands that the individual
make a choice. The individual is no longer confronted by a fate-determined social
world but by a multiplicity of choices, as if he or she stands before a multiplicity
of realities: that which was once perceived as a whole must now be seen as a system of
variables. And it is the same fragmentation that the individual is exposed to. In his
studies of Musil and The Man without Qualities, sociologist P. L. Berger (1988) has
demonstrated what consequences Musil’s position has for the modern subject:
It is becoming y increasingly more difficult to regard the ‘‘I’’ as the center of a single individual’s
actions. Instead, these actions are now regarded as events, which, without the doing of the
individual, befalls him, and which may be explained by either exterior (social) or interior (organic
or psychic) causes. The Cartesian ‘‘I,’’ that used to proclaim its ‘‘cogito ergo sum,’’ has been
Innovation and Organization 77
In The Man without Qualities, Musil depicts the ideological confusion and collapse
that characterized the final years of the Austro-Hungarian double-monarchy from
1913 to 1914. A lack of substance and overall vision marked the picture of the times,
the officer and the civil servant are stuck in their ‘‘partial values’’ and have lost their
sense of the possible. The novel’s main character — Ulrich, the man without qualities
— should not be seen as without character; he is above all the man of possibilities.
He is not immobilized in a determined role but, rather, he is open and creative, he sees
reality as hypothesis, he lives in the conjunctive, he experiments.7
But if there is such a thing as a sense of reality — and no one will doubt that it has its raison
d’eˆtre — then there must also be something that one can call a sense of possibility. Anyone
possessing it does not say, for instance: Here this or that has happened, will happen, must
happen. He uses his imagination and says: here such and such might, should or ought to happen.
And if he is told that something is the way it is, then he thinks: Well, it could probably just as
easily be some other way. (Musil, 1979, p. 12, vol. 1)
It is the next step’s ethics that Musil aspires to: an exacting way to be a person. One
might also use the term ‘‘self-creating,’’ where the direction is not fixed. Not the
strategic, but the tactical is in focus. This sense of the possible, moreover, runs
strongly in German romantics like Fichte and Schelling who operate with their
indeterminate person who bears the mark of being ‘‘not yet’’; or the art which in a
manner of speaking approaches the ‘‘uncreated creating.’’
In the same way as Musil, Luhmann also takes to task all forms of substance
thinking. As the central character in Musil’s novel Ulrich says ‘‘The world has not
only one meaning, it has uncountable meanings’’ and ‘‘the one binding truth
disintegrates into hundreds of relative truths,’’ it is therefore natural that
mathematics’ absence of substance becomes the yardstick for Musil’s central
character: ‘‘One and the same case has a hundred sides, each side a hundred
relationships.’’ In the same way mathematics (especially that of Spencer Brown) takes
a central position in Luhmann’s system theory, and from the standpoint of
Whitehead and the pragmatists Luhmann also rejects the substance school of
thought. When Whitehead in Process and Reality (1978) seeks to understand what
‘‘process’’ is, he rejects Kant’s substance-oriented logic that indeed stands as a special
instance of one-sided predicate logic. Process — Whitehead says — can only be
understood from ambiguous predicates. An ‘‘event’’ — or that something happens —
is without substance, and a subject–predicate logic must therefore be undergone
because it is one-sided, i.e., the notion that we are dealing with an unalterable subject
that initiates changes must be abandoned.
7
The story tells that Ulrich’s interest for technological innovations almost became an ideological fashion:
‘‘From the first moment when he entered the engineering lecture room, Ulrich was feverishly biased. What
does one still want with the Apollo Belvedere when one has the new lines of a turbo-dynamo or the
smoothly gliding movements of a steam engine’s pistons before one’s eyes?’’ (Musil, 1979, p. 37, vol. 1).
78 Tore Bakken et al.
It is precisely this process aspect that Karl Weick wishes to clarify, and this with a
view toward organizations. Like the autopoiesis concept, he seeks to show how
organizations operate, and he does this by seeing evolutionary concepts such as
variation, selection, and retention as processes. Instead of using the word variation,
Weick applies the term enactment8 to describe the active role the organization plays in
the production (Gestaltung) of its relevant environment. Organizational processes are
thus the distinctive feature of being mutually influenceable (interdependent) as well as
‘‘loosely coupled.’’9 For
multiple realities, in turn, cause loosely coupled systems because individuals share few variables,
share weak variables, and differ in their perceptions of covariation among these variables. The
existence of multiple realities is not just a byproduct of enactment: it is the major consequence of
bounded rationality. (Weick, 1982, p. 385)
With the term enactment one can now describe organizational innovations and
their processes in a more satisfactory way because endogenous and exogenous factors
can be thematized together. An innovation now becomes understood as an active
‘‘production,’’ i.e., as an organizational activity where the surrounding, partly
uncertain environment (e.g., the wants of the customer) gives a ‘‘new’’ Gestalt. With a
‘‘product innovation’’ an organization ‘‘produces’’ its customer base anew. A novelty
in the form of an innovation is thus not to be seen as a reaction of an organization to a
demand from the environment, but rather an active variation. An enacting
organization can choose that which it deems meaningful to continue actively to
constitute relevant environments like before, or it can choose to do things differently.
It can thus profit an organization to increase the variation by ‘‘producing’’ its
environments (one acts differently from before), or by reducing its environment (one
acts exactly like before). This is what Weick (1969) calls doing choice: ‘‘Knowing what
I know now, should I act differently?’’ (p. 60). This means that relevant environments
are called in a different way in an innovation process. Different departments in an
organization can thus have a different influence on this active constitution. The
concept ‘‘loose coupling’’ clarifies the simultaneous presence of certain, unambiguous,
rationally plannable and the more uncertain, ambiguous, non-rationally plannable
aspects in relation to innovation processes. The concept points out that many parts of
an organization operate simultaneously according to set plans, but that others do not,
and this accentuates that these parts can develop a separate identity from the
organization. Certain parts of the organization can be described as loosely coupled
with other parts, and can thus adjust locally to environments without affecting the
rest of the organization.
8
The notion has a strong biological basis and is different from symbolism and ‘‘connectionism’’ because it is
a more radical interpretation of the network models: The autonomy of the construction of the system and
its order and information is emphasized. It is underlined that the development of complex systems is open
where the organisms in a way enact their complementary environment. This point is also made in the theory
of autopoietic systems.
9
Cf. Weick (1982): ‘‘Loose coupling exists if (1) A affects B (suddenly rather than continuously), (2)
occasionally (rather than constantly), (3) negligibly (rather than significantly), (4) indirect (rather than
direct), and (5) eventually (rather than immediately)’’ (p. 380).
Innovation and Organization 79
Precisely this process aspect, if one re-specified it for systems in general, is also
what Luhmann is at pains to describe. But processes here are to be understood as
irreversible events.10 Process is concerned with likely/unlikely events, something
which makes the concept sensitive to all forms of risk problematic (cf. Luhmann,
1991). It becomes a way of dealing with complexity: the system can orientate itself to
differences and link their operations to them. This is temporalizing of complexity, i.e.,
the system’s elements cannot base themselves on repetition or routine, but must
change according to internal and external demands. It must be prepared for the fact
that something else is joining the fray (connection operations). The system can only
actualize ‘‘instant’’ connections, and from moment to moment it produces new
situations where repetition or change is possible. This is autopoietic systems’ way of
functioning: systems of this kind are therefore inherently restless, they are exposed to
an ‘‘endogenous’’ dynamic, and they compel themselves constantly to learn structures
that are compatible with this. In this respect one can say that autopoietic systems are
produced from unstable elements that vary over short periods of time, or indeed that
when it comes to events they have no independent duration, but already within
their own production vanish. In this way systems with temporalized complexity are
continually left to disintegrate.
Based on this experience of temporalized complexity, Luhmann made the meaning
concept the very basis of his system theory. That which constitutes meaning for
Luhmann is not, as for Weber, the significance in events, but rather a difference,
namely the difference between actuality and possibility. Luhmann thinks that an
autopoietic, self-referential system cannot have a stabile basis, and the system’s
operations can therefore only be a departure point for further operations. Every
operation consists in ‘‘drawing a distinction’’ and calls forth further markings of one
side of the distinction. At a particular instant something is indicated (actualized) while
at the same time pointing toward other meanings. That which at a given instant in
time prevails is thus unstable, it will break off or collapse, and it compels us to select
from the domain of the possible something new that can be actualized in the next
instant if the autopoiesis is to be continued. Meaning is thus a continual actualization
of possibilities. The measure for what is to be done is no longer the perfect, but the
possible. At any given moment there is a reference to a horizon or a surplus of further
possibilities, which cannot all be actualized at the next moment. Consequently one
must select. The connections that are not selected, however, do not disappear, but are
maintained as possibilities, i.e., they can be actualized at a later moment. This means
that a system can permit errors, go back to the point of departure, and choose
another road. The possibilities are after all inexhaustible. In this way the unstable and
the restless compel a system into continual self-change, i.e., to become self-innovatory.
According to this difference between actuality and possibility one can say with
Luhmann that ‘‘everything could be different.’’ But at the same time, no matter how
much one chooses, the together-possible (co-possible) will always exceed the together-
realizable (compatible). The romantic ‘‘everything could be different’’ can quickly
10
This contrary to structures which make time reversible.
80 Tore Bakken et al.
become melancholy resignation: ‘‘and almost nothing can I change’’ (Luhmann, 1971,
p. 44).11 But this is the modern risk society’s paradox: the difficulty of changing
something is due quite simply to the tempo and the dynamics of risk society. The risk
society is driven forward at such a high tempo that it has become unchangeable in a
way fraught with risk. We know this from organizational theory: no programming of
larger changes is in a position to relate to the incalculable complexity the programs
are involved in (cf. Luhmann, 2000, 330ff.).
Innovations are possible only to the extent that decision-making processes are able
to come up with implementable alternatives. The first step toward such a shift is to
study organizational changes from an evolutionary theoretical perspective rather
than from a planning theoretical perspective. This is not only because an evolutionary
theoretical approach operates with a more complex use of concepts, but also
because value concepts can be studied in their functional context through such a
perspective, i.e., studied on the level of self-descriptions of organizational systems.
Value concepts cannot give a complete picture of the actual operations or structures
of an organization. With Brunsson (2002) one could say that such concepts organize
political modes of speaking, but by no means describe the reality they deal with.
For Luhmann (1996) it is a question not only of seeing decisions as a choice from
within a range of alternatives, but also to broaden that perspective. In addition one
must ask how alternatives are arrived at in a world that is as it is, i.e., how is it
possible through a decision to bring about something that was not there before in a
world where what happens, happens, and where what does not, does not? Luhmann’s
concern consists in relating decision making to the time dimension. This means that
we need to assume a difference between past and future, and this at the same time
makes a difference between what is past and what is future. The decision causes this
difference as a result of the decision to turn out differently from how it would have
turned out if one had not taken a decision, it ‘‘causes,’’ or better: the change of the
difference is attributed to the decision, however, much the actual inscrutabilities and
the complex causal relationships race by. The decision thus makes itself visible by
attributing to itself, and in this way past and present can be treated equally. States
that are as they are, or which are becoming that which they are becoming to be, break
11
This back and forth between a romantic sense of possibilities and a more skeptical orientation is typical in
Luhmann’s authorship.
12
Values are contrafactic stabilized expectations. According to Luhmann they are (1973, p. 40) not very
fruitful in modern organizations. Values prevent us from grasping the cognitive aspects in organizations.
Innovation and Organization 81
up into differences. And this makes possible a reentry of time into time.13 The past
and future time horizons thus become seen in relation to each other and therefore
integrated. It matters not in the slightest that a decision cannot change the past or
determine the future. And despite this there begins, thanks to the reentry of time into
time with every decision, a new history. The future thus for us becomes unknown. But
the future’s unknowability is at the same time the greatest resource for the decision,
and indeed a precondition for organizations as such, for decisions base themselves on
the fact that no one knows what the future holds. One can set goals because one does
not know how things will look in relation to what the future will bring. Of course we
can come up with some relatively stable assumptions; e.g., that the Oslo fjord will
continue to exist. But the existence or nonexistence of the Oslo fjord is not really an
area for decisions. Although when one plans a tunnel under the Oslo fjord one
produces a niche for the future’s unknowability, and again thanks to the reentry of
time into time one is able certainly to take decisions. The more the society is prepared
to recognize niches of this kind, the more obvious it will become in the future that
decisions become more transitory. But when with these decisions there always begins
a new history, the decision perspectives intensify the unknown aspect of decisions.
As opposed to natural and cultural philosophers since Bacon and Vico, history is just
as fully unpredictable because it is created by people. And it is on the cards that this
happens not only through information, but also through imagination. Decisions
make themselves dependent on surprises because they are surprises. They can change
expectations by being contra-inductive — i.e., innovative.
Again: ‘‘Everything could be different’’ and ‘‘almost nothing can I change.’’ This
paradox is typical for organizational systems, for organizational systems show more
than other systems this two-sidedness of redundancy and variation (cf. Ahlemeyer,
1997; Ashby, 1956; Bateson, 1972).
To obey a rule, to make a report, to give an order, to play a game of chess, are customs (uses,
institutions). To understand a sentence means to understand a language. To understand a
language means to be master of a technique. (y199, p. 81)
13
The notion of reentry (Spencer Brown) is based on the semantics of deparadoxation.
82 Tore Bakken et al.
14
Cf. Luhmann (2000): ‘‘But they [the decision premises] focalize the communication on the differences that
have been established in the decision premises. This makes it probable that one will observe future decisions
with regard to these preset premises, from a viewpoint of consideration/non-consideration and conformity/
deviance, rather than unfolding the complete complexity of the situation anew every single time’’ (p. 224,
own translation).
Innovation and Organization 83
Even so, we know that organizations innovate. When organizations are regarded as
social systems with recursive decision production, it means not merely programming,
routine and selective rigidity, but it also accentuates changes in such decision
premises. Programs can be changed; expertise and communication channels can
be worked out anew; job positions can be redefined or done away with, etc. The
innovation rate in the modern economy is intense when one thinks of new products
and the increasing production diversity (Baecker, 2007). Indeed, today the supply rate
has increased so markedly that demand cannot keep pace with it. US studies show
that customers can no longer tell the difference between new and the old cars when
they attend motor shows. In this way the time one has to adapt to new products grows
shorter and shorter. The same is true for computers. Too much re-adaptation leads to
time loss for computer users. Experience and familiarity are also benefits. This is why
big companies have become so concerned with ‘‘compatibility’’ between product
generations. New concepts must to a large degree base themselves on the old, and
in this way one meets a problem in ‘‘pre-announcement.’’ Trade fairs not only exhibit
new products, but also announce the products for the next generation that will
contain all the improvements not yet incorporated into the new products being
launched. In this way novelty is imagined into the future, something that can have
unfortunate consequences: this form of temporalizing leads to the collapse of market
mechanisms through the distortion of purchasing decisions.15
This dynamic stems not first and foremost from the organizational systems but
rather from the functional systems, especially the economic functional system that is
set up for the feedback mechanism profit and loss, and where the market through the
15
It is this kind of problem which has forced Japanese trade authorities to require the producers to lengthen
the production cycles among the chip producers, and in this way lower the speed of innovation. But of
course, this demands institutional regulations (more distinct rules of the game). But this is not easy to gain.
84 Tore Bakken et al.
prices becomes a highly effective mechanism for reducing the complexity when it
comes to resources, motivation, needs, decisions, and products for a manageable
format. The market forms an environment that provides effective mechanisms for
evaluating results and decisions and for learning from mistakes. Organizations see
and calculate connections for their own operations and utilize the complexity of the
environment to transform uncertainty into risk in the form of investment risk,
purchasing risk, payment risk, etc.16
This is also expressed internally in organizations through change projects
running concurrently, which every day challenge secure structures and demand new
orientations. This holds out the prospect of an exchange of patterns and structures
that compels organizational members to accept more and more of the new. But how
can this be understood through system theory terms?
To understand innovations in organizations it is above all important to connect
with evolution theory. For Luhmann (1997), the different components of evolution
(variation, selection, re-stabilization) become linked to the different components of
autopoiesis (p. 413ff.). The term variation takes in ‘‘coincidence,’’ ‘‘innovation,’’ and
‘‘enactment’’; selection refers to ‘‘elimination,’’ ‘‘adaptation,’’ and ‘‘rebuilding’’; and
re-stabilization encompasses ‘‘reproduction’’ and ‘‘institutionalization.’’
Evolutionary variation challenges the action theorist’s conception that change is
linked to the actions of strongly motivated individuals. Variation for Luhmann means
that the elements within a system vary, i.e., it is the communications that vary.
Variation consists of a divergent reproduction of the elements through the system’s
elements, i.e., in an unexpected, surprising communication. This also takes in
‘‘coincidence,’’ which here should not be seen as causal-theoretical, or, as for Hegel,
as a counter-term for necessity. For Luhmann, coincidence should be seen as a form
of coherence between system and environment that evades the system’s synchroniza-
tion (control, systematization). No system — or organizational system — can control
for all causalities. In Luhmann’s sense coincidences should be seen as a system’s
ability to take advantage of ‘‘events’’ that are not produced by the system itself
(i.e., not in autopoiesis’s network). They are at once possibilities, chances, and
dangers. They can be constructive while also destructive. They can turn out to be
‘‘order from noise,’’ or as structural connections that channel irritations from the
environment. Even so, variation falls into place under the system’s autopoiesis; it
ensures that the communication continues — if one with freer connective possibilities
and an inherent tendency toward conflict. But like all operative elements in dynamic
systems, communication and deviant communication (innovation) are situationally
determined and rapidly lose their meaning. The concept of variation therefore gives
no answer to the question of why great epoch-making ideas and inventions arise,
for evolution makes no great leaps, even though in retrospect others might observe it
as doing so. The social reality is extremely conservative and does not negate the
16
Cf. Luhmann (2000): ‘‘The function systems proceed from inclusion, exclusion is unplanned and, so to
speak, just happens. In organizations, the opposite is the case. Here, everyone is excluded — membership is
not a natural right — since inclusion must proceed in a highly selective fashion’’ (p. 392, own translation).
Innovation and Organization 85
existing in favor of the unknown, if chances for consensus are not proven or even
tested for the same situation. A variation can therefore make it through in favorable
cases, but it is not possible to pose the question of why things should be so and not
otherwise.
Re-stabilization. This refers to the state of the evolutionary system after a positive
or negatively charged selection. It is a question of durability in a social system
differentiation. The innovated structures must at a point in time fit in with the system
and be made compatible with their relationships with their environment. In 1789 the
Paris uprising was observed as ‘‘revolution’’ and this later had consequences for the
development toward a representative democracy. The codification of the legal system,
the liberation of an economic system with its own forces, the secularization of
religion, and the privatization of the extended family can be seen as re-stabilizations
of the revolutionary innovations. But revolution can also be harnessed negatively, as
in Prussia where one ended up with a re-stabilization of a culture-state program for
schools and colleges. Therefore, variations can vanish unobserved, while selections
are normally held on to in a system memory. However, the point here is that a
rejected variation in the long run can have greater effect than an implemented
innovation. In any event, the term re-stabilization represents sequences for building in
structural changes in a structurally determined operational system, and this bases
itself on an insight that this happens through variations and selections, and always
through the system’s own operations. In every case selection (positively or negatively)
leads to an increase in the system’s complexity and therefore the system must react
with re-stabilizations. Banks are examples of re-stabilizations of the money economy,
and burst the old maxim about reciprocity. The ‘‘new-time’’ state earns its
re-stabilization from a long-prepared political centralization. Re-stabilization always
carries with it a stability principle, but always with an additional solution: that which
will abolish atomic power stations are confronted by the question: How will we
generate power another way? With the transition from the re-stabilization function to
functional systems stability becomes a dynamic principle and indirectly a central
stimulus for variation. Functional systems remain prepared for change under
the condition of functional equivalence and a net superiority of possible new forms.
86 Tore Bakken et al.
Also, while they do not themselves initiate innovations, they have a high potential for
reacting to innovations with innovations. This applies to a greater extent if within
functional systems organizations are formed that can change themselves and their
decision-making practices through decisions. Luhmann (1997) describes this in the
following way:
As early as the pronounced stratified order of the middle ages corporations, such as churches,
monasteries, orders, cities, guilds and universities, assumed innovative functions — at first
because they, thanks to their corporative stability, could maintain themselves as member
communities outside the system of estate. Society had already started experimenting with forms
of dynamic stability, that were not anticipated in its form of differentiation. But exactly this
division of labor between the corporations also implied that their potential for innovation
remained restricted to themselves and, in the transition to the modern world, they were perceived
as rather rigid and inflexible [institutions]. The system of estates and corporations was gradually
replaced by the order of organizations in functional systems; only then the primary societal
subsystems were given the opportunity to develop a conditioned dynamic stability. (p. 493, own
translation).
In the course of this evolutionary process the functional systems realign their mode
of selection more and more toward fundamentally unstable criteria. Provided the
channeling of the re-stabilization takes place within organizations, i.e., through
decisions, resistance toward novelty will also be produced.
6 Conclusion
In today’s climate of forced dynamic and social change there runs the classic solution
pattern in organizations, namely the reproduction of ‘‘more of the same’’ problem:
the same — that is to say the repetition of a pattern — at the same time embraces
change of the pattern by allowing for the growth of new emergent structures.
Innovations are thus important even though they have their limitations (seen from
a system-theoretical perspective). But innovation can only assume the form of being a
disturbance of experience and routine, and in this sense innovation is unlikely
(Luhmann, 2000, p. 162).
Proposals for innovations almost by necessity invoke conflicts, and it would be
foolish of organizational sociologists not to emphasize the importance of precisely
these aspects of the reality. Every innovation or reform is about ‘‘creating or utilizing
previously unrecognized social spaces’’ the organizational theorist P. Herbst observed
(cf. Herbst, 1976, p. 48). These empty spaces will no doubt be formally unregulated
areas in an organization. But we can ask with Luhmann if such empty spaces still
exist, and if all reform and innovatory work in organizations to an increasingly
greater degree is no more than the entering of already occupied spaces. It is therefore
perhaps time to reflect upon the question: How will organizations in the future endure
with their self-inflicted innovations?
In a follow-up chapter in this volume, we carry out a deeper exploration of the
topic launched in this chapter by setting the theme of autopoietic understanding of
Innovation and Organization 87
References
Ahlemeyer, H. W. (1997). Mehr des Neuen statt mehr desselben?. Vom Umgang mit Innovation
in organisierten Sozialsystemen. Sozialwissenschaften und Berufspraxis, SUB, Heft 4,
pp. 329–340.
Ashby, W. R. (1956). An introduction to cybernetics. London: Chapman and Hall.
Baecker, D. (2000). The distinction theoretical shift in systems theory. An interview with Dirk
Baecker. Distinktion (1), 123–131.
Baecker, D. (2007). Studien zur nächsten Gesellschaft. Frankfurt am Main: Suhrkamp.
Bakken, T. & T. Hernes. (Eds). (2003). Autopoietic organization theory. Drawing on Niklas
Luhmann’s systems theory. Copenhagen: Liber, Copenhagen Business School Press.
Bakken, T., & Hernes, T. (2008). Autopoiesis. In: R. Thorpe & R. Holt (Eds), Sage dictionary
of qualitative management research (pp. 33–36). London: Sage Publications.
Bateson, G. (1972). Steps to an ecology of mind. New York: Ballantine Books.
Beckert, J. (1997). Grenzen des Marktes: Die sozialen Grundlagen wirtschaftlicher Effizienz.
Frankfurt am Main: Campus Verlag.
Berger, P. L. (1988). Robert Musil und die Errettung des Ich. Zeitschrift für Soziologie, 17(2),
132–142.
Brunsson, N. (2002). The organization of hypocrisy: Talk, decisions and actions in organizations.
Oslo: Abstrakt Forlag.
Giddens, A. (1990). The consequences of modernity. Cambridge: Polity Press.
Günther, G. (1980). Die historische Kategorie des Neuen. In: G. Günther (Ed.), Beiträge zur
Grundlegung einer operationsfähigen Dialektik (Vol. 3, pp. 183–211). Hamburg: Felix Meiner
Verlag.
Herbst, P. (1976). Alternatives to hierarchies. Leiden: Martinus Nijhoff.
Joas, H. (1992). Die Kreativität des Handelns. Frankfurt am Main: Suhrkamp.
Krohn, W. & G. Krüppers. (Eds). (1992). Emergenz. Die Enstehung von Ordnung, Organisation
und Bedeutung. Frankfurt am Main: Suhrkamp.
Luhmann, N. (1971). Politische Planung. Wiesbaden: Westdeutscher Verlag.
Luhmann, N. (1973). Zweckbegriff und Systemrationalität. Frankfurt am Main: Suhrkamp.
Luhmann, N. (1975). Soziologische Aufklärung (Vol. 2). Wiesbaden: Westdeutscher Verlag.
Luhmann, N. (1978). Organisation und Entscheidung. Wiesbaden: Westdeutscher Verlag.
Luhmann, N. (1995). Social systems. Palo Alto, CA: Stanford University Press.
Luhmann, N. (1988). Die Wirtschaft der Gesellschaft. Frankfurt am Main: Suhrkamp.
Luhmann, N. (1991). Soziologie des Risikos. Berlin, NY: Walter de Gruyter.
Luhmann, N. (1995). Sich im Undurchschaubaren bewegen. Zur Veränderungsdynamik
hochentwickelter Gesellschaften. In: R. Grossmann, E. Krainz, & M. Oswald (Eds),
Symposium veränderungen in organisationen, Gabler Verlag, Wiesbaden.
Luhmann, N. (1996). Entscheidungen in der Informationsgesellschaft. Manuscript, Berlin.
Luhmann, N. (1997). Die Gesellschaft der Gesellschaft. Frankfurt am Main: Suhrkamp.
Luhmann, N. (2000). Organisation und Entscheidung. Wiesbaden: Westdeutscher Verlag.
Maturana, H., Varela, F., & Uribe, R. (1982). Autopoiese: die Organisation lebender Systeme,
ihre nähere bestimmung und ein Modell. In: H. Maturana (Ed.), Erkennen: Die Organisation
und Verkörperung von Wirklichkeit. Braunschweig/Wiesbaden: Friedr. Vieweg und Sohn.
88 Tore Bakken et al.
Musil, R. (1979). The man without qualities. London: Secker and Warburg.
Varela, F. (1979). Principals of biological autonomy. New York, NY: North Holland.
Watzlawick, P. (1985). Management oder — Konstruktion von Wirklichkeit. In: G. Probst,
H. Siegwart (Eds), Integriertes management. Bausteine des systemorientierten Managements
(pp. 365–376). Stuttgart.
Weick, K. (1969). The social psychology of organizing. Reading, MA: Addison-Wesley.
Weick, K. (1982). Management of organizational change among loosely coupled elements. In:
P. S. Goodman, et al. (Eds), Change in organizations: New perspectives in theory, research,
and practice. San Francisco, CA: Jossey-Bass.
Whitehead, A. N. (1978). Process and reality. An essay in cosmology. New York, NY: Free
Press.
Wittgenstein, L. (1967). Philosophical investigations. Oxford: Basil Blackwell.
Chapter 5
1 Introduction
Ellis, 2006; Goldspink & Kay, 2007, 2008). How they are distinct is made clear from
the application of autopoietic theory.
We use autopoiesis to better understand the reciprocal interplay between the micro
behavior of agents on the one hand, and the resulting pattern of behaviors at the
macro-level on the other. These emergent macro-structures are somewhat robust
patterns associated with particular groups of agents. They have traditionally been
referred to using terms like ‘‘institutions,’’ ‘‘norms,’’ and, the focus of our interest
here, ‘‘organizations.’’ These patterns do not result from upward causation only, as is
the case with particle interaction, for example, but rather they include a downward
causal path: constraining the scope of action of the very agents which give rise to
them. In other words, macro-level patterns have micro level effects. This has been
referred to as ‘‘immergence’’ (Castelfranchi, 1998).
Fuchs and Hofkirchner (2005, p. 33) take an emergentist perspective in his
classification of alternative approaches to the micro–macro relationship within social
theory distinguishing between individualist approaches such as agency theory,
‘‘sociologism’’ including systems theories which he argues are top-down deterministic,
dualistic approaches and dialectical approaches.
Most social theory falls into one or other of the first two categories. These theories
work with a dichotomous view of macro and micro: focusing attention on just one level
or the other and failing to address their relationship. This is consistent with Weik’s
(2006) view. She has argued that social theory can be divided into three categories:
dualist, duality, and theories which avoid or deny the separability of micro and macro.
In social science the micro–macro problem is also referred to as the problem of
structure and agency. Structure emerges from the agency of social agents and at the
same time constrains it, but neither determines the other. Weik argues that in most
social theory this micro level capacity for partial independence is commonly attributed
to intention or purpose — the debate being about how ‘‘free’’ agents are to exercise these
with respect to structure. Structure implies a repetitive relation between two or more
individuals with different theorists positing different dimensions to that relation — viz.
shared knowledge, functions, routines, constraints reciprocal expectations, power or
force, rational choice, identity need, habit or rule following. ‘‘Some of these definitions
overlap, some have been taken together to form several levels of structure embedded
in one another y and some are, of course, contradictory’’ (Weik, 2006, p. 3). Another
area of confusion relates to where these structures are considered to reside. ‘‘The most
prominent candidate is, of course the individual mind’’; however, alternatives include
the brain, body, human essence, act (habitus), and language. Finally, the mode of
influence between levels is often unspecified: is it causal or something else?
In short, social theory has attempted to resolve the problem using a wide range of
conflicting theoretical stances, none of which have proven satisfactory. Those that
come closest are in Fuchs dialectical category. These too are diverse, including but
not restricted to Marxist dialectical materialism, the critical theory of Habermas, the
critical realism of Bhaskar, and the structuration theory of Gidden. A few incorporate
the theory of autopoiesis with some drawing directly on Maturana and Varela’s
original work, whilst others have adopted a more Luhmannian perspective. What
then does it offer?
Autopoiesis and Organizations 91
2 Autopoiesis
subset of behaviors which we refer to as the drivers of the change dynamics of the
organization. These are the relatively small number of behaviors (including linguistic
utterances) which generate and maintain a particular attractor. These attractors are
of course the cultural norms and institutions which combine to support the higher
order attractor, which we refer to as ‘‘the organization.’’
For many managers, detecting these drivers is an intuitive process or one based
on experience; however, more systematic research methods may also be used to
surface them. Once the key drivers influencing such patterns have been identified, the
manager can take action to disrupt those that appear to support undesired stability
and/or stimulate those that might support desired change.
While developments in complex systems and social simulation have advanced our
ability to map complex dynamics, this has generally been in systems where agents
have limited cognitive capacity (Sawyer, 2003, 2005). While developments in these
techniques hold promise for the future, at the current time there are few techniques
that support our understanding of dynamics that result from the reflexive emergence
associated with human agents (Goldspink & Kay, 2007, 2008). It is therefore
necessary to use more conventional research methods to gain insights into the
operations of organizations. A range of methods have been developed for the study of
linguistic interaction. Some focus on mapping the denotative content of utterances,
while others are concerned with the illocutionary or pragmatic force of language
as a basis for direct influence (Searle, 1969; Habermas, 1976).1 In the following two
cases we illustrate techniques that can be employed to surface the change dynamics
from which the organization emerges, focusing primarily on alternative methods for
linguistic analysis.
1
We are examining this in another case study on normative self-organization in the Wikipedia; see
Goldspink (2007).
Autopoiesis and Organizations 95
4.1 Case Study One: Financial Services Trust and Innovation Potential
The research context was a small business unit within a large financial services
institution. The business unit in which the case study was conducted was led
by a general manager and a six heads of departments, each with multiple direct
reports in a strongly hierarchical structure. Each department in the business unit
was responsible for the management of different outsourcing arrangements
and contracts with suppliers. The leadership team was concerned at the low level
of collaboration between the different departments and the effect this had on
innovation and the quality of decisions making. In response they designed a small
intervention to facilitate collaboration across the business unit. The task involved
bringing together senior managers from the different departments to solve a set
problem.
The managers were asked to establish a taxonomy against which the top 100
suppliers could be categorized, according to whether they were strategic (bringing
new capability), aligned (providing improved capability to an existing strategy), or
standard (providing supply to a nonstrategic function). It was intended that the
taxonomy would form the basis for new relationship management models.
Participation in the project was voluntary and undirected: those who volunteered
to participate were expected to self-organize in order to clarify and generate strategies
to address the problem. The voluntary nature of participation resulted in only about
half of the potential participants taking part.
The outcome of the project was seen by most people associated with it, including
the general manager, to be unsatisfactory, both in terms of the proposed solution and
the collaboration achieved. The group working on the project fragmented into two
subgroups with each advocating incompatible solutions. The fact that such a
relatively simple task could not be completed came as a shock to the general manager,
who suspected there were deeper issues at play. We were asked, as people independent
of the institution, to explore the reasons why the exercise failed. Our brief was
to understand the factors affecting the group’s ability to collaborate: why couldn’t
a group of intelligent, experienced managers, organize themselves to complete a
relatively simply problem-solving activity?
4.1.1 Methodology. Eleven senior managers took part in our study drawn from a
group of 18 possible participants. Participants were selected at random from a list of
all the senior managers. Six out of the eleven interviewees had taken part in the
exercise, whilst the others, although aware of it, had either specifically chosen not to
be involved, or had sent a representative from their team.
We sought to gain an understanding of the recent history of the interactions, the
environment, and how both individual (micro) sense-making and (macro) institu-
tional structures combined to limit collaboration. To achieve this, a methodology
which combined narrative and repertory grid methods was employed. Both narratives
and the repertory grids were collected in a single interview which lasted on average
about one and half hours.
96 Chris Goldspink and Robert Kay
It has, however, been argued to be at the core of the functioning of human meaning
making — the narrative mode of thought (Bruner, 1991a; Dautenhahn, 2002). Bruner
observes that there is a sense in which:
y narrative, rather than referring to ‘reality’ may in fact create or constitute it y . (1991, p. 13)
From this perspective, narrative data provides an account both of how people
interpret past events and how those interpretations play a role in embedding
particular ways of thinking and knowing in the culture of the organization — how
they come to be constitutive of the organizational reality. When we construct
narratives we place ourselves as a character, even if it is one of innocent bystander.
A narrative can reveal a lot about the part and future role an actor may play. We can
and do of course revise our narratives. We will, however, be very reluctant to change
the central character — ourselves: the grand narrative that is our sense of identity.
Narrative data then provides insight into the relationship between events — i.e., how
the observer/participant sees the ways events are linked in time. More than this, and
significantly for this study, it captures individual and collective accounts of the
interplay between individual behavior and collective consequences. These accounts
play a part in the maintenance of existing order and/or to reflect the basis for change
in established routines by revealing compartmentalization in the linguistic domains.
In this case study a very simple narrative collection was undertaken. This involved
asking participants to recall two recent collaboration experiences with which they
had been involved within the institution: one a positive experience and the other
a negative experience. Not all participants were able to think of two stories that they
felt were worth telling, and as a result 14 stories were collected out of a possible 22.
The stories were analyzed with the participant at the time of the interview. Six key
events were selected that ‘‘stuck in their mind.’’ These events were equivalent to what
David Snowden (2000) would describe as an anecdote. Breaking the stories down into
anecdotes supported analysis of the stories as a whole, but also identified discrete
events for subsequent thematic analysis across narratives. Eighty-four separate
anecdotes were collected and clustered according to commonalities in their content,
i.e., common words, depiction of similar events, etc.
construal and forms a hierarchical system of more or less tightly held conceptual
distinctions which orientate behavior. Kelly saw this construct system as dynamic —
being constantly modified as the agent acts in the world and attempts to be effective
within it.
While a construct system is specific to the individual and forms the basis of that
individual’s agency, it is a product of his/her history of interaction in the current
and other social domains. Constructs low in the hierarchy have fewer dependent
connections with other constructs and can be surrendered or modified more readily
than those at the top of the hierarchy. Superordinate constructs form primary
orientating distinctions: they are associated with worldviews, and individuals will
generally be reluctant to change them as they have profound implications for the way
he/she sees and orientates him/herself in the world. Kelly (1963) argues that all social
processes necessarily involve the mutual construal of others construction and that
this gives rise to some commonality of construction (consensuality) in that domain of
interaction.
Repertory grid (Fransella et al., 2004; Jankowics, 2004) is one of a family of related
methods developed by Kelly and others to make Personal Construct Theory
operational. In the context of this case study, repertory grid offered a means for
mapping both individual (micro) and collective (macro) patterns of construal within
a particular social domain. Furthermore, grid analysis supports the development of
metrics, which allow some prediction of how willing or likely individuals would be to
change their construal and thus how responsive they may be to alternative change
interventions.
Repertory grids collect fine-grained data about individuals sense-making about
some target. While the data is fine grained it is also sharply focused, so the challenge
in using grid as a means for data collection is to ensure that the data converges well
onto the topic of inquiry. Critical here are the choice of items of experience (elements)
that will be used to ‘‘elicit’’ ‘‘constructs’’ and the focus question used during
elicitation (Jankowics, 2004). Elements need to be tangible items of experience (i.e.,
time-bound events, things, or people). For this exercise we chose to use relational
descriptors as prompts and to have the respondents supply specific people who
matched the descriptor.2 These people then became the elements in that respondent’s
grid. Each respondent would have different individuals, but individuals who were
selected against criteria were common to all respondents. Respondents were asked
to identify eight colleagues from within the senior manager team who matched the
following descriptions:
2
In a more recent related study, which focused on innovation rather than collaboration and trust as with the
case study reported here, in this latter case ‘‘innovation events’’ were taken from the narratives and used as
elements.
98 Chris Goldspink and Robert Kay
4.1.2 Analysis. All the people involved selected and described the same negative
experience — the exercise in generating collaboration discussed earlier. As might
be expected the narratives captured quite distinct and different accounts and
interpretations of events: unique personal histories of the shared experience. These
narratives provided anchoring events against which the individual sense-making
of the participants (as revealed by the repertory grids) could be interpreted. They
also revealed the wider environmental factors and historical sequence, as well as the
individuals reading of cultural rules, norms, and institutional practices, which they
believed influenced the outcome.
Grids were analyzed using the software package Idiogrid (Grice, 2002). Patterns in
the relationship between elements and constructs were examined using Principal
Component Analysis. This enabled us to identify, for each respondent, the type of
person he/she was likely to share information with compared to those with whom
he/she would be unlikely to share; what type of person he/she would trust compared
to those he/she would not trust etc. It also revealed the degree of association between
the element classes; if likelihood to ‘‘share information’’ was closely associated with
‘‘trust’’ or based on different factors in a relationship for example.
According to Kelly, a person’s construct system provides him/her with a basis for
hypothesizing about consequences of his/her and others’ actions. Tight construal
(as indicated by a high mean correlation between constructs in the grid) would suggest
that a respondent would have relatively unvarying predictions based on his/her
construal of a situation. In other words, the characteristics the respondent attributes
to individuals would, from his/her perspective, be expected to provide good
prediction of the collaborative behavior of others. Loose construal, by contrast,
Autopoiesis and Organizations 99
would suggest a person with more flexible views, someone open to surprise. Inferences
can therefore be drawn about a respondent’s openness to change. In addition, an
ordination score can be used to reveal the location of a construct within the
respondent’s construct hierarchy, with higher scores suggesting higher ordination or
more meaningful (and abstract) constructs (Landfield & Cannell, 1988). Individuals
are less likely to be willing to change higher order constructs as they have significant
implications for how they make sense of the world (Kelly, 1963; Bannister &
Fransella, 1989).
Combining the results. A comparative analysis of the results of the two data sets
was undertaken on two levels. Firstly, individual stories were mapped to individual
repertory grids. These two data sets revealed insight into which constructs in each
individuals meaning system primarily orientate their construal of events and guide
their action. Secondly, the narrative clusters emerging from the thematic analysis of
the stories were mapped to the output from the group grid analysis.
Usually repertory grid analysis is undertaken at the individual level; however,
in this instance we conducted a thematic analysis across the constructs of the entire
group (see Jankowics, 2004 for a systematic process for doing this). This analysis
provided insight into how each agent made sense of their situation and the degree
to which there were commonalities to this sense-making. Mapping these two
together revealed the areas of common construal around a distinct series of events.
It also means we could see the depth with which that construal is held and therefore
also which dimensions of the social system’s patterns can easily change, and those that
will not.
Observations. From the combined analysis it was possible to discern the primary
distinctions that orientated respondents toward one another and influenced their
willingness to collaborate. These distinctions appeared to form the basis for the
creation of subgroups within the broader team, where people of like characteristics
have a much higher propensity to trust and collaborate with each other rather than
those they perceived as being different. The combination of depth with which these
constructs were held and the degree to which they were shared across the group
strongly drove the eventual outcome of the particular activity we studied, i.e., the
group that was supposed to be collaborating split to create subgroups closely aligned
to the constructs described above.
What is interesting here is that overtly all the participants wanted to collaborate,
and indeed initially did collaborate around the problem they had been set, thus
creating a new pattern of interaction that had not existed before. However, over a
relatively short period, this new pattern broke down with a slightly modified version
of the preexisting pattern of interaction reemerging. In the evidence collected there
is a clear explanation for this. Individuals were construed through established
constructs, and these influenced subsequent behavior. As there was nothing in the
design of the intervention which was directed at challenging or disrupting the existing
ways of making sense of the situation and, in particular, nothing powerful enough to
100 Chris Goldspink and Robert Kay
compel the need to reconsider deeply held constructs, no change was achieved. On the
contrary, the existing patterns reappeared in a slightly modified form.
4.1.3 Conclusion on Case Study One. This case selected for this research centered
on an intervention designed to address a limited capacity for innovation in a senior
management team — i.e., a perceived inability for managers to bring new ideas,
understandings, and capabilities to challenging situations. We have examined the
reasons for the failure of this intervention by seeking to better understand the way in
which individuals contribute to maintaining current patterns in the organization and
how the intervention failed to address these. This represented a move away from
approaches which treat ‘‘organizations’’ in a reified way to a complex systems view
focusing in particular on understanding the interplay between macro and micro levels.
The intervention initially used to try to build collaboration in this work unit,
assumed that collaboration was not occurring due to formal structural inhibitors
(institutional silos and or physical distance) and/or lack of opportunity. It was
anticipated that providing different people from different backgrounds with the
opportunity to work on a common project would be all that was required to overcome
the problem of lack of collaboration. This proved too simplistic as it failed to identify
the way in which individual and collective sense-making around who and when to share
information or trust had developed within the organization and had come to constrain
the range and type of relationships members were prepared to participate in.
The data gathered using both narrative and repertory grid methods revealed a more
complex picture. The senior management group was shown to have formed a set of
ways of interpreting their environment which limited their willingness to engage on
the basis of three dimensions of relationship. These were not related to the formal
structure or to physical proximity directly (although these would have influenced
the formation and maintenance of the dimensions found) but were culturally
stable dimensions which had become self-maintaining attractors. This combined
with a pattern of tight construal contributed to a very stable system whereby
individuals sense-making reinforced cultural patterns which shaped interaction so
as to reinforce individuals sense-making in a manner which restricted the possibility
of change.
This analysis supported the argument that organizational behavior is a complex
product of the interplay between individual agency and institutional structure and
that these come together to form phenomenal domains. We have argued that unless
insights can be gained into the drivers which support attractors in these domains
intervention is likely to be ineffective. We have shown how conventional methods, in
this case narrative and repertory grid technique, may be combined to help locate these
drivers in the linguistic domain pertaining to the particular context.
How people influence one another and converge on common expected patterns of
behavior
The emergence and role of social constructs which have become somewhat ‘‘reified’’
within a particular consensual domain (rules and explicit norms) in an open
volunteer community where there is little to no hierarchy and limited capacity for
formal sanction and which must continue to attract and retain agents if it is to
survive (is in a sense self-maintaining and producing)
How these norms and rules are generated and maintained within behavioral and
linguistic domains
The relationship between goal, technical artifacts, and social structures and the
exercise of individual agency within the resulting domains
102 Chris Goldspink and Robert Kay
As this study was not concerned with the editing activity but with the self-
organizing and self-regulating phenomena which make it possible, the Discussion
pages of a sample of controversial and featured articles were analyzed. Controversial
articles were chosen as they were more likely to involve the need to resolve conflict
and hence place greater demand on effective normative regulation; featured articles
by contrast may be so rated due to the attainment of a higher level of consensus
among participants.
The activity on the Discussion pages comprises a series of ‘‘utterances’’ or speech
acts between contributors about editing activity and the quality of product. The only
means for editors to influence one another’s behavior (to structurally couple) is
through these utterances. On the face of it then, these pages should provide a fertile
source to support analysis of how self-organization was occurring and to identify the
agent characteristics and mechanisms involved.
It was anticipated that the process may involve quite subtle use of linguistic cues.
Accordingly sampled pages were coded to a high level of resolution using the Verbal
Response Mode (VRM) taxonomy (Stiles, 1992). VRM is very attractive where there
is a need (as in this case) to capture many of the subtleties of natural language use that
derive from and rely on the intrinsic flexibility and ambiguity of natural language,
yet map them to a more formal or axiomatic system needed for computer simulation.
A range of additional codes were applied, including: whether a listener accepted or
‘‘validated’’ an utterance; the explicit invocation of norms or rules; the associated
deontic command; and the style and focus (subject) of the utterance.
For the study we randomly selected a sample of Discussion pages associated with
both controversial and featured articles. At the time of the study (May/June 2007)
there were 583 articles identified by the Wikipedia community as controversial and
approximately 1900 as featured. The analysis reported here is based on a sample of 19
controversial and 11 featured articles. The most recent three pages of discussion were
selected for analysis from each Discussion page associated with the article included in
the sample.
These were subjected to detailed coding using the Open Source qualitative analysis
software WeftQDA. Both qualitative and quantitative analysis was performed. The
latter was undertaken by reprocessing the coded utterances such that each utterance
constituted a case and each applied code a variable associated with that case. This
data set was then analyzed using SPSS and MLWin.
Domain C
Domain A
Wikipedia Domain
Domain B
4.2.3 What kind of phenomenal domain emerges within Wikipedia? To think about
what is happening in the domain of Wikipedia, we can usefully draw on Habermas
theory of pragmatics. For Habermas, a successful speech act would be one in which
the listener both comprehends and accepts the validity claims made by the sender
and thus enters into the intended relationship. The tests of validity include
104 Chris Goldspink and Robert Kay
comprehensibility, truth, sincerity, and rightness. Thus for Habermas, a speech act
only serves to support the maintenance of effective communicative exchange to the
extent that it is held as valid by listeners. At the level of the individual agent, what is
held to be valid will largely be a product of its past participation in one or more
phenomenal domains with the norms or rules typical of that domain. Habermas
distinguishes between communicative acts and strategic action. The former is action
based on consensus while the latter implies action resulting from the exercise of power
or compulsion. The latter is not possible in Wikipedia as there are very few means for
compulsion or exercise of formal or authoritative power. The intrinsic openness
of Wikipedia means that the majority of exchanges can be expected to conform to
the qualities of communicative acts — i.e., bounded and influenced by normative
behavior rather than through the exercise of formal authority, power, or coercion.
The existence of community is central to establishing such an environment as the
heterogeneity of social backgrounds and experiences of participants coming together
incidentally around the task would likely fail to have sufficient power to provide
coherence to the relationships unless it had the opportunity to converge locally
around an accepted set of behavioral regulators. Do we see any evidence of this type
of regulator?
The absence of any expression of acknowledgment of emotions and/or similarity of
attitude (homophilly) among many contributors suggests that Wikipedia lacks many
of the qualities of verbal exchange that would identify it as strong community.
Possibly it therefore fails to constitute a distinct consensual domain. It is more
consistent with being a place to share coordination of a task. This could suggest that
the goal is the primary orientating point. However, the lack of quality of discourse
needed to achieve consensus is more indicative of a brief encounter between different
and established milieus which struggle to find common understanding rather than of a
community committed to a common goal (Becker & Mark 1997). This might suggest
that the primary influence of the utterance strategies employed by agents is the
consensual domain/s to which they belong in their wider life — not the immediate
environment of the Wikipedia. If this were the case then we would expect to see
speech acts which are a minimal accommodation: are minimally concerned with
establishing understanding and aimed at a pragmatic accommodation or satisficing of
presenting demands from different editors. Certainly this is one way of interpreting
the patterns observed in the data. Similarly we would expect to find that local norms
and rules had little effect and that social behavior was primarily influenced by the
socialized ‘‘norms’’ consistent with the editors’ primary domains — that is to say —
brought in from outside the Wikipedia.
4.2.4 Conclusions on Case Study Two. In this case we are particularly confronted
with the epistemic implications of the theory base we are following. Where do
consensual domains begin and end? Does the communicative activity in Wikipedia
give rise to a distinct phenomenal domain or can it only be understood by
appreciating the domains with which its participants are involved outside of the
Wikipedia? As Hejl long ago noted, the attributions of closure to social domains
(as compared to physical ones at the level of biological entities) is an epistemic act not
Autopoiesis and Organizations 105
an ontological one, and it reinforces the view that social systems are not autopoietic in
and of themselves.
Hejl (1984) distinguished between self-maintaining systems and self-referential
systems. He argued that functionally autonomous entities (such as organizations) are
abstract; they are self-referential but as they do not ‘‘self-produce’’ in a physical
domain, they should therefore be considered as self-maintaining but not autopoietic.
Thus both Varela and Hejl identify social systems as belonging to the broader
class of autonomous, operationally closed, and self-organizing/self-referential systems
but not as autopoietic. Further, the concept of autopoiesis only offers new insight
into systems that do self-produce in a physical domain: biological systems as per
the genesis of the concept. In relation to other classes of system the concept of
operational closure and self-organization are sufficient and equivalent.
To revisit some fundamentals, the criteria Maturana and Varela (1980) used to
distinguish autopoietic systems are:
1. their principal output is themselves, i.e., they are first and foremost self-producing;
2. they bring forth their own boundary as a result of their ongoing process of self-
production;
3. they are operationally closed and are therefore autonomous — their response to
perturbation being entirely determined by their structure;
4. in the case of composite unities there is mutual dependence between the levels of
autopoiesis — the continued autopoiesis of the components of a composite unity is
dependent on the maintenance of the autopoiesis of the composite unity and vice
versa.
and that restricting the definition to tangible boundaries ‘‘serves no useful purpose’’
(1991, p. 322). Mingers, in addressing this point, states:
A physical boundary has a spatial dimension forming a barrier between inside and outside.
This is not the case for a membership-type boundary; some members are not nearer the outside
than others. (Mingers, 1995, p. 128)
This is highly suggestive of naturalistic enquiry. The cases we have presented here
involved a degree of this in that the methods stayed close to the language usage of the
contributors, and they were involved in the choice of anecdotes. In the second case the
natural language was again used in order to find evidence of points of relative closure.
In social systems then, boundaries are defined by observers and it matters where we
draw them. This is not to say that we cannot gain some empirical clues as to where
we may usefully draw them and the Wikipedia case provides an example of the type of
data that may be used for this purpose and the implications of drawing it in different
places.
5 Overall Conclusions
The first case suggests that it may be possible to map the key distinctions which
characterize and contribute to the coherence of particular linguistic domains. The
challenge is in gaining sufficient initial lead to know where to look closely. We have
identified several conventional tools which can be used. Elsewhere we have also
outlined a model to assist with the interpretation of the resulting findings (Goldspink
et al., 2008). These are relatively easy to use and have modest data needs — a great
Autopoiesis and Organizations 107
References
Bannister, D., & Fransella, F. (1989). Inquiring man: The psychology of personal constructs.
London: Routledge.
Barandiaran, X. (2005). Behavioral adaptive autonomy. A milestone on the ALife route to AI?
San Sebastian, Spain: Department of Logic and Philosophy of Science, University of the
Basque Country.
Barandiaran, X., & Moreno, A. (2006). On what makes certain dynamical systems cognitive:
A minimally cognitive organization program. Adaptive Behavior, 14(2), 171–185.
Becker, B., & Mark, G. (1997). Constructing social systems through computer mediated
communication. Sankt Augustin, Germany: German National Research Center for
Information Technology.
Browning, L., & Boudes, T. (2005). The use of narrative to understand and
respond to complexity: A comparative analysis of the Cynefin and Weikian models. E:CO,
7(3–4), 7.
Bruner, J. (1991a). The narrative construction of reality. Critical Enquiry, 18(1), 20.
Bruner, J. (1991b). Self-making and world making. Journal of Aesthetic Education, 25(1), 11.
Burrell, G., & Morgan, G. (1980). Sociological paradigms and organizational analysis. London:
Heinemann.
Castelfranchi, C. (1998). Simulating with cognitive agents: The importance of cognitive
emergence. In: J. S. Sichman, R. Conte & N. Gilbert (Eds), Multi-agent systems and agent
based simulation. Berlin: Springer.
Clayton, P., & Davies, P. (2006). The re-emergence of emergence: The emergentist hypothesis
from science to religion. Oxford: Oxford University Press.
Cosmelli, D., et al. (2007). Neurodynamics of consciousness. In: P. D. Zelazo, M. Moscovitch
& E. Thompson (Eds), The Cambridge handbook of consciousness. Cambridge: Cambridge
University Press.
Dautenhahn, K. (2002). The origins of narrative: In search of the transactional format of
narratives in human and social animals. International Journal of Cognition and Technology,
1(1), 26.
Davies, P. (2006). The physics of downward causation. In: P. Clayton & P. Davies (Eds),
The re-emergence of emergence: The emergentist hypothesis from science to religion. Oxford:
Oxford University Press.
De Jaegher, H., & Di Paolo, E. A. (2007). Participatory sense-making: An enactive approach to
social cognition. Phenomenology and the Cognitive Sciences, 6(4), 485–507.
Di Paolo, E. A., & Lizuka, H. (2007). How (not) to model autonomous behavior. Biosystems,
91, 409–423.
Di Paolo, E. A., et al. (2007). Horizons for the enactive mind: Values, social interaction and
play. In: J. Stewart, O. Gapenne & E. A. Di Paolo (Eds), Enaction: Towards a new paradigm
for cognitive science. Cambridge, MA: MIT Press.
Duijn, M. V., et al. (2006). Principles of minimal cognition: Casting cognition as sensorimotor
coordination. Adaptive Behavior, 14(2), 157–170.
Autopoiesis and Organizations 109
Ellis, G. F. R. (2006). On the nature of emergent reality. In: P. Clayton & P. Davies (Eds),
The re-emergence of emergence: The emergentist hypothesis from science to religion. Oxford:
Oxford University Press.
Fransella, F., et al. (2004). A manual for repertory grid technique. Chichester: Wiley.
Fransella, F. B. (1977). A manual of repertory grid technique. London: Academic Press.
Fuchs, C., & Hofkirchner, W. (2005). The dialectic of bottom-up and top-down emergence in
social systems. TripleC, 1(1), 22.
Gaines, B. R. (1981). Autopoiesis: Some questions. In: M. Zeleny (Ed.), Autopoiesis: A theory
of living organization. New York, NY: North Holland.
Goldspink, C. (2007). Normative self-regulation in the emergence of global network
institutions: The Case of Wikipedia. ANZSYS07, Auckland, New Zealand.
Goldspink, C., & Kay, R. (2003). Organizations as self organizing and sustaining systems:
A complex and autopoietic systems perspective. International Journal General Systems, 32(5),
459–474.
Goldspink, C., & Kay, R. (2004). Bridging the micro–macro divide: A new basis for social
science. Human Relations, 57(5), 597–618.
Goldspink, C., & Kay, R. (2007). Social emergence: Distinguishing reflexive and non-reflexive
modes. AAAI Fall Symposium: Emergent Agents and Socialities: Social and Organizational
Aspects of Intelligence, Washington.
Goldspink, C., & Kay, R. (2008). Agent cognitive capability and orders of emergence. In:
G. Trajkovski, & S. Collins (Eds.), Agent-based societies: Social and cultural interactions.
Hershey, PA: Information Science Reference.
Goldspink, C., et al., (2008). Organizational change: Revealing the micro–macro patterns
underlying social system dynamics in a financial services context. First ICC Workshop on
Complexity in Social Sciences, ISCTE Lisbon, Portugal.
Gotham, K. F., & Staples, W. G. (1996). Narrative analysis and the new historical sociology.
The Sociological Quarterly, 37(3), 20.
Grice, J. W. (2002). Idiogrid: Software for the management and analysis of repertory grids.
Behavior Research Methods, Instruments, and Computers, 34, 3.
Habermas, J. (1976). Some distinctions in universal pragmatics: A working paper. Theory and
Society, 3(2), 12.
Hejl, P. (1981). The definition of system and the problem of the observer: The example of the
theory of society. In: G. S. Roth (Ed.), Self-organizing systems: An interdisciplinary approach
(pp. 171–185). New York, NY.
Hejl, P. (1993a). Culture as a network of socially constructed realities. In: A. F. Rigney (Ed.),
Cultural participation: Trends since the middle ages (pp. 227–250). Amsterdam: John
Benjamins Publishing Company.
Hejl, P. M. (1984). Towards a theory of social systems: Self-organization, self-maintenance,
self-reference and syn-reference. In: P. Ulrich (Ed.), Self-organization and management of
social systems: Insights, promises, doubts and questions (pp. 60–78). Berlin: Springer-Verlag.
Hejl, P. M. (1993b). Culture as a network of socially constructed realities. In: A. Rigney &
D. Fokkema (Eds), Amsterdam: John Benjamins Publishing Company.
Jankowics, D. (2004). The easy guide to repertory grids. London: Wiley.
Kelly, G. A. (1963). A theory of personality. New York, NY: Norton.
Kelso, J. A. S. (1995). Dynamic patterns: The self-organization of brain and behavior.
Cambridge, MA: MIT Press.
Landfield, A. W., & Cannell, J. E. (1988). Ways of assessing functionally independent
construction, meaningfulness, and construction in hierarchy. In: J. C. M. M. L. Shaw (Ed.),
Cognition and personal structure (pp. 67–90). New York, NY: Praeger.
110 Chris Goldspink and Robert Kay
1 Introduction
Maturana and Varela (1980, p. 78f) provided the following definition of autopoiesis:
‘‘An autopoietic machine is a machine organized (defined as a unity) as a network of
processes of production (transformation and destruction) of components that produces
the components which: (i) through their interactions and transformations continuously
regenerate and realize the network of processes (relations) that produced them and
(ii) 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.’’ This definition shows that for Maturana and Varela, autopoietic systems are
systems that define, maintain, and reproduce themselves. The notion of machine that
they employ in the definition might seem a bit misleading because we tend to think of
machines as mechanistic and nonliving, but Maturana and Varela (e.g., 1987) in later
publications have preferred to speak of autopoietic organizations.
Social systems are systems that are based on the interactions of living systems.
Maturana considers them as higher-order systems. The question therefore arises if
these systems are also autopoietic systems. The paper at hand will discuss this
question and try to give an answer that is critical of the one given by the main
representative of the theory of social autopoiesis — Niklas Luhmann.
According to Niklas Luhmann, the first and still most prominent thinker on social
autopoiesis, organizations are a variety of social systems besides interaction systems
and societal systems. As there has been much discussion on the question whether
social systems in general can be said to be autopoietic and if so to what extent,
we resume this discussion.
The rise of the importance of the sciences of complexity can be interpreted
as a turn toward the conception of reality as complex, dynamic, and networked.
In biology, Maturana and Varela (1987) have been two of the most important
scholars who are well known for the application of complexity thinking to living
systems. They argue that the differentia specifica of living systems is that they can
maintain and reproduce themselves by dynamically producing their own components
and with them a systemic unity. Conceiving a system as autopoietic means to stress
that it is dynamic and self-creating. The question arises if it is possible to generalize
this concept and to apply it to social systems and what advantages or disadvantages
such an endeavor brings. In this paper, we discuss two basic possibilities for
considering social systems as self-producing systems.
First, we discuss Niklas Luhmann’s approach of self-referential systems, which can
be considered as the most important approach of autopoietic social theory. Second,
based on a critique of Luhmann, we introduce an alternative approach that we term
critical social systems theory.
1
http://www.inteco.cl/biology/ask9707-1.htm
Autopoiesis and Critical Social Systems Theory 113
The consequence of Luhmann’s exclusion of humans and their interests from his
theory is a blindness for social problems that created an affirmative uncritical theory
that describes society as it is, not also as it could be. Luhmann (1984) explicitly argues
that his theory is not a social problems approach. So Luhmann (1996a), e.g., claims
that the mass media can’t manipulate humans because they, just like every system,
would construct a legitimate reality. The function of the mass media for him is that
they provide topics for communication and hence advance the autopoiesis of society.
There is no analysis of simplification, scandalization, and emotionalization as media
tactics, one-dimensional reporting, staged media events, the role of the Internet in the
mass media, media monopolies, and so on. For Luhmann, there are no problematic
aspects of the mass media — and of contemporary society at a whole.
The dramatic implications of Luhmann’s theory become most apparent in his
discussion of protest movements. He argues that social movements are alternatives
without alternatives (Luhmann 1996b, p. 75ff.), that they protest against the
functional differentiation of society (p. 76), operate within society against society
(pp. 103, 204), have no alternatives to offer (p. 104), fetishize opposition and
alternative thinking (p. 159), are made up by a notoriously mentally instable public
(p. 204), stage provocation as end in itself (p. 206), possess no analytical depth and
don’t know why something is as it is (p. 207), stage protest as pseudoevents (p. 212),
are a form of refractory communication against communication (p. 214), constitute a
disturbing aspect of modern society (Luhmann 1984, p. 545), and act as negators that
weaken the affirmation of society (ibid., p. 549ff.).
For Luhmann, protest movements are reactive, aimlessly, and dangerous. Each
protest movement has values and certain political goals; hence, it wants to change
society. Social movements are not reactive but active and proactive. Luhmann’s
characterization aims at discrediting protest; if the latter is not seen as a positive function
of society, alternatives are considered as undesirable. A society that forestalls critique
seems close to a totalitarian society; a theory that considers critique and opposition as
undesirable is affirmative and seems accordingly close to a totalitarian theory. The role
of sociology in society is critique and reflection of society; a pure description of society as
it is as the best form of society is uncritical and affirmative. For Luhmann, the function
of protest movements is that they convert the negation of society in society into
operations (ibid., p. 214). According to Hegel, a contradiction can be interpreted as not
purely negative but a determinate negation, i.e., a contradiction results in the negation of
the negation; it is sublated and produces positive results. Protest movements then can be
considered as a negation of existing structures and values, but they strive for changing
society, i.e., for a negation of the negation and for sublation. They are movements
because they move society and want to guarantee dynamic change.
Based on a dualistic concept of system and environment, Luhmann can neither
explain how ecological problems are caused nor how they could be solved; he is only
interested in how society communicates about ecological problems (ecological
communication) and argues that ecological problems are only problems because
society communicates them as problems (Luhmann, 2004, p. 63), which suggests a
radical constructivist perspective that doubts the existence of real problems. In such
an approach, ecological problems are not real but only constructed.
The Habermas/Luhmann debate has shown that there is a difference between critical
thinking and the thinking of Luhmann (Habermas & Luhmann, 1971). Habermas’s
Autopoiesis and Critical Social Systems Theory 115
main criticism of Luhmann is that the latter considers society as instrumental and
describes it as it is and not as it could be. Luhmann is only interested in describing
society, whereas Habermas argues that ignoring social problems and aspects of how to
improve society and how to advance human interests and human emancipation means
to reduce sociology to the logic of instrumental and functional reason. Habermas says
that Luhmann ignores the intersubjective and democratic dimensions of social
relationships, i.e., that consensus and participation can be achieved by communicative
action in ideal speech situations that satisfy the four validity claims of truth,
truthfulness, rightness, and comprehensibility. Luhmann argues that modern society is
too complex for allowing discursive decision taking. It is no wonder that based on a
system/human dualism, he is blind for social problems and human interests. Luhmann
(1984, p. 114) argues in this context that he does not pursue a social problems
approach. We agree with Habermas’s criticism in this respect. Luhmann constituted a
methodological antihumanism, whereas critical theories have always been forms of
methodological humanism. Critical theory is about analyzing how to change society,
for Luhmann social theory is about describing society. This is a crucial difference. We
stick to Horkheimer’s view that theory should not have an interest in ‘‘the preservation
of contemporary society but in its transformation into the right kind of society’’
(Horkheimer, 1937/2002, p. 218). Such theories try to show conditions and hindrances
for the emergence of a ‘‘society without injustice’’ (221) that is shaped by ‘‘reason-
ableness, and striving for peace, freedom, and happiness’’ (222), ‘‘in which man’s
actions no longer flow from a mechanism but from his own decision’’ (229), and that is
‘‘a state of affairs in which there will be no exploitation or oppression’’ (241).
An alternative view of how autopoiesis can be applied to social systems is the
critical social systems theory approach.
A critical social systems theory is a critical theory of social systems. It combines the
stance of critical theory as represented by, e.g., Habermas and Herbert Marcuse and
the Frankfurt School philosophers like Ernst Bloch — a theory which has its roots
in the weltanschauung of Karl Marx — and a system theoretical view, in particular,
science of complexity insights provided by Evolutionary Systems Theory (EST)
applied to the domain of social systems and going back to General System Theory
(GST) as inaugurated by Ludwig von Bertalanffy among others.
We present this approach by discussing three aspects of critical social systems:
design, modeling, and methodology one by one.
Design is concerned with the relation of theory and technology, theory and practice.
It’s the context of application, in which scientific knowledge is used for solving
problems and is transformed into technologies, whether material or ideational.
It addresses the opposition of normative versus descriptive.
116 Christian Fuchs and Wolfgang Hofkirchner
This question had been contested with considerable amount of attention in the
second half of the last century. Positivism tried to exclude this context by terming it
a factor external to science. Nowadays, in social science there seems to be a consensus
on rejecting the ideology of value-free science.
Not so with Luhmann. Luhmann’s theory is nonnormative, i.e., it avoids to discuss
and criticize societal problems. By doing so, however, it becomes affirmative because
just describing society as it is means to leave it unquestioned and give dominant
groups the opportunity to positively refer to this theory in their endeavor to uphold
asymmetric power relations.
Critical thinking is not entirely new to systems theory. If we equate the beginnings
of systems science with Ludwig von Bertalanffy’s GST, then systems science has been
normative from its very beginning. Bertalanffy’s GST is a humanistic one. Thus all his
descriptions of humans and social systems serve the function to help to formulate
guidelines for acting toward humane norms and values (see Hofkirchner, 2005).
Approaches like Critical Systems Thinking (CST) have been grounded in
Habermas’ version of critical theory. Two of the five commitments of CST are
critical awareness and dedication to human emancipation (Jackson, 2003). CST
rests ‘‘upon Habermas’ theory of human interests as mediated through the system
of system methodologies’’ (Jackson, 2003, p. 83). CST is ‘‘dedicated to human
emancipation and seeks to achieve for all individuals the maximum development of
their potential’’ (ibid., p. 85). It especially tries to advance the emancipatory interest
(which is one fundamental human interest besides the technical and the practical
interest) of humans by ‘‘denouncing situations where the exercise of power, or other
causes of distorted communication, are preventing the open and free discussion
necessary for the success of interaction’’ (ibid., p. 85).
CST sees itself in the service of a more general emancipatory project (ibid., p. 86).
‘‘Critical systems thinking, and the thrust of Total Systems Intervention (TSI)
therefore, is emancipatory in that it seeks to achieve for all individuals, working
through organizations and in society, the maximum of their potential. (y) The
exercise of power in the social process can prevent the open and free discussion
necessary for the success of interaction. Human beings have, therefore, an
‘emancipatory’ interest in freeing themselves from constraints imposed by power
relations and in learning, through a process of genuine participatory democracy,
involving discursive will formation, to control their own destiny’’ (Flood & Jackson,
1991, p. 95f).
Critical social systems thinking can easily be based upon EST — a term by which a
theory of complex, dynamic, nonlinear, open, self-organizing systems is denoted.
Evolutionary systems design principles encourage to make use of the systems’
dynamic and stress the point that knowing about nonlinearity and sensitivity may
help to choose those inputs that trigger developments in the overall self-organization
process of the system that are favorable to those who make the inputs. System
processes may be facilitated or may be dampened. Also it is important to influence the
general setup of the system only and abandon instructions down to every detail so
that relative autonomy is granted to the subsystems. Being critical can be ascribed to
this theoretical framework when applied to social systems in that it is normative while
Autopoiesis and Critical Social Systems Theory 117
doing justice to the factual at the same time. For it includes not only an account of the
potential that is given with the actual, but also an evaluation of the potential which
sorts out the desired.
In a philosophical perspective, this deliberate activism is not a practicism that
guides action according to the maxim that all that is feasible shall be realized thereby
assuming that it is desired too. Nor is this kind of activism a utopian or romantic
wishful thinking that holds that what is desired is feasible too. Both practicism and
wishful thinking believe in total controllability and result in expensive brute-force
interventions. Nor is this kind of activism an inactivism that believes in total
uncontrollability, condemns any kind of intervention and fails to reconcile the
feasible and the wishful. On the contrary, it takes responsibility for producing the
unity of the feasible and the wishful. And it does so by working out the ascendance
from the potential given now to the actual to be established in the future as well as
the ascendance from the less good now to the better then which altogether yields the
Not-Yet in critical theorist Ernst Bloch’s sense (see, e.g., Bloch, 1967). These
processes aimed at the Not-Yet are at the core of the dynamic of social self-
organization. By the notion of the Not-Yet Bloch tried to salvage the idea of
utopia — it is not any longer a nowhere deprived of the possibility to get there but a
future that can be glimpsed and anticipated in what is already possible here and now.
Why is it especially important today to advance a critical approach? We consider it
irresponsible if social theory is watching as a bystander as the world is increasingly
getting out of human control. Due to the existence of global problems, we argue that
a critical social systems theory is needed. There is evidence that late-modern society
is characterized by culminating antagonisms between economic feasibility and social
usefulness of technological products, between economic growth and ecological
sustainability, and between economic freedom (of markets) and social equity.2
2
Income inequality measured as the relation of the mean income of the upper and the lower quintile has
decreased in the years 1995–2000 in the EU15 countries, but it has increased from 4.5 in 2000 to 4.8 in 2005
(Eurostat Online). The higher this measure, the higher the income disparity between the poorest and the
richest. In the EU25 countries, it has increased from 4.5 in 2000 to 4.9 in 2005. In 2000, the richest 5%
Europeans owned 35.7% of the worldwide wealth (Davies et al., 2006, table 10a). The at-risk-of-poverty
rate after social transfers measured by 60% of median equivalized income after social transfers has risen
from 15% in 1998 to 16% in 2005 in the EU15 as well as the EU25 countries (Eurostat Online). Income
inequality as measured by the Gini coefficient has increased from 29 in 1998 to 31 in 2005 in the EU25
countries and from 29 in 1998 to 30 in 2005 in the EU15 countries (Eurostat Online). The in-work at risk of
poverty rates for part time workers was 11% in the EU25 and 10% in the EU15 countries in 2005 (Eurostat
Online). The increase in income inequality, job insecurity, and poverty risk has been accompanied by a
polarization between capital and labour: Whereas the average profit rate has increased by 39.4% in the
years 1987–2007 in the EU15 countries (net returns on net capital stock, European Commission Annual
Macro-Economic Database), the wage share has in the same time span decreased by 7.5% (Compensation
per employee as percentage of GDP at current market prices, European Commission Annual Macro-
Economic Database). It is hence reasonable to assume that during the last years and decades, economic
growth has been accompanied by a rise of relative wage decreases, income inequalities, and poverty risks.
Hence, we assume that such a form of economic growth, i.e., the unhindered expansion of capital
accumulation, is not compatible with social sustainability.
118 Christian Fuchs and Wolfgang Hofkirchner
Critical theory doesn’t accept existing social structures as they are, it is not
interested in society as it is, but in what society could be and can become. It
deconstructs ideologies that claim that something can’t be changed and shows
potential counter tendencies and alternative modes of development. That the negative
antagonisms are sublated into positive results is not an automatism, but depends on
the realization of practical forces of change that have a potential to rise from the
inside of the systems in question in order to produce a transcendental outside that
becomes a new whole. All critical approaches in one or the other respect take the
standpoint of oppressed or exploited classes and make the judgment that structures of
oppression and exploitation benefit certain classes at the expense of others and hence
should be radically transformed by social struggles.
We understand the notion of critical theory in the sense of approaches that are
oriented on maximizing human potentials and realizing societal conditions that give
advantages to all humans. Such theories are human-centered; they have human needs
and the goal of a good life for all as their central concern. This endeavor also includes
criticizing societal conditions that limit human potentials as unjust.
If critical theory means human-centeredness as normative quality, then such a
theory needs to put humans also in the center of theory itself. That global problems
like global war, the ecological crisis, rising inequality, precarious labor and living
conditions, etc., have emerged is an indication for the assumption that under the
given societal conditions, human-centeredness is only the essence, but not the reality
of society — to paraphrase Hegel. Not all humans benefit, only certain classes benefit
at the expense of the large majority. Human-centeredness implies that society shall be
designed in a way that allows all humans to realize a maximum of their potentials and
to live a good life. As this is not the case today, human-centeredness implies the
critique of contemporary society and the normative claim for societal transformation.
Contemporary capitalist society is not human-centered, but capital- and power-
centered — money capital and political power have colonized human interests and
caused an alienation of society from its human essence.
Our approach therefore aims at human-centeredness, but at analyzing social systems
nonetheless as dynamic and self-producing. What is the basic advantage of the
application of a transposed notion of autopoiesis to social systems for a critical theory? If
social systems are conceived as dynamic, fundamental social change can be conceived as
a potential development. This is particularly important today because neoliberal scholars
and politicians tend to argue that there are no alternatives to neoliberalism (the ideology
of ‘‘there is no alternative,’’ TINA) in particular and capitalism in general. Dynamics
means that change is an inherent feature of society. If change is taking place permanently,
then it is likely that fundamental change can also occur and is an option that humans can
pursue. This holds not only for societal systems, but for organizations as well.
The basic onto-epistemological question is: are models constructs that are subjective,
kind of arbitrary, and can’t be corroborated because of the lack of an authority that
Autopoiesis and Critical Social Systems Theory 119
would decide upon truth or untruth because this decision, in turn, would need a
legitimation — and so forth ad infinitum — or are models made for mapping
reality by which objectivity would enter the scene? The answers distinguish
between a constructivist and a realist stance. Modeling is about the relation of
theory to reality and by that about the relations of theories to each other. It is
the context of justification in which scientific knowledge is critically exposed to
possible refutations and corroborated in as far as it is not refuted and theories are
comparatively assessed.
Unlike today’s radical constructivism, Bertalanffy’s GST supported the idea that
we are dealing with real-world systems and not with mere constructs. However, there
is also a constructivist part in his GST perspective, for he appreciated the fact that it is
models we construct in dealing with reality and that it is models that determine
how we perceive reality. He called his view ‘‘perspectivism’’ which is neither
absolutism nor nihilism. He stated that, e.g., a fly, a dog, or a human being has only
limited knowledge of the world, but that this knowledge has some validity because
otherwise the fly, the dog, the human would not have been able to survive for long
(see Hofkirchner, 2005).
Evolutionary systems modeling principles take as starting point that self-
organization takes place in phases that yield different levels of real-world systems.
Evolutionary systems undergo stages. The stage model of systems evolution is based
upon the principle of emergentism and the principle of asymmetrism. Emergence
takes place in transitions in which by the interaction of proto-elements systems are
produced. Asymmetry describes the suprasystem hierarchies in which subsystems are
encapsulated.
The ontological perspective of EST — a term coined by Ervin Laszlo (1987),
Vilmos Csanyi (1989), and Susantha Goonatilake (1991) — as a theory about
evolving systems is the result of the merger of systems theory and evolutionary theory
which nowadays not only applies to living and human/social systems but also to
physical systems, i.e., to the cosmos itself.
EST aims at distinguishing between different levels of self-organization, i.e., self-
organization has aspects that are common to all types of systems as well as aspects
that are unique to a concrete type of system. There are systems and processes that
manifest patterns. Pattern is form, i.e., a superstructure that refers to a basis that
refers to the superstructure, and so on. These are macro- and micro-levels that coexist
and influence each other which is more important than the influence from outside.
The system is produced by its elements, and the system constrains and enables its
elements at the same time. As this works by dissipation of entropy, Ilya Prigogine
(1980) called the emerging structures ‘‘dissipative.’’ The fluid particle in the Bénard
convection cell — i.e., a hexagonal pattern emerging from conduction in liquids
of high viscosity if exposed to a temperature gradient that exceeds a certain critical
value — is prompted to contribute to the cell structure that emerges from the
activities of all particles. This is said to be true of all self-organizing systems on a
physical and chemical level.
Then there are systems and processes that are able to maintain the form they show,
i.e., to hold the form stable while matter is changing. This is the case with all living
systems. This is why Maturana and Varela (1980) called them ‘‘autopoietic.’’
120 Christian Fuchs and Wolfgang Hofkirchner
By stressing the fact that in living systems the elements that constitute the system
produce new elements by which the system can be constituted, Maturana and Varela
denoted that living systems are systems that produce themselves by constraining and
enabling their elements to produce new elements that produce the systems. Put it that
way, it becomes clear that EST can consider autopoiesis, i.e., living self-organization,
as evolutionary follow-up of dissipative self-organization in physical and chemical
systems and as physical and chemical basis for biotic self-organization processes.
Autopoiesis can be looked upon as a particular way of universal self-organization,
characteristic of living systems.
The same figure of thinking evolutionary systemically is applied by EST when it
comes to the human/social level. Human/social system and processes are viewed as
systems and processes that change their form in a rather deliberative way, i.e., they
are endowed with the capability to transcend themselves, invent themselves. This is
what Erich Jantsch (1987) was pointing at when talking of ‘‘re-creative’’ systems at
the human/social level.
Re-creation means that social systems do not only have the capacity to modify
themselves (as physical and chemical self-organizing systems do) and to essentially
maintain themselves (as living self-organizing systems do), but they also have the
capacity to reinvent themselves, to shape themselves, to produce a specific character
by which the individuals that are parts of a social system can strive to realize
themselves in a more or less self-determined way. That is to say, systems at
the evolutionary stage of human society are just another — but new — way of
metabolism nonhuman living systems carry out (just as systems at the evolutionary
stage of living beings are another way of making use of energy than nonliving
material systems do): re-creation is a particular way of autopoiesis which is a
particular way of dissipative self-organization.
In this vein, there is no problem to include human beings in social systems, but it is
a requisite to do this. Though humans are ‘‘produced’’ by humans in a biological
sense, they are also produced as social beings, as members of social systems, by the
actions they carry out under the constraints and enablers social systems represents to
them. So autopoiesis is clearly there, and it is amended in so far as transformations of
social systems can occur.
Applying EST models to social systems means to give an answer to how to
relate individuals and society — the central theme each general sociological theory
revolves around and which is known today as the duality of agency and structure
(see Reckwitz, 1997). Given this basic duality, there are four ways of conceiving
of their relationship. The first way is individualism which can be classified as
downward reductionism because it gives priority to individual phenomena over
societal ones as action theory does. The second way is a kind of reverse reductionisms
which better may be called downward projectivism typical of structuralism that
reverses the priority relationship, which means that properties of the higher, the
macro-, level (society) are projected onto, or extrapolated to, properties of the lower,
the micro-, level (individuals and their actions). The third way is a dualistic view
that takes for granted the independent existence of structures and agency and cuts
Autopoiesis and Critical Social Systems Theory 121
individuals free from societal structures which is the way Luhmann chose
(see Hofkirchner, 2006).
The approach of critical social systems theory that we have tried to ground
during the past years (cf., e.g., Hofkirchner, 1998, 2007; Fuchs & Hofkirchner,
2003, 2005; Fuchs, 2008a) is a fourth way to conceive of this relationship
and starts from the human-centered argument that human beings as such are
creative social beings that cocreate social reality together with others. Society is
conceived as a large-scale system of networked social systems that is based on
the dialectic of social structures and human actors. By social actions, social
structures are constituted and differentiated. The structure of a social system is
made up by the total of regularized social behavior and relations that are
continuously reproduced over certain time spans. By social interaction, new qualities
and structures can emerge that cannot be reduced to the individual level. This is a
process of bottom-up emergence that is called agency. Emergence in this context
means the appearance of at least one new systemic quality that cannot be reduced to
the elements of the system. So this quality is irreducible and it is also to a certain
extent unpredictable, i.e., time, form, and result of the process of emergence cannot be
fully forecast by taking a look at the elements and their interactions. Social structures
also influence individual actions and thinking. They constrain and enable actions.
This is a process of top-down emergence where new individual and group properties
can emerge.
The whole cycle is the basic process of systemic social self-organization that can
also be called re-creation because by permanent processes of agency and constraining/
enabling, a social system can not only maintain and reproduce itself but also
transform itself, i.e., create itself anew (see Figure 1, Hofkirchner, 1998, cf. also
Fuchs, 2008a). It again and again creates its own unity and maintains itself. Social
structures enable and constrain social actions as well as individuality and are a result
of social actions (which are a correlation of mutual individuality that results in
sociality). We term social systems due to their dependence on human creativity and
self-producing re-creative systems.
This approach is dialectical because it conceives social systems as an interconnec-
tion of human actors and social structures. Actors and structures on the one
hand are different, on the other hand actors form and are part of certain social
structures and social structures condition and hence become part of human actions.
The relationship can be conceived as being based on difference, unity, and
interdependence. Individuals and society are interdependent (none of them can be
understood without the other), they oppose each other (none of them is fully
understandable by understanding the other), and they build a systemic hierarchy
(society plays the dominant role). Dialectics is said to apply whenever two correlates
build a mutually dependent relationship between themselves as opposites in an
asymmetrical way.
In Luhmann’s approach, the unit of social autopoiesis is communication. In our
approach, the unit of social autopoiesis ( ¼ re-creation) is human actors permanently
reproducing and/or transforming social structures. Society doesn’t produce and
122 Christian Fuchs and Wolfgang Hofkirchner
3
Mingers’ theoretical works form a very important contribution to social theory because he tries to connect
aspects of social self-organization with modern sociological theories. Mingers wants to combine Luhmann’s
with Giddens’ theory and says that society is mutually related to the interactional domain where people
interact. ‘‘Society selects interactions and interactions select society — this is their form of organizational
closure. We can choose to observe society, and see networks of communications triggering further
communications, and forming self-bounded subsystems that persist and reproduce over time. Or, we can
focus on particular episodes of interaction between individuals and groups’’ (Mingers, 1999, p. 38). If one
observes society or a social system, one will not find either communications or interacting individuals, but
both at once. Separating communications and individuals into two separate domains results in a rather
dualistic and non-consistent conception. Communication and social interactions do not constitute separate
domains, they are part of the structure that relates social groups and individuals, they exist in-between
individuals as a connecting mechanism. To avoid shortcomings, one could conceive social structures as a
unity of social relationships that take place in and through interaction and communication and social forms
such as rules and resources. As long as communications are defined as components of a social system, it is a
very hard or nearly impossible task to integrate the theories of Luhmann and Giddens. We prefer to define
individuals as social beings and components of social systems in such a way that society produces man as a
social being just like man produces society as a necessary condition for his/her social being.
124 Christian Fuchs and Wolfgang Hofkirchner
their works to the sciences of complexity (for such an endeavor cf. Fuchs, 2003a,
2003b). Bourdieu argues that there is a ‘‘dialectical relationship between the objective
structures and the cognitive and motivating structures which they produce and which
tend to reproduce them, (y) these objective structures are themselves products of
historical practices and are constantly reproduced and transformed by historical
practices whose productive principle is itself the product of the structures which it
consequently tends to reproduce’’ (Bourdieu, 1977, p. 83). For Bourdieu, the concept
that establishes the connection between structures and agency is that of the habitus.
Giddens formulates the dialectic as duality of structure: ‘‘According to the notion of
the duality of structure, the structural properties of social systems are both medium
and outcome of the practices they recursively organize’’ (Giddens, 1984, p. 25).
One aspect that these approaches have in common is that they consider themselves
as dynamic critical realist theories that are not naive, but dynamic and acknowledge
the importance of active humans and their social relations in society. However, not all
of these approaches are critical. The approach, which most clearly can be considered
as a critical theory, is Bhaskar’s Dialectical Critical Realism. For Bhaskar, there is a
normative feature in dialectical thinking that he terms Moral Realism. Its central
feature would be absenting absence. ‘‘This encompasses the absenting of constraints,
including ills generally, which comprise lack of freedoms. (y) Dialectic is the process
of absenting constraints on absenting absences (ills, constraints, untruths, etc.)’’
(Bhaskar, 1993, pp. 102, 297). Dialectic would be the axiology and pulse of freedom
(Bhaskar, 1993, pp. 378, 385). ‘‘Dialectic is the yearning for freedom and the
transformative negation of constraints on it’’ (Bhaskar, 1993, p. 378). Bhaskar
stresses a quality of critical thinking that is inherent in Marxian thinking: the critique
of all domination because it sets limits on human potentials. ‘‘Theory is capable of
gripping the masses as soon as it demonstrates ad hominem, and it demonstrates
ad hominem as soon as it becomes radical. To be radical is to grasp the root of the
matter. But, for man, the root is man himself. (y) The criticism of religion ends with
the teaching that man is the highest essence for man — hence, with the categoric
imperative to overthrow all relations in which man is a debased, enslaved, abandoned,
despicable essence, relations which cannot be better described than by the cry of a
Frenchman when it was planned to introduce a tax on dogs: Poor dogs! They want to
treat you as human beings!’’ (Marx, 1844, p. 385).
Critical theory is materialistic in the sense that it addresses phenomena and
problems not in terms of absolute ideas and predetermined societal development, but
in terms of resource distribution and social struggles. Reality is seen in terms that
address ownership, private property, resource distribution, social struggles, power,
resource control, exploitation, and domination. In such an endeavor, a reactualized
notion of class is of central importance (cf. Fuchs, 2008a, Chapter 7.3). To make a
materialistic analysis also means to conceive society as negativity. To identify
antagonisms means to take a look at contradictory tendencies that relate to one and
the same phenomenon, create societal problems and require a fundamental systemic
change in order to be dissolved.
Our critical social systems modeling approach is a non-constructivist one because
we find it difficult to conceive society as just a construct of the human mind
Autopoiesis and Critical Social Systems Theory 125
(as, e.g., argued by Ernst von Glasersfeld, 2008, cf. also the comment on Glasersfeld’s
notion of society by one of the authors of the paper at hand, Fuchs, 2008b). The
regularized patterns of society that we encounter and cocreate in everyday life and
that seem to enable continuous social activity are evidence that we can be confident
that others exist and are potential partners of communication in an overall shared
space that is termed society and that is created by many individuals together
and hence is not independent of these individuals, but also not reducible to their
cognition, as they require others with whom they mutually create that space. This
space is objective in the sense that it is cocreated by humans who in their social
relationships create supra-individual regularized patterns of interaction that they
can rely on in everyday life and that makes social activity work. Society is not
independent of individuals, but also not as radical constructivism seems to claim only
subjectively cognitively constructed.
Therefore, our notion of self-producing social systems is realistic, it assumes that
social reality exists objectively and is recognized and transformed by humans who are
parts of social reality and form this reality in interaction with others. Our approach
could be classified as a variety of critical realism (cf. Bhaskar, 1975, 1993, 1998).
the speculative natural philosophy of antiquity. Rather, it can and must assimilate the
knowledge gained from research in every discipline in a historical process which rises
from the abstract to the concrete.
This way of thinking was applied when Flood and Jackson (1991) dealing with the
variety of approaches in the systems movement itself came up with their so-called
System of Systems Methodologies, which they called ‘‘complementarism’’ and,
after slight modification in the tradition of their CST, was, e.g., termed ‘‘discordant
pluralism’’ (Gregory, 1996). Complementarism or discordant pluralism does not
mean that anything goes. ‘‘There is a need for debate about what are ‘‘good’’
arguments and what are not, and for discussion about how we can choose between
different positions that are conflicting’’ (Gregory, 1996, p. 54). Though ‘‘different
perspectives and systems methodologies should be used in a complementary way to
highlight and address different aspects of organizations, their issues and problems’’
(Jackson, 2003, p. 285), they are brought together in a constellation that does not give
way to a reduction to a common denominator, but serves as the basis for a discourse
that, as Gregory points out, is ‘‘not a relativistic chaos of unrelated factors, but
a dialectical model’’ (Jay, 1984, p. 15, cited in Gregory, 1996, p. 54). The question of
different perspectives is framed ‘‘in a way that recognizes the legitimacies of each
position’’ involved. It ‘‘is a third perspective through which the legitimacies of each
value system can be brought together in a critically systemic discourse.’’ This may
include that ‘‘such a constellation may legitimately eliminate elements of otherness
that have been identified as illegitimate’’ (Gregory, 1996, p. 55).
While the System of Systems Methodologies is confined to the tool box of systems
approaches, Mingers made substantial contributions to multi-method research and a
pluralist methodology in the realm of sociology and social science and information
systems outside systems thinking when, in drawing upon Bhaskar, elaborating a
philosophical position called ‘‘critical pluralism’’ (see Mingers, 2001a). He argues for
detaching research methods from the paradigm they are espoused with by convention
and assign them a role in concrete research tasks independent of the traditional
paradigms. But, other than Mingers contends, we believe that putting them into a new
context means and shall mean their integration on a meta-level system of methods.
Combining both the ideas of a System of Systems Methodologies and Critical
Pluralism, i.e., extending the systematization attempt beyond systemic methods
to other sociological, social scientific, and other methods is what is needed, when
it comes to critical social systems thinking in our opinion. On a meta-level, a
methodology can be built that is a system of methods that, in turn, originate
from different theoretical angles, but undergo a process of critical reconsideration
in order to suit a common methodological umbrella. The underlying way of
thinking is a dialectical account of unity and diversity or identity and difference.
Ways of thinking can be seen as ways of considering how to relate identity and
difference. The dialectical one — which is opposed to reductionism, projectivism, and
disjunctivism — establishes identity in line with the difference; it integrates both sides
of the difference (yielding unity) and it differentiates identity (yielding diversity); it is
a way of thinking that is based upon integration and differentiation; it is opposed to
both unification and dissociation and yields unity and diversity in one — unity in
Autopoiesis and Critical Social Systems Theory 127
diversity and diversity in unity. As French social thinker and systems philosopher
Edgar Morin puts it: ‘‘It means understanding disjunctive, reductive thought by
exercising thought that distinguishes and connects. It does not mean giving up
knowledge of the parts for knowledge of the whole, or giving up analysis for
synthesis, it means conjugating them. This is the challenge of complexity which
ineluctably confronts us as our planetary era advances and evolves’’ (1999, p. 19).
Thus, a critical social systems methodology shall be capable of doing justice to
methods other than system methods and including them as well. It is critical in that it
combines different methods to consider properly the multidimensionality of the world.
The concept of re-creation or social autopoiesis takes into account that phenomena
don’t have linear causes and effects, but are complex, dynamic, and open to the future.
Systems carry certain development potentials in them that at the same time pose
positive and negative potentials that are realized or suppressed by human social
practice. Dialectical reasoning means acknowledging the existence of contradictions
and the search for these contradictions. Dialectical analysis in this context means
complex dynamic thinking, realism an analysis of real possibilities and a dialectic of
pessimism and optimism. In a dialectical analysis, phenomena are analyzed in terms
of the dialectics of agency and structures, discontinuity and continuity, the one and
the many, potentiality and actuality, global and local, virtual and real, optimism and
pessimism, essence and existence, immanence and transcendence, etc.
4 Conclusion
In this paper, we have argued for a turn in systems theory and social autopoiesis
theory away from constructivism and functionalism toward critical thinking,
dialectics, and human-centeredness. The predominant application of autopoiesis to
social systems is Niklas Luhmann’s social systems theory. We identified the exclusion
of humans from social systems as the main problem of Luhmann’s theory.
This separation of systems and humans results in an affirmative approach that
neglects social problems. Critical social systems theory sees humans at the center of
social systems, it argues that humans coproduce and reproduce social structures,
which condition further human actions, by which again structures emerge and
are reproduced, etc. This dynamic, dialectical process was termed re-creation.
Re-creation is an autopoietic process because the unity of human actors and social
structures that constitutes sociality is permanently reproduced and reemerging.
The acuteness of global societal problems requires that today social theory is not just
descriptive and analytical, but also normative and in the interest of oppressed groups
and individuals. Therefore, we argued that human-centeredness should also be seen as
an important critical feature of contemporary social theory.
Are social systems autopoietic? Yes, but we suggest an understanding that is
human-centered and therefore departing from Luhmann’s interpretation. We argue
that humans permanently create the unity of human actors and social structures,
i.e., human sociality, in society. What is permanently created in society is the
128 Christian Fuchs and Wolfgang Hofkirchner
fundamental quality of humans, their sociality. Society reproduces and produces man
as a social being, and man reproduces and produces society by socially coordinating
human actions. Man is creator of, and created by, society; society and humans
produce each other mutually. We try to frame social autopoiesis as a process, in
which we find a dialectic of social structures and human actors. Luhmann’s focus on
communications and structures as unit of autopoietic reproduction is in our approach
replaced by the unity of structure and actors. We have argued that this focus allows
to build a critical autopoietic theory of organizations and society. The gain of a
reinterpretation of autopoiesis that is connected to thinkers like Giddens, Bourdieu,
or Bhaskar is a critical focus that we miss in Luhmann’s theory.
References
Archer, M. S. (2002). Realism and the problem of agency. Alethia, 5(1), 11–20.
Beyerle, M. (1994). Staatstheorie und Autopoiesis: Über die Auflösung der modernen Staatsidee
im nachmodernen Denken durch die Theorie autopoietischer Systeme und der Entwurf eines
nachmodernen Staatskonzepts. Frankfurt/Main: Peter Lang.
Bhaskar, R. (1975). A realist theory of science. London: Verso.
Bhaskar, R. (1993). Dialectic: The pulse of freedom. London: Verso.
Bhaskar, R. (1998). The possibility of naturalism. A philosophical critique of the contemporary
human sciences. New York, NY: Routledge.
Bloch, E. (1967). Das Prinzip Hoffnung (Vol. 3). Frankfurt: Suhrkamp.
Bourdieu, P. (1977). Outline of a theory of practice. Cambridge, MA: Cambridge University Press.
Csanyi, V. (1989). Evolutionary systems and society. A general theory of life, mind and culture.
Durham, NC: Duke University Press.
Davies, J.B., Susanna Sandstrom, S., Anthony Shorrocks, A., & Edward, N. Wolff, E. N.
(2006). The world distribution of household wealth. URL (consulted June 2008). Available
at: http://www.iariw.org/papers/2006/davies.pdf
Flood, R. L., & Jackson, M. C. (1991). Creative problem solving: Total systems intervention.
New York: Wiley.
Fuchs, C. (2003a). Some implications of Pierre Bourdieu’s works for a theory of social self-
organization. European Journal of Social Theory, 6(4), 387–408.
Fuchs, C. (2003b). Structuration theory and self-organization. Systemic Practice and Action
Research, 16(4), 133–167.
Fuchs, C. (2008a). Internet and society: Social theory in the information age. New York, NY:
Routledge.
Fuchs, C. (2008b). Sociology, dynamic critical realism, and radical constructivism.
Constructivist Foundations, 3(2), 97–99.
Fuchs, C., & Hofkirchner, W. (2003). The dialectic of bottom-up and top-down emergence in
social systems. tripleC (http://triplec.at), 3(2), 28–50.
Fuchs, C., & Hofkirchner, W. (2005). Self-organization, knowledge, and responsibility.
Kybernetes, 34(1–2), 241–260.
Giddens, A. (1984). The constitution of society. Outline of the theory of structuration.
Cambridge: Polity Press.
von Glasersfeld, E. (2008). Who conceives of society? Constructivist Foundations, 3(2), 59–64.
Goonatilake, S. (1991). The evolution of information. Lineages in gene, culture and artefact.
London: Pinter.
Autopoiesis and Critical Social Systems Theory 129
Gregory, W. (1996). Dealing with diversity. In: R. L. Flood & N. R. A. Romm (Eds), Critical
systems thinking, current research and practice. New York, NY and London: Plenum Press.
Habermas, J., & Luhmann, N. (1971). Theorie der Gesellschaft oder Sozialtechnologie?
Frankfurt/Main: Suhrkamp.
Hofkirchner, W. (1998). Emergence and the logic of explanation. Acta Polytechnica
Scandinavica. Mathematics, Computing and Management in Engineering Series, 91, 23–30.
Hofkirchner, W. (2005). Ludwig von Bertalanffy. Forerunner of evolutionary systems theory.
In: J. Gu & Chroust, G. (Eds), The New Role of Systems Sciences for a Knowledge-Based
Society, Proceedings of the First World Congress of the International Federation for Systems
Research. Kobe, Japan, ISBN 4-903092-02-X, 6pp.
Hofkirchner, W. (2006). Towards a post-Luhmannian social systems view. International
Sociological Association: Proceedings of the XVI World Congress of Sociology: The Quality of
Social Existence in a Globalizing World. Durban. CD-ROM.
Hofkirchner, W. (2007). A critical social systems view of the Internet. Philosophy of the Social
Sciences, 37(4), 471–500.
Horkheimer, M. (1937/2002). Traditional and critical theory. In: Critical theory (pp. 188–252).
New York: Continuum.
Jackson, M. C. (2003). The origins and nature of critical systems thinking. In: G. Midgley (Ed.),
Systems thinking volume IV (pp. 77–92). London: Sage.
Jackson, M. C. (2003). Systems thinking: Creative holism for managers. Hoboken, NJ: Wiley.
Jantsch, E. (1987). Erkenntnistheoretische Aspekte der Selbstorganisation natürlicher Systeme.
In: S. J. Schmidt (Ed.), Der Diskurs des Radikalen Konstruktivismus (pp. 159–191). Frankfurt:
Suhrkamp.
Jay, M. (1984). Adorno. Cambridge, MA: Harvard University Press.
Laszlo, E. (1987). Evolution — The grand synthesis. Boston, MA: New Science Library.
Luhmann, N. (1984). Soziale Systeme. Frankfurt/Main: Suhrkamp.
Luhmann, N. (1988). The autopoiesis of social systems. In: F. Geyer & J. Van der Zouwen
(Eds), Sociocybernetic paradoxes (pp. 172–192). London: Sage.
Luhmann, N. (1996a). Die Realität der Massenmedien. Opladen: Westdeutscher Verlag.
Luhmann, N. (1996b). Protest: Systemtheorie und soziale Bewegungen. Frankfurt/Main: Suhrkamp.
Luhmann, N. (2004). Ökologische Kommunikation. Opladen: Westdeutscher Verlag.
Marx, K. (1844). Einleitung zur Kritik der Hegelschen Rechtsphilosophie. In: MEW, Vol. 1
(pp. 378–391). Berlin: Dietz.
Maturana, H. (1987). Biologie der Sozialität. In: S. J. Schmidt (Ed.), Der Diskurs des Radikalen
Konstruktivismus (pp. 287–302). Frankfurt/Main: Suhrkamp.
Maturana, H. R. (1980). Man and society. In: F. Benseler, P. Hejl & W. Kock (Eds),
Autopoiesis, communication, and society: The theory of autopoietic systems in the social
sciences (pp. 11–32). Frankfurt: Campus.
Maturana, H. R., & Varela, F. (1980). Autopoiesis and cognition. Dordrecht: Reidel.
Maturana, H. R., & Varela, F. J. (1987). The tree of knowledge. Boston, MA: Shambhala.
Mingers, J. (1995). Self-producing systems. Implications and applications of autopoiesis.
New York, NY: Plenum.
Mingers, J. (1999). Information, meaning, and communication: An autopoietic approach to
linking the social and the individual. Cybernetics & Human Knowing, 6(4), 25–41.
Mingers, J. (2001a). Combining IS research methods: Towards a pluralist methodology.
Information Systems Research, 12(3), 240–259.
Prigogine, I. (1980). From being to becoming. San Francisco, CA: Freeman.
Reckwitz, A. (1997). Struktur. Zur sozialwissenschaftlichen Analyse von Regeln und
RegelmäXigkeiten. Opladen: Westdeutscher Verlag.
PART III
APPLICATIONS OF AUTOPOIESIS IN
ORGANIZATION THEORY
Chapter 7
1 Introduction
Many recent studies have voiced the growing concern that the body of knowledge that
springs from organization science is hardly taken notice of in management practice.
This has given rise to urgent calls for making organization research more relevant
to practitioners and an intensive debate on how to realize this aim has set in
(e.g., Hodgkinson, Herriot, & Anderson, 2001; Rynes, Bartunek, & Daft, 2001;
MacLean & MacIntosh, 2002; Baldridge, Floyd, & Markoczy, 2004; Van de Ven &
Johnson, 2006). In most of the existing literature one can identify three main reasons
for the observable lack of connection between organization research and practice:
research is not sufficiently focused on the ‘‘real’’ problems of practitioners (e.g.,
Rynes, McNatt, & Breetz, 1999), research results are not properly disseminated
to practitioners (e.g., Spencer, 2001), and the language of science is not properly
translated into the language practitioners’ use (e.g., Starkey & Madan, 2001; Van de
Ven & Johnson, 2006). The underlying assumption is that if scientists redressed
these shortcomings, their findings would be utilized by practitioners and thus the gap
between theory and practice would be bridged.
The aim of this chapter is to contrast this recent debate on the relation between
science and practice with an analysis from the perspective of Niklas Luhmann’s
theory of autopoietic systems. According to this perspective, the lack of any transfer
Niklas Luhmann (1927–1998) was one of the most influential sociologists to have
drawn on the concept of autopoiesis. The two Chilean cognitive biologists Humberto
Maturana and Francisco Varela had introduced the concept of autopoiesis in the
early 1970s to conceptualize life, i.e., the aspect that distinguishes what they called a
living from a nonliving ‘‘machine’’ (Varela, Maturana, & Uribe, 1974). They write:
Central to the concept of autopoiesis is the idea that a system is produced and
reproduced by interactive processes among its components. In other words, through
its components the system reproduces itself.
In contrast to allopoietic systems, none of the elements of autopoietic systems are
produced by agents external to the system. All processes of autopoietic systems are
produced by the system itself and all processes of autopoietic systems are processes of
self-production. In this sense, one can say that autopoietic systems are operatively
closed. There are neither elements entering the system from outside nor vice versa.
A system’s operative closure, however, does not imply a closed system model. It only
implies that no operations can enter or leave the system. Autopoietic systems are,
Productive Misunderstandings 135
nevertheless, also open systems: all autopoietic systems have contact with their
environment (interactional openness). Living cells, e.g., depend on an exchange of
energy and matter between themselves and their surroundings, without which they
could not exist. The contact with the environment, however, is regulated by the
autopoietic system; the system determines when, what and through what channels
there is an exchange of energy or matter between itself and the environment
(undoubtedly, there are some external forces that might influence the system directly,
e.g., radioactive radiation, which can destroy parts of the system. These influences,
however, can never determine what operations take place in a system).
Luhmann (1986, 1995) argued that the concept of autopoiesis, if abstracted from
its biological references, could also be applied to other domains, particularly to the
social domain. In contrast to other social scientists who used the concept of
autopoiesis only metaphorically (e.g., Morgan, 1997) or who tried to apply it directly
to the social domain (e.g., Beer, 1980; Robb, 1989; Zeleny & Hufford, 1992),
Luhmann first abstracted it into a general concept on a transdisciplinary level, and
then redefined it as the specific concept of autopoiesis with reference to particular
types of nonbiological systems (Luhmann, 1995; for an overview of different
applications of autopoiesis to the social domain; see Mingers, 1995). Apart from
living systems Luhmann identifies two additional types of autopoietic systems: social
systems and psychic systems (or minds). While living systems reproduce themselves
via biological processes, social systems reproduce themselves via communication
processes, and psychic systems via mental processes. Whereas the elements of living
systems are physical substances, those of social and psychic systems are elements of
meaning. In the following we will concentrate on Luhmann’s theory of social systems.
For Luhmann the elements of social systems are communications (Luhmann, 1986,
1995). Yet, in contrast to the conventional notion of communication as the transfer of
meaning from a sender to a receiver, Luhmann conceptualizes it as the unity of three
components: (1) information, (2) utterance, and (3) understanding. ‘‘Information’’
refers to the question of what is being communicated while ‘‘utterance’’ concerns the
question of how and why it is communicated. Yet, the central component of
communication is ‘‘understanding,’’ which is absent from most other conceptualiza-
tions of communication. Understanding is the distinction between information
and utterance. For a communication to be understood, the information has to be
distinguished from the utterance: what is being communicated must be distinguished
from how and why it is communicated. It is the understanding which determines
the other two components, i.e., the information and the utterance. In this context,
Luhmann (1995, p. 143) writes: ‘‘Communication is made possible, so to speak, from
behind, contrary to the temporal course of the process.’’
Luhmann argues that communication conceptualized as the unity of utterance,
information, and understanding cannot be produced by a human being alone; a single
individual might produce an utterance containing a particular piece of information,
but he or she cannot contribute the element of understanding as well. This means
that it always takes at least two individuals to co-produce this unity. Consequently,
communication is conceptualized as an emergent phenomenon that arises from the
contact between different individuals.
136 David Seidl
communication, in this sense, is a communication that has been accepted by the other
scientific communications as ‘‘true’’ — it is a ‘‘coded truth.’’
The scientific system is operatively closed in that scientific communication is only
produced by the network of other scientific communications. Scientific communica-
tion cannot draw on nonscientific communications in order to substantiate any
scientific claims. For example, a communication that was substantiated with reference
to a newspaper article would be considered unscientific and thus not be incorporated
in the network of scientific communications. The processing of scientific commu-
nications is guided by theories and methodologies that constitute the structures
(or ‘‘program’’) of the scientific system. Theories and methodologies define the
‘‘rules’’ of what constitutes an acceptable scientific communication; i.e., they define
how scientific communications can be related to other scientific communications, and
thus, ultimately, whether or not a particular communication is treated as true or false,
or whether it should be ignored as unscientific. They also determine how to construct
scientific communications from empirical observations. Different theories and
methodologies will lead to different scientific communications. In line with the
concept of autopoiesis, the theories and methodologies are not introduced from
outside but are themselves the product of scientific communications. New theories
and methodologies are developed on the basis of existing theories and methodologies.
Whether or not new theories and methodologies are considered true or false, and thus
whether one can substantiate further scientific communications, depends entirely on
the network of scientific communications (Luhmann, 1990).
As a consequence of this, scientific discourses are necessarily highly ‘‘stylized’’
(Astley & Zammuto, 1992); they construct abstract variables that are meaningful only
in a scientific context as they have mostly very little to do with how the ‘‘same’’
phenomena are treated elsewhere. An example of this is the way in which the concepts
of ‘‘performance’’ and ‘‘success factors’’ are constructed in the management sciences
(March & Sutton, 1997). In addition to the construction of idiosyncratic variables, the
scientific discourse forces a communication also into relating its variables to each
other in an idiosyncratic manner. Thus, within the scientific discourse, a phenomenon
is structured differently from the way in which it would be structured within any other
discourse. To give an example, within the scientific discourse, one assumes explicitly
counter-factual situations and works with ceteris paribus clauses. As Luhmann writes:
The assumption of ceteris paribus is the condition of isolating the objects of research, but like the
presuppositions of model-formation it is a consciously false assumption. Only through false
assumptions can true knowledge be attained. (Luhmann, 1989, p. 81)
An economic communication is, e.g., the placing of an order. In this case the
meaning of the communication for further economic communications is not its
truth or falsity but the payment associated with it. Other economic communications
connect to this communication with regard to its effects on payment. The
communication may be rejected by ensuing communications, if it is considered to
lead to increasing expense (i.e., to be unprofitable), or, conversely, it may be accepted,
if considered to lead to increasing revenues (i.e., to be profitable). Again, what
qualifies as profitable/unprofitable is determined entirely by the economic
discourse itself. As in the case of the scientific system, the processing of economic
communications is guided by specific structures. The structures of the economic
system are budgets and balances. They define the rules for economic communication
and determine what money can be spent for what purposes. Again, these structures
are not introduced from outside but are themselves the product of the economic
communications.
The third type of system that needs to be examined is the organization. Unlike the
other two systems (science and economy), organizations belong to a type that is
very different from that of functional subsystems of society. As described above,
organizations are systems that reproduce themselves on the basis of decision
communications. To appreciate this it is necessary to clarify Luhmann’s concept of
decision (Luhmann, 2000, 2005b). In contrast to other conceptualizations of decision-
making in the literature, for Luhmann decisions are decision communications; it is
not that decisions are first made and then communicated.
Decisions are a very peculiar form of communication: they are ‘‘compact
communications’’ (Luhmann, 2000, p. 185) which communicate their own con-
tingency. In contrast to an ‘‘ordinary’’ communication, which only communicates
a specific content that has been selected (e.g., ‘‘I love you’’), a decision communicates
also — explicitly or implicitly — that there are alternatives that could have been
selected instead (e.g., ‘‘We are buying machine A and not machine B’’). They
communicate not only what has been decided but also that it has been decided. This
has significant implications for the dynamics of decisions. In the transition from
one decision to the next, the uncertainty of the first decision situation — i.e., the
uncertainty about the consequences of the given alternatives — disappears. For the
second decision it is irrelevant what the initial decision situation looked like.
The second decision can take the chosen alternative as a clear point of reference
without having to evaluate the first decision situation; i.e., the first decision has been
‘‘decided’’ and does not have to be ‘‘decided’’ once more. As such, every decision
makes possible extremely complex decision processes by producing stable points of
reference for ensuing decisions.
As in the case of the other two systems described above, the processing of decisions
is guided by particular structures, which Luhmann refers to as ‘‘decision premises.’’
These decision premises define what decisions come about. There are different types
of decision premises. Decision programs or ‘‘plans’’ are such an example: a strategic
plan defines, e.g., a general direction for future decision-making. These decision
premises are, again, not introduced from outside but are the product of the
organization’s decision processes. Decision premises result themselves from decision
140 David Seidl
processes; e.g., strategic plans are the outcome of decision-making processes, which
themselves are guided by other decision premises, such as those concerning decision-
making competences (Luhmann, 2005b).
The three systems that we have described (science, economy, and organization) are
very different in the way they process meaning, as has hopefully become clear by now.
In the following section we will examine what consequences this has for the
possibilities of transferring meaning from science to ‘‘practice.’’
transfer more or less the entire scientific system into the other system. But even if this
was possible, the meaning of the communication in another system would necessarily
differ from its meaning in the original system, as the entire complex would be
interpreted according to a different code.
Drawing on Luhmann’s systems theory, Kieser and Nicolai (2004) described how
the system of science constructs the problems that it analyzes in a self-referential way
that has very little to do with the problems faced by practitioners. This is inevitable
as the problems are, by definition, framed differently in the domains of science and
‘‘practice,’’ even if scientists and practitioners cooperate. Kieser and Nicolai write:
[T]he negotiation of a problem definition [y] has to be seen as a communicative process that
depends on agreeing on a specific frame of reference. In the case of science, frames of reference
are derived from extant theories. In the case of performance studies, sometimes they are triggered
by a problem that plagues practitioners – for example, by the question of whether the existence
of formal procedures of strategic planning are correlated with organizational performance.
Soon, however, the discussion between researchers via their publications creates new and
different problems, and the problem that initiated the scientific discourse gets lost from sight.
(2004, p. 276)
Even when it is the same individuals who ‘‘participate’’ in the scientific and
‘‘practical’’ discourses, they cannot transfer meaning from one discourse to another;
‘‘their’’ communications and actions are determined rather by the logic of the
particular communication systems (Luhmann, 1986). Nicolai (2004) demonstrated
this impressively in his study of Porter’s work, which is widely considered a prime
example of applied research. Rather than crossing the boundaries of the scientific and
the ‘‘practical’’ discourses, the economics-based scientific parts and the ‘‘applied’’
parts of Porter’s work are presented more or less autonomously from each other.
Kieser and Nikolai conclude:
In short, in discourses in which researchers try to establish the validity of theories on the basis of
scientific criteria, science necessarily disconnects itself from discourses in which practitioners
evaluate the usefulness of a concept. Again, the practice-oriented researcher finds himself or
herself thrown back on that self-referential stream of communication that is typical for scientific
discourses and is perceived as detached from the real world. (2004, p. 277)
As we have argued in the last section, a direct transfer of scientific results into practice
is not possible due to the different logics of communication. However, this does
not mean that management science is irrelevant to management practice. Both the
systems of economy, in general, and business organizations, in particular, are
influenced by management science. In this sense one can speak of the ‘‘practical
relevance’’ of science whenever science makes a difference to practice — even though
it does not make the same difference as it makes for science itself. Hence, scientific
knowledge can be said to be of relevance to business organizations if it has
some relevance to decision-making, or more to the point, if it makes a difference to
142 David Seidl
The particular form of relevance for decision-making can vary. For example,
scientific knowledge might contribute to defining the decision situation, deciding
between alternatives or enforcing a decision. Thus, whether a scientific theory is true
or not is irrelevant to ‘‘practice’’; the question is whether the referral to a particular
theory helps achieve certain aims. Luhmann illustrates this with an example from
psychology:
With regard to the question of applicability it is irrelevant whether the Oedipus complex really
exists; what counts is whether somebody who is skilled in identifying it is able to combine
situations and therapies in a successful way. (Luhmann, 1993, p. 323; my translation)
As Teubner explains, one system cannot receive input of meaning from another
system; it merely reconstructs elements of another system according to its own logic.
This internal reconstruction is, however, its very own construct, which is different
from the original one. Luhmann writes:
Non-identical reproduction thus means: a change of meaning through re-contextualization,
through integration into a new neighborhood, through triggering of different associations.
Whether the infused element was true or false quickly loses its relevance. (Luhmann, 1993, p. 330;
my translation)
Linguistic ambiguity [y] gives conceptual terminology great flexibility of application, allowing
words to take on new meanings in the context of a different language game. (Astley & Zammuto,
1992, p. 453; emphasis added)
References
Ashforth, B. E., & Gibbs, B. W. (1990). The double-edge of legitimation. Organization Science,
1, 177–194.
Astley, W. G., & Zammuto, R. F. (1992). Organization science, manager, and language games.
Organization Science, 3, 443–460.
Productive Misunderstandings 147
Baldridge, D., Floyd, S., & Markoczy, L. (2004). Are managers from Mars and academicians
from Venus? Toward an understanding of the relationship between academic quality and
practical relevance. Strategic Management Journal, 25, 1063–1074.
Bateson, G. (1972). Steps to an ecology of mind. New York: Ballantine.
Beer, S. (1980). Preface. In: H. M. Maturana & F. G. Varela (Eds), Autopoiesis and cognition:
The realization of the living (pp. 63–72). Dordrecht: Reidel.
Benders, J., & Van Veen, K. (2001). What’s in a fashion? Interpretive viability and management
fashions. Organization, 8, 33–53.
Brunsson, N., & Olsen, J. (1993). The reforming organization. London: Routledge.
Gibbons, M., Limoges, C., Nowotny, H., Schwartzman, S., Scott, P., & Trow, M. (1994).
The new production of knowledge: The dynamics of science and research in contemporary
societies. London: Sage.
Hodgkinson, G., Herriot, P., & Anderson, N. (2001). Re-aligning the stakeholders in
management research: Lessons from industrial work and organizational psychology. British
Journal of Management, 12, 41–48.
Kieser, A. (1997). Rhetoric and myth in management fashion. Organization, 4, 49–74.
Kieser, A. (2002). On communication barriers between management science, consultancies and
business companies. In: T. Clark & R. Fincham (Eds), Critical consulting (pp. 206–227).
Oxford: Blackwell.
Kieser, A., & Leiner, L. (2009). Why the rigour–relevance gap in management research is
unbridgeable. Journal of Management Studies, 46, 516–533.
Kieser, A., & Nicolai, A. (2005). Success factor research: Overcoming the trade-off between
rigor and relevance?. Journal of Management Inquiry, 14, 275–279.
Kieser, A., & Wellstein, B. (2008). Do activities of consultants and management scientists affect
decision making by managers?. In: G. P. Hodgkinson & W. H. Starbuck (Eds), The Oxford
handbook of organizational decision making (pp. 493–516). Oxford: Oxford University Press.
Luhmann, N. (1986). The autopoiesis of social systems. In: F. Geyer & J. Van d. Zeuwen
(Eds), Sociocybernetic paradoxes: Observation, control and evolution of self-steering systems
(pp. 172–192). London: Sage.
Luhmann, N. (1989). Ecological communication. Cambridge: Polity Press.
Luhmann, N. (1990). Die Wissenschaft der Gesellschaft. Frankfurt a.M.: Suhrkamp.
Luhmann, N. (1993). Theoretische und praktische probleme der anwendungsbezogenen
sozialwissenschaften. In: N. Luhmann (Ed.), Soziologische aufklärung 3: Soziales system,
gesellschaft, organization (pp. 321–334). Opladen: Westdeutscher Verlag.
Luhmann, N. (1994). Die Wissenschaft der Gesellschaft (2nd ed). Frankfurt: Suhrkamp.
Luhmann, N. (1995). Social systems. Palo Alto, CA: Stanford University Press.
Luhmann, N. (2000). Organization und entscheidung. Opladen: Westdeutscher Verlag.
Luhmann, N. (2002). How can the mind participate in communication?. In: W. Rasch (Ed.),
Theories of distinction: Redescribing the descriptions of modernity (pp. 169–184). Palo Alto,
CA: Stanford University Press.
Luhmann, N. (2003). Organization. In: T. Bakken & T. Hernes (Eds), Autopoietic organization
theory. Drawing on Niklas Luhmann’s social systems perspective (pp. 31–52). Copenhagen:
Copenhagen Business School Press.
Luhmann, N. (2005a). Communication barriers in consulting. In: D. Seidl & K.-H. Becker
(Eds), Niklas Luhmann and organization studies (pp. 349–362). Copenhagen: CBS Press.
Luhmann, N. (2005b). The paradox of decision making. In: D. Seidl & K.-H. Becker (Eds),
Niklas Luhmann and organization studies (pp. 85–106). Copenhagen: Copenhagen Business
School Press.
148 David Seidl
MacLean, D., & MacIntosh, R. (2002). One process, two audiences: On the challenges of
management research. European Management Journal, 20, 383–392.
March, J. G., & Sutton, R. I. (1997). Organizational performance as a dependent variable.
Organization Science, 6, 698–706.
Maturana, H. (1978). Biology of language: The epistemology of reality. In: G. Millar &
E. Lenneberg (Eds), Psychology and biology of language and thought: Essays in honour of Eric
Lenneberg (pp. 27–63). New York, NY: Academic Press.
Maturana, H., & Varela, F. (1980). Autopoiesis and cognition: The realization of the living.
Dordrecht: Reidel.
Mingers, J. (1995). Self-producing systems: Implications and applications of autopoiesis.
New York, NY: Plenum.
Mohe, M., & Seidl, D. (2007). The consultant–client relationship: A systems-theoretical
perspective. Munich Business Research No. 2007–2.
Morgan, G. (1997). Images of organization. Thousand Oaks, CA: Sage.
Nicolai, A. (2004). Bridges to the ‘real world’: Applied science fiction or a ‘schizophrenic tour
de force’?. Journal of Management Studies, 41, 951–976.
Ortmann, G., & Salzmann, H. (2002). Stumbling giants: The emptiness, fullness, and
recursiveness of strategic management. Soziale Systeme, 8, 205–230.
Robb, F. (1989). Cybernetics and suprahuman autopoietic systems. Systems Practice, 2, 47–74.
Rynes, S. L., Bartunek, J. M., & Daft, R. L. (2001). Across the great divide: Knowledge
creation and transfer between practitioners and academics. Academy of Management Journal,
44, 340–355.
Rynes, S., McNatt, B., & Breetz, R. (1999). Academic research inside organizations: Inputs,
processes, and outcomes. Personnel Psychology, 52, 869–898.
Seidl, D. (2007). General strategy concepts and the ecology of strategy discourses: A systemic-
discursive perspective. Organization Studies, 28, 197–218.
Seidl, D., & Becker, K. (2006). Organizations as distinction generating and processing systems:
The contribution of Niklas Luhmann to organization studies. Organization, 13, 9–35.
Spencer, J. (2001). How relevant is university based scientific research to private high-
technology firms?. Academy of Management Journal, 44, 432–440.
Starkey, K., & Madan, P. (2001). Bridging the relevance gap: Aligning stakeholders in the
future of management research. British Journal of Management, 12, 3–26.
Teubner, G. (2000). Contracting worlds: The many autonomies of private law. Social and Legal
Studies, 9, 399–417.
Van de Ven, A., & Johnson, P. (2006). Knowledge for theory and practice. Academy of
Management Review, 31, 802–821.
Varela, F., Maturana, H., & Uribe, R. (1974). Autopoiesis: The organization of living systems,
its characterization and a model. Biosystems, 5, 187–196.
Zbaracki, M. (1998). The rhetoric and reality of total quality management. Administrative
Science Quarterly, 43, 602–636.
Zeleny, M., & Hufford, C. (1992). The application of autopoiesis in social analysis:
Are autopoietic systems also social systems?. International Journal of General Systems, 21,
145–160.
Chapter 8
1 Introduction
In examining what role autopoietic theory might play in furthering the agenda of
process-based organizational research, it is worth noting that the biological notion of
autopoiesis and derivative concepts have already achieved limited recognition in the
broad organization studies field. A perennial debate has evolved around the question
of whether organizations can and/or should be considered autopoietic (see Luhmann,
1986; Zeleny & Hufford, 1992; Mingers, 1992; Robb, 1989; Kay, 2001). Beyond that,
the general approach seems to involve taking some defined aspect of autopoiesis
and employing this to shed light on some defined aspect of organizational life.
Thus, Krogh and Roos (1998) use the concept of autopoiesis to expound, discuss, and
illustrate a distinctive perspective on organizational knowledge; Luhmann (1990) and
Teubner (1984) use autopoiesis to create awareness of how the circularity and self-
referentiality of legal, and social systems more generally, can prevent renewal and lead
to a failure in adapting to problems in society. Autopoiesis has been used to enhance
our understanding of how the functioning of computers relate to the evolution of
human language, thought and action, (Winograd & Flores, 1987). In management,
the concept of autopoiesis has been used, largely in a metaphorical sense, to
understand the firm as a living evolving system that is characterized by ‘‘flux and
transformation’’ (Morgan, 1986). In the therapeutic professions, various writers use
autopoiesis to show how circular sets of self-reinforcing conversations can create
severe dysfunctions with individuals (Efren, Lukens, & Lukens, 1990), in families and
in other tightly knit social groups (Dell, 1982, 1985; Hoffman, 1988; Goolishian &
Winderman, 1988). Elsewhere in organization studies, Kay (1997) applies autopoiesis
to the facilitation of organizational change, and Beer (1981) uses the term
‘‘pathological autopoiesis’’ in understanding threats to organizational viability.
Applications of the autopoiesis concept such as these raise important new
questions about organizations and the behavior that takes in and around them.
Despite this, there remains a sense that more could be done to exploit the full
potential of this body of knowledge, that it could, and perhaps should have a higher
profile than it currently does. More ambitiously perhaps, one might even entertain the
thought that autopoietic theory might provide a coherent paradigm for theorizing
and studying organizational phenomena more generally.
Against this background, the main purpose of the chapter is to sketch out what
might potentially be a much more significant role for autopoietic theory. The
argument revolves around aligning it with a major trend that has been underway
within organization theory for some time, a trend that appears to be becoming
increasingly influential. I am referring here to the shift toward process thinking and
the corresponding ‘‘turn to language’’ that is a key aspect of it.
In what follows, the chapter begins by outlining the broad characteristics of
process thinking in organizations, and — within this frame — what is meant by the
‘‘turn to language.’’ It then considers two process and language-based approaches
which have been highly influential in organization studies. These are Foucauldian
discourse analysis and social constructionism. The chapter then turns to autopoietic
theory with a view to showing how it can build upon and extend these approaches
and provide answers to questions that remain theoretically underdeveloped in these
other approaches.
If the number of papers, edited works, symposia, workshops, and conferences devoted
to this issue is any guide (see, e.g., Gergen, 1992; Hosking & McNamee, 2002; Townley,
1993; McKinlay & Starkey, 1998; Burr, 1995; Knights & Morgan, 1995; Shotter, 1995;
Wood et al., 2006), there has been a major renewal of interest in organizational process
thinking and, within this, in the specific role played by language. With process thinking,
it is clear that this has a significantly different meaning today than has been the
case previously. In the past — often within a broad ‘‘systems’’ perspective — this
has generally been taken to be about the various financial, material, human, and
informational flows that take place within the boundaries of the firm that sustain the
transformation of a defined set of environmental inputs into outputs. All of this is done
in support of some purpose. On this account, the organization, its boundaries, and the
phenomena that occur within and beyond these are taken as preexisting givens. There is
a strong emphasis on explaining such phenomenon in terms of cause–effect logic with
a view to maximizing control in support of taken-for-granted organizational goals.
Classical management theory, the functionalist perspective in organization theory, and
the so-called ‘‘hard’’ systems approach all reflect these sorts of concerns.
Plugging the Theoretical Gaps 151
In a manner that is somewhat reminiscent of how autopoietic theory has been applied
in the field of organization studies, the perspective known as ‘‘discourse analysis’’ —
based on Foucault’s work — has been appropriated for fulfilling a diverse range of
tasks in the same field, (see, e.g., Boje, 1994; Burrell, 1988; Clegg, 1994; Fox, 2000;
Knights & Morgan, 1991; Townley, 1993). Fundamentally, and in an organizational
context, these publications build upon and illustrate Foucault’s problematization of
the relationship between language and action. This takes discourse analysis to be a
practice which enables transformation not only in the ‘‘economy of discourse,’’ in
terms of regimes of truth, but also ‘‘in the administration of scarce resources’’
(Foucault, 1972, p. 120). In other words discourse analysis is a political matter.
It deals with the relationship between language and power relations and thereafter
with access to material resources.
Foucault specifically refuses ‘‘analysis couched in terms of the symbolic field or the
domain of signifying structures’’ in favor of an analytic model of war or battle:
‘‘relations of power, not relations of meaning’’ (Foucault, 1980, p. 114). Thus, he is
interested in questions of what communication does rather than what it means.
As Shapiro (1981) points out, Foucault’s concept of discourse is inevitably strategic.
That is, irrespective of whether or not it is intended, discourse has political effects.
The implication for organization scholars is that an analysis of organizational
discourse is inseparable from an analysis of power relations. By extension, definitions
of discourse are inseparable from wider social processes.
Plugging the Theoretical Gaps 153
4 Social Constructionism
For many years, organizational research has drawn on the contributions of social
psychology. However, following the 1966 publication of Berger and Luckmann’s
seminal ‘‘The Social Construction of Reality,’’ one of the traditions in that discipline,
social constructionism, has become particularly influential (see, e.g., Gergen, 1985,
1995; Shotter, 1997a, 1997b).
Despite its social psychology origins, social constructionism is now an inter-
disciplinary approach that includes Wittgenstein’s (1953) work in philosophy; the
contribution of ethnomethodologists such as Garfinkel (1984, 1996) and post-
structuralists such as Derrida (1976) and Deleuze (2001). In essence, social
constructionism moves beyond the traditional dualism that characterizes the debate
between subjectivism and objectivism to focus on the process of social interchange.
The (experiential) terms by which the world is understood are regarded as social
artifacts, products of historically situated interchanges among people. Within the
parent discipline of psychology, this represents a significant move from an individual
epistemology to a social one.
In social constructionism, the focus on the individual ‘‘mind’’ is replaced by a focus
on relationships. And because the locus of explanation shifts from the mind to the
processes of human interaction, social constructionist research is mainly concerned
with explicating those processes by which people account for the world in which they
live (see, e.g., Gergen, 1995; Bayer, 1988). Its emphasis is on the way we negotiate
meanings in our lives, and its practices stress how language fashions subjective
experience. In organizational research, and mirroring developments in psychology,
this relational emphasis on language has signaled a departure from an individualist
perspective of organization to a communal one, or, to draw on Gergen’s (1999) terms
‘‘an alternative discourse to the discourse of the self-contained individual.’’ Harre and
Gillet (1994) discuss how the pragmatics of everyday language serve to render certain
types of social relationships more salient than others. Gergen also relates this
154 John Brocklesby
Shifting attention away from the basic mechanism of interaction between systems in
general, let us now consider, more specifically, those that involve human beings and,
in particular, which involve, in some manner or other, language.
We have already noted some key assumptions about language that are shared by
the three perspectives discussed here. First, it is the primary mechanism through
which we construct or constitute and explain our worlds; second, it is something that
we do with others, i.e., language develops intersubjectively; and third, it is linked
to community practices, i.e., it is about actions and ‘‘doings,’’ it is not an abstract
symbolic system of communication about a preexisting world.
Conventional wisdom has it that the fundamental aspect of language is the speech
act itself (see, e.g., Searle, 1983). In autopoietic theory, the fundamental aspect is
action; vocabulary and speech are derived from action but come later in the piece.
On Maturana’s view, languaging is associated with particular types of behavior, or
particular ‘‘doings.’’ The most basic operation of this occurs when some entity does
something on the consequences of an initial coordination of behavior between it and
another entity which, by definition, must be structurally coupled, at least to some
minimum degree (see Maturana, 1988a). In other words, there is a ‘‘coordination of a
coordination of behavior.’’ From this basic behavioral coordination, further recursions
of behavioral coordination result in language becoming increasingly complex and
sophisticated. Thus, in human communities such as organizations, objects ‘‘arise’’ as
tokens for highly specific behavioral coordinations.
In his classes and public seminars, Maturana uses a number of examples to
illustrate this process. The designation ‘‘taxi,’’ for instance, connotes the coordinated
sequence of actions that are involved in carrying someone from one place to another
(usually in a motor vehicle) in return for the payment of money. Similar coordinated
actions underpin ostensibly simple designations such as ‘‘desk,’’ ‘‘computer,’’ as well
as more complex ones such as ‘‘the annual retreat,’’ ‘‘the board,’’ ‘‘the retirement
function,’’ ‘‘the brand,’’ etc. The same can be said of more abstract entities such as
‘‘fairness,’’ ‘‘equity,’’ ‘‘integrity,’’ ‘‘ethicality,’’ and so on. Across different organiza-
tions and different cultures, these are all anchored in highly specific and recursive
coordinated behaviors.
Plugging the Theoretical Gaps 159
Having arisen through this process, we often lose sight of how such ‘‘objects’’ are
anchored in actions. Thus: ‘‘y objects take place as distinctions of distinctions that
obscure the co-ordination of actions that these co-ordinate’’ (Maturana, 1988a, p. 47).
At this point in time, ‘‘the taxi,’’ ‘‘the CEO,’’ ‘‘the brand,’’ and everything else become
entities ‘‘in-themselves,’’ which, of course, is how we tend to live them.
Although language might appear to us as being about symbols that represent
‘‘things out there,’’ fundamentally it is about doing. Entities and objects correspond
to ‘‘doings,’’ and language is a flow of coordinations of coordinations of action.
Language is about communal action and is a concrete phenomenon of the living. As
we coordinate our behaviors in different ways; as, through various forms of linguistic
interaction, we make new distinctions; and as we come up with new tokens for specific
behaviors, we are continually weaving linguistic networks with other people.
So much for the basic building blocks of language. As observers, we do more than
simply operate with simple one-word descriptors of ‘‘things’’; we combine these
descriptors in highly creative communicative processes that, in the process literature,
are variously known as conversation, narratives, discourse, ‘‘storylines,’’ ‘‘text,’’ and
so on and so forth.
Whereas Foucauldian analysis and social constructionism are somewhat vague on
what exactly these things involve, Maturana has theorized the matter in some detail.
On his account, the key concept is what he refers to as ‘‘conversation.’’ This involves
two processes: first, ‘‘languaging,’’ which we have already discussed; second, what he
describes as ‘‘emotioning.’’ Emotioning refers to ‘‘bodily dispositions for actions’’
(Maturana, 1988a, p. 42). Such predispositions can be both individual and shared, so
on this account, emotioning is a social construct as well as an individual one.
Languaging and emotioning are ‘‘braided’’ processes, i.e., each process can
affect the other. This can be seen in everyday conversations where the specific
distinctions that people use can invoke an emotional response, and/or where people’s
distinctions reflect the emotion or predisposition of the moment. This latter point is
very important. The distinctions that we use to explain and understand our worlds are
never isolated from bodily processes on the one hand and from social processes on
the other.
Importantly, as conversations flow through an interweaving of distinctions and
emotions, the interpretations and meanings that people attach to situations can alter.
How people construe and feel about situations, and how they subsequently act in
relation to a situation depends — at least to some extent — on the specific distinctions
that they employ and on the flow of their emotions from moment to moment.
Typically, this is something that we recognize in our own daily life experiences and the
basic idea is acknowledged in process research. How one goes about defining
(and then researching) whatever it is that generates ‘‘meaning,’’ however, is less clear.
There are of course some exceptions where this matter has been theorized in
some detail. A good example is Kelly’s so-called ‘‘Theory of Personal Constructs’’
160 John Brocklesby
which although relatively autonomous as discourses may overlap and permeate each
other. What is less clear is how this ‘‘overlapping’’ works.
Maturana would agree that in organizational settings, many of our structural
couplings are with other members of the organization and many of the conversations
that take place there are relatively independent. Some may even be unique to the
specific organizational context. However, as human beings, we all have multiple
structural couplings and we participate in an infinitely large number of different
conversations. Each one of these has its own braided flow of distinctions and
emotions that have been learned through recurrent interactions over time and which
alter subject to the dynamic flow of the conversation. This means that just as thoughts
and descriptions within a single conversation are subject to change depending upon
the flow of the conversation, they can also alter as the observer shifts from one
conversation to another.
But how do these conversations interact? Visually this can be explained using the
image of a wheel where there is a hub and a large number of intersecting spokes.
The spokes represent intersecting conversations; the hub represents the ‘‘bodyhood’’
of the observer. The mechanism is then very simple: if the structure of bodyhood
allows it, each conversation — subject to the principle of structure determinism —
will trigger structural change. This then alters the way the observer participates in
other conversations. Through this theoretically very simple process, we can explain
a number of important observations, e.g., how one conversation can affect another,
how language can create change and learning, why some conversations are more
powerful than others, and so on.
This mechanism also draws attention to another possible theoretical contribution
of autopoietic theory which I shall now consider. This has to do with the relationship
between the individual and the social.
As we have seen, the various bodies of thought discussed here assert that the process
of creating meaning is primarily a social activity. For social constructionism, this
has signaled a major departure from an individualist perspective of meaning to a
communal one. This contrasts with the strongly individualistic emphasis in many
Western cultures and traditional views of knowledge that have invited people to see
themselves as the center of their actions in a natural state of independence.
However, while this shift in emphasis from the individual to the social is
understandable, if it is taken to imply that the individual is relatively unimportant, it
does raise some interesting questions. For example, if we accept that organizational
and other communication processes do intersect and affect one another, then we
have to explain how this happens. On their own, commonsense logic dictates
that communication processes such as conversations, narrative, and text cannot
‘‘talk to one another’’; so there has to be some mediating mechanism. That can only
be the individual observer. Moreover, even if we do assert the social basis of language,
something, other than itself, has to make language possible in the first place and
162 John Brocklesby
Biology then impacts upon but does not determine what happens in the relational
domain. At the same time, relational encounters impact upon but do not directly
imprint themselves upon biology. At one extreme, an acrimonious exchange between
two people may increase the blood pressure of the people concerned. Longer term,
if the exchanges are repeated, it could trigger cardio-vascular disease; one of the
participants might even suffer a heart attack and die. At the other extreme, using
various meditation techniques and through processes of ‘‘self talk,’’ someone might
learn how to reduce muscular and nervous tension to good long-term effect. In both
cases, there is a structural transformation of the system in line with changes in
relational circumstances and in the medium itself. Indeed this is the basis for
evolutionary development and learning (see Maturana, 1988a, p. 74; Winograd &
Flores, 1987, pp. 44–47). However, such changes are always subject to the system’s
structure determinism.
This is a never-ending iterative process since the changed structure then generates
the parameters which govern subsequent social behavior. This elaboration of the
nature of the relationship between the individual and the social is a highly distinctive
feature of autopoietic theory.
5.5 Power
Let me now turn to what can be said about autopoietic theory in the light of the
strong emphasis that is placed on power in Foucauldian discourse analysis. While
Foucault’s fundamental intertwining of discourse and power is not something that
Maturana discusses per se, the overall conceptual framework, key aspects of which
have been outlined above, does open theoretical space for such an explanation.
In simple terms, the link between power and meaning hinges first on the proposition
that conversations — and languaging and emotioning more specifically — occur in
networks of structural coupling. Recall that this concept asserts a dynamic form of
adaptation or mutual adjustment between a living system and a medium, or between
two or more living systems. In this case, the most obvious and relevant example is
the structural coupling between two or more human beings. However, the same logic
could be applied, as it is in Foucault’s writings, to the relationship between different
fields of inquiry, disciplines, and institutions.
Irrespective of the context, it would be erroneous to presume that this coupling is
an equal one, so the potential for a power imbalance always exists. In the vernacular
it could be said that one system ‘‘calls the shots,’’ so where there is ‘‘adjustment,’’
inevitably one party will adjust more than the other. Such logic applies to all
interacting systems; potentially, there is always one party that is relatively dominant,
one that is relatively subservient.
Thereafter, recall that conversations reflect the exigencies of the structural
couplings; powerful actors are able to initiate some conversations and close down
others; powerful actors can choose to invite or not invite others to participate
(i.e., become structurally coupled); powerful actors can seize control over the
distinctions that are employed and control the ‘‘emotional flow’’ by ‘‘controlling the
164 John Brocklesby
agenda’’ and otherwise shaping the various predispositions that open up some
possibilities and close down others. Finally, since people’s meanings and explanations
arise through the various couplings and the conversations that take place there, it is
logically quite straightforward with autopoietic theory to craft a link between power,
meaning, and knowledge.
Although this inextricable linking of power and meaning needs to be taken very
seriously, one wonders why Foucauldian discourse analysis elevates power relations
to the level that it does. Recall that conversations involve a braiding of emotioning
and languaging, which means that people’s distinctions reflect emotions more
generally, not predispositions to subjugate, dominate, and/or control in particular.
In that sense one could claim that it is emotion that is the primary ‘‘always-already’’
condition; it is not, as Foucault appears to suggest, power.
As I have already said, how human beings think, how they act, and what they do
partly depends on their emotional flow, and some of this reflects biological dynamics.
At the same time, emotional predispositions also arise in coexistence with others as
we go about living together. For the researcher who seeks to reveal the implicit
predilections and preferences that are embodied in organizational cultures, policies, and
practices, the possibility arises that, alternatively, they may be grounded in emotions
such as love, mutual respect, friendship, indifference, or in some other disposition.
The researcher does not necessarily have to be swayed by the profound pessimism of
Foucault which points toward the negative and hegemonic impact of totalizing
managerial discourses. Foucault’s seminal work on prisons reminds us that many of
our organizations do indeed share some of the characteristics of such institutions. But
some do not, and there are no compelling reasons why organizational researchers
should formulate projects on the premise that power is at the root of everything.
6 Conclusion
Despite its popularity in some quarters, autopoietic theory has not achieved the sort
of profile in the field of organization studies that one might have expected. Typically it
has been employed to shed light on specific issues that are of interest to a limited
number of organizational scholars. To date it has not been opened up to a wider
potential audience through being aligned with a more general organizational research
paradigm. This chapter has suggested that the recent turn toward a process and
language-based approach in organizational studies provides fertile ground for
autopoietic theory to play a more central role. Specifically, I have argued that it
can plug some important theoretical gaps that exist in this literature. In particular, the
chapter has sought to demonstrate that autopoietic theory can be helpful in specifying
a number of important things more precisely than is the case elsewhere. These include:
the process through which human beings interact, the process through which they
influence one another, and through which they change over time; the action and
behavioral basis of language; the nature of conversations and dialogue and the role
that these play in generating meaning; and the relationship between the individual
Plugging the Theoretical Gaps 165
and the social. In addition, autopoietic theory is able to contextualize the relationship
between power, meaning, and action by showing how power may not always be the
dominant emotional predisposition that governs social relations.
References
Barrett, F. J., Thomas, G. F., & Hosevar, S. P. (1995). The central role of discourse in large-
scale change: A social construction perspective. Journal of Applied Behavioral Science, 31(3),
352–372.
Barry, D., & Elmes, M. (1997). Strategy retold: Toward a narrative view of strategic discourse.
Academy of Management Review, 22(2), 429–452.
Bayer, B. (1988). Re-enchanting constructionist inquiries. In: B. Bayer & J. Shotter (Eds),
Reconstructing the psychological subject (pp. 1–20). London: Sage.
Beer, S. (1981). Brain of the firm (2nd ed.). Chichester: Wiley.
Birch, J. (1991). Re-inventing the already punctured wheel: Reflections on a seminar with
Humberto Maturana. Journal of Family Therapy, 13, 349–373.
Blackler, F., Crump, N., & McDonald, S. (2000). Organizing processes in complex activity
networks. Organization, 7(2), 277–300.
Boje, D. (1994). Stories of the storytelling organization: A postmodern analysis of Disney as
Tamara-land. Academy of Management Journal, 38(4), 997–1035.
Burr, V. (1995). An introduction to social constructionism. New York, NY: Routledge.
Burrell, G. (1988). Modernism and PostModernism II — The Contribution of Michel.
Chia, R., & Holt, R. (2006). Strategy as practical coping: A Heideggerian perspective.
Organization Studies, 27(5), 635–655.
Chia, R., & MacKay, B. (2007). Post-processual challenges for the emerging strategy-as-
practice perspective: Discovering strategy in the logic of practice. Human Relations, 60(1),
217–242.
Clegg, S. (1994). Weber and Foucault: Social theory for the study of organizations.
Organization, 1(1), 149–178.
Cunliffe, A. L. (2001). Managers as practical authors: Reconstructing our understanding of
managerial practice. Journal of Management Studies, 38(3), 351–371.
Deleuze, G. (2001). Difference and Repetition. London: Athlone.
Dell, P. (1982). Family therapy and the epistemology of Humberto Maturana. Family Therapy
Networker, 6(4), 39–41.
Dell, P. (1985). Understanding Bateson and Maturana: Toward a biological foundation for the
social sciences. Journal of Marital and Family Therapy, 11(1), 1–20.
Derrida, J. (1976). Of grammatology. Baltimore, MD: John Hopkins University Press.
Eden, C. (1992). Strategy development as a social process. Journal of Management Studies,
29(6), 799–811.
Efran, J., Lukens, M., & Lukens, R. (1990). Language, structure, and change-frameworks of
meaning in psychotherapy. New York, NY: W.W. Norton and Co. Inc.
Foucault, M. (1972). The archaeology of knowledge. London: Routledge.
Foucault, M. (1980). Truth and power. In: C. Gordon (Ed.), Power/knowledge: Selected
interviews and other writings, 1972–1977 Michel Foucault. New York: Pantheon.
Foucault, M. (1983). Afterword: The subject and power. In: H. Dreyfus & P. Rabinow (Eds),
Michel Foucault: Beyond structuralism and hermeneutics (pp. 208–226). Chicago, IL:
University of Chicago Press.
166 John Brocklesby
Fox, S. (2000). Communities of practice, Foucault, and actor network theory. Journal of
Management Studies, 37(6), 853–868.
Garfinkel, H. (1984). Studies in ethnomethodology. Cambridge, MA: Polity Press.
Garfinkel, H. (1996). Ethnomethodology’s program. Social Psychology Quarterly, 59(1), 5–21.
Gergen, K. (1985). The social constructionist movement in modern psychology. American
Psychologist, 40(3), 266–275.
Gergen, K. (1992). Organization theory in the postmodern era. In: M. Reed & M. Hughes
(Eds), Rethinking organization (pp. 207–226). London: Sage.
Gergen, K. (1995). Global organization: From imperialism to ethical vision. Organization, 2,
519–532.
Gergen, K. (1999). Social constructionism and relational action. Theory and Psychology, 9, 113–115.
Gergen, K., & Thatchenkery, T. (1996). Organization science as social construction:
Postmodern potentials. Journal of Applied Behavioral Science, 32(4), 356–377.
Goolishian, H. A., & Winderman, L. (1988). Constructivism, autopoiesis and problem
determined systems. Irish Journal of Psychology, 9(1), 130–143.
Harre, R., & Gillet, G. (1994). The discursive mind. Thousand Oaks, CA: Sage.
Hoffman, L. (1988). A constructivist position for family therapy. Irish Journal of Psychology,
9(1), 110–129.
Hosking, D. M., & McNamee, P. (2002). The social construction of organization. Copenhagen:
Liber Herndon.
Kay, R. (1997). Applying autopoiesis to the facilitation of organizational change. Australian
and New Zealand Systems Conference, Brisbane.
Kay, R. (2001). Are organizations autopoietic? A call for a new debate. Systems Research and
Behavioral Science, 18(6), 461–477.
Kelly, G. (1955). The psychology of personal constructs: A theory of personality. New York, NY:
Norton.
Knights, D., & Morgan, G. (1991). Corporate strategy, organizations, and subjectivity:
A critique. Organization Studies, 12(2), 251–273.
Knights, D., & Morgan, G. (1995). Strategy under the microscope: strategic management in IT
in financial services. Journal of Management Studies, 32(2), 191–214.
Krogh, G. V., & Roos, J. (1998). Knowing in firms: Understanding, managing and measuring
knowledge. London: Sage.
Luhmann, N. (1986). The autopoiesis of social systems. In: F. Geyer & J. V. D. Zouwen (Eds),
Sociocybernetic paradoxes: Observation, control, and evolution of self steering systems
(pp. 172–192). London: Sage.
Luhmann, N. (1990). The cognitive program of constructivism and a reality that remains
unknown. In: W. Krohn, G. Kuppers & H. Nowotny (Eds), Selforganization: Portrait of a
scientific revolution (p. 270). Dordrecht: Kluwer.
Maturana, H. (1970). Biology of cognition. In: H. Maturana & F. Varela (Eds), Autopoiesis and
cognition. Dordrecht: Reidel.
Maturana, H. (1980). Man and society. In: F. Benseler, P. M. Hejl & W. K. Kock (Eds),
Autopoiesis, communication, and society: The theory of autopoietic systems in the social
sciences (pp. 11–32). Frankfurt: Campus Verlag.
Maturana, H. (1988a). Reality: The search for objectivity or the quest for a compelling
argument. Irish Journal of Psychology, 9, 25–82.
Maturana, H. (1988b). Ontology of observing: The biological foundations of self consciousness
and the physical domain of existence. American Society for Cybernetics, Felton, CA.
Available at: http://www.inteco.cl/biology/ontology/index.htm
Maturana, H. (1993). Videotaped seminar on language and cognition. Melbourne: St. Kilda.
Plugging the Theoretical Gaps 167
An Autopoietic Understanding
of ‘‘Innovative Organization’’
Tore Bakken, Tor Hernes and Eric Wiik
1 Introduction
to the question why new structures appear. Such assumptions has led to claims that
Luhmann’s theory is relevant only to the study of routines and not to innovative
processes (cf., Beckert, 1997, p. 347), and that it prevents a satisfactory understanding
of the phenomenon of innovation (Joas, 1992; Giddens, 1996).
We would argue differently, and say that autopoietic theory offers a way of
conceptualizing how systems reproduce themselves in the face of novelty, further that
it is the expected possibility of connecting to novelty that drives systems forward. The
possibility of novelty is a central part, both of reproducing central features, and
producing features for future operations. Possibilities for novelty arise as systems, as
part of their recursive reproduction, draw distinctions amid a changing environment.
The system reproduces itself recursively, pointing forward to possible connections,
and at the same time connecting to previous operations. It is in this sense that a
system may be understood as a ‘‘historical machine’’ (cf., von Foerster, 1993), or a
‘‘system-in-an-environment-with-a-history’’ (cf., Luhmann, 2005, p. 25).
We would argue that an autopoietic theory of organization is in fact also a theory of
innovation. Without the possibility of novelty, autopoietic organization is hardly
possible. A second argument we make in this chapter is that, contrary to much of
the literature on organization and innovation, an autopoietic view does not consider
the degree to which innovation takes place. Instead it considers how the nature of
communication shapes expectations, thus influencing the search for novelty. If we
assume that different functions within an organization operate according to different
modes of communication, we may come to a different understanding of how the
organization engages with novelty. Key to this understanding is that different
organizational functions operate with different degrees of redundancy in their
communication.
2 Key Concepts
We base our discussion on selected concepts from Luhmann’s autopoietic theory.
A key concept is expectations which Luhmann developed from early works in the
psychological and sociological literature. One feature of self-referential systems is that
they generalize meanings of their elements into expectations:
The concept of expectations points to the fact that the referential structure of meaningful objects
or themes can only be used in a condensed form. Without this condensation the burden of
selection would be too great for connecting operations. Expectations are formed by the
intervening selection of a repertoire of possibilities, by whose light one can orient oneself better
and, above all, more quickly. (Luhmann, 1995, p. 96)1
1
What Luhmann is looking for is how to generalize expectations, but not in Parsons’ fashion where there is
a strong link between norms and expectations which stabilizes the expectations. For Luhmann generalizing
constrains what is possible, while making visible other possibilities. In this way, Luhmann applies a new
term which he calls’’redundant complexity’’ as an alternative to the more well-known term ‘‘organized
complexity.’’
An Autopoietic Understanding of ‘‘Innovative Organization’’ 171
Expectations serve to drive the system forward while connecting the system
with its past. Expectations are proper to the system and rooted in its past, they
take place in the present; what were at one point in time expectations for the future
become past experiences at a later point in time. They do not, however, contain any
certain predictions, expectations are tentative. As Luhmann (1995, p. 110) suggests, a
social system is an autopoietic system and hence there is no basal certainty about
states or prediction of behavior to be built on. There is not control of what will
happen.
Social systems operate in a world of double contingency, which implies that
the system will never have complete knowledge of what the reactions will be to its
own actions.2 The problem of double contingency was raised by Parsons (1951),
who worked from the assumption that it would be solved by shared norms
between actors.3 Luhmann argued that the assumption of shared norms was
inadequate for explaining how systems deal with double contingency. Under
uncertainty, instead, systems stabilize expectations (Luhmann, 1995, p. 110), and,
depending on the outcomes, they will correct their actions within their own repertoire
of behavior.
A second concept is that of communication.4 A social system, being based on
actions, needs communication in order to sustain itself, as communication provides
meaning to actions. In Luhmann’s own words, communication serves to inundate the
system (of actions) with meaning. It is a kind of ‘‘self-excitation’’ (Luhmann, 1995, p.
171) that helps the system sort order from noise, and thus cope with the challenges
posed by double contingency. Not only does communication help the system cope with
the problem of contingency, but it also sensitizes the system to unexpected and
disturbing events within the repertoire of understanding that constitutes the system
(Luhmann, 1995, p. 172). It is the ways in which communication sensitizes the system
to unexpected and disturbing events that in this chapter we see as characterizing the
system’s engagement with novelty, which is sometimes called ‘‘innovation’’ in the
organization literature.
A third concept that we draw upon is that of redundancy. In Luhmann’s
conception, social systems are seen as unlikely phenomena in the sense that they have
to constantly produce and reproduce themselves. They are based on operations which
disappear as soon as they are produced, and consequently depend on their own ability
to reproduce themselves. The system’s ability to reproduce itself lies not in repetition,
but rather in connectivity. In an uncertain world, connectivity entails connecting to
something that is yet to occur, and then cope with the consequences of that
2
Although Luhmann concedes that a completely indeterminate situation (i.e., ‘‘pure’’ double contingency)
never occurs in our societal reality (Luhmann, 1995, p. 119).
3
In fact, this was also offered as an explanation in systems theory (Emery & Trist, 1965) for what happens
between organizations and their environments, when uncertainty and the rate of change become too high
for the organization to keep pace with the environment.
4
Communication consists of information, utterance, and understanding. Information is understood as ‘‘a
difference that makes a difference’’ (Bateson, 1972); utterance refers to the how and why of the
communication; and understanding is the making of a distinction between information and utterance.
172 Tore Bakken et al.
occurrence. The ability of a system to cope with the unexpected depends on the degree
of redundancy that it contains. Redundancy is a concept from systems theory that
Luhmann applied to communication. Redundancy implies a surplus of informational
possibilities and thus protects the system from the danger that something will be lost
(Luhmann, 1995, p. 171). Redundancy may be interpreted as a surplus of connectivity
enabled by communication.
A fourth concept that we draw upon from Luhmann’s autopoietic theory is that of
themes. Social systems, according to Luhmann, develop thematic structures to which
contributions are made. Themes define what may or may not be acceptable
contributions. Thus, themes define what is considered important or not important. In
this chapter we apply the notion of themes to different areas of activity within an
organization. More precisely we work from the idea that areas of strategy,
administration, and technology development represent examples of different themes,
each exhibiting different characteristics of communication.
With few exceptions (e.g., Van de Ven, Polley, Garud, & Venkatamaran, 1999), works
on innovation and organization tend to consider innovation to take place to a smaller or
lesser extent. Thus, some writers distinguish ‘‘radical innovations’’ (Leifer et al., 2000)
and others distinguish ‘‘disruptive technologies’’ (Bower & Christensen, 1995) from the
supposedly more mundane attempts made at exploring novelty in organizations. Thus,
organizations are seen as being more or less innovative. Whereas some organizations are
seen to be endowed with qualities enabling them to beat the competition by staying
ahead, others are characterized as less daring and willing to engage with novelty.
Such views of innovation find their basis in an ontological view of ‘‘organizations’’
that has dominated in organization theory for decades. This assumption in
organization theory has given rise to views of innovation as being a matter of
degree, thus paving the way for assuming that one could correlate basic features of an
organization with its ability to innovate.
Based on older forms of systems theory, traditional organization theory has assumed
a boundary between organization and environment, which has been assumed to be
more or less open. The underlying assumption has been that the environment and the
organization act upon one another from either side of this ‘‘line’’ or organizational
boundary. But more often than not, the organization has been understood to be the
passive part that has to align itself to the demands in the environment (Weick, 1979;
Tsoukas & Chia, 2002). Thus, expressions such as ‘‘strategic fit’’ have made their entry
into the terminology of organizational understanding. The view has implications for
how change is understood and studied. In particular, correlation presupposes
correlation of states of the organization and the environment (Hernes, 2008). In other
words, the state of the organization is compared to the state of the environment at the
same time. Consequently, organizations in rapidly changing environments have been
expected to follow suit if they are not to be threatened with extinction.
An Autopoietic Understanding of ‘‘Innovative Organization’’ 173
The experience of something as new and the assertion that something is new, consequently only
indicates a decision to use up to now redundant possibilities for structural formation. The label
‘‘new’’ is thus an element of the system’s self description. It emphasizes discontinuity in order to
demolish traditions and to reorganize connective capability. Novelty, then, is always about a
system’s relation to itself, and therefore also to the old. (Our translation)
engages in. Instead it makes more sense to ask how the organization engages with the
novelty that presents itself as a possibility, but without ‘‘losing itself.’’
This is where literature on ‘‘innovative organization’’ is in need of alternative
thinking. It is noticeable how the literature describes organizations as exhibiting basic
features making them more or less innovative. Thus, for example, bureaucratic
features are seen as making an organization generally adverse to innovation. An early
contribution in this respect is Victor Thompson’s (1965) paper on how bureaucratic
features of organizations correlate negatively with their capacity for innovation. The
ability of engaging with novelty has also been correlated with traits of an
organization’s culture (e.g., Peters & Waterman, 1981; Tidd et al., 2001), absorptive
capacity (Cohen & Levinthal, 1990; Zahra & George, 2002), or dynamic capabilities
(Eisenhardt & Martin, 2002; Winter, 2002).
they are not fulfilled. On the other hand, if fulfilled and hence confirmed, they become
all the more entrenched and eventually lead to an excessive bias of expectations. In
tightly connected systems, precise expectations are not able to incorporate complexity
and unpredictability from the outside world. Luhmann’s point is that precision in
expectations may actually make the system inaccurate5 because precision makes it
more likely that expectations are not met, thus engendering disappointment, which
may have dire consequences for the system.
An autopoietic reading of the 3M example above might focus on the projection of
expectations rather than the ingenuity of Fry or any inherent organizational
characteristic. Fry’s expectations were precise, in the sense that only the right sort of
glue would enable the realization of Post-it, and thus add novelty to the
organization’s line of products. If his precise expectations had not been met (by
chance in this case), the Post-it would probably not have been launched, at least not
there and then.
The question that poses itself is how the communication that frames expectations
influences the ways in which organizations deal with novelty. We will assume that the
distinguishing features of different functions within an organization are given by their
modes of communication, and especially how they incorporate different degrees of
redundancy in their communication. More precisely we consider the degree of
redundancy in communication to influence the way in which novelty is engaged with.
As a tentative metaphor, consider an adjustable beam of light from a torch. When set
in a wide mode, the beam lights up more in the vicinity of the torch, its circumference
is large, and it captures a higher number of objects than if it is set on ‘‘narrow.’’ It is
less focused in the ‘‘broad’’ mode than in the ‘‘narrow’’ mode. In the narrow mode, on
the other hand, objects can be seen that are further away from us, but at the expense
of the number of objects that can be lit up in the vicinity of the torch. Let us assume
that the torch was used to find a specific object and not merely to light up around the
holder of the torch. In a ‘‘broad’’ mode, the light would help us find objects that we
half expected to be there. They were close to us anyway, so finding them was not such
a great surprise. In the ‘‘narrow’’ mode, however, the object we look for are further
away, and the risk of not finding them is greater due to the narrower beam. On the
other hand, finding them is associated with greater satisfaction because the search was
more exclusive and focused. We will return to this example in the last part of the
chapter.
Broadly speaking the communication process relies on two factors for the
reproduction and change in social systems: First, for any communicative system to
uphold itself, it relies on redundancy of communication. Redundancy in this context
means that the system produces more than it needs for its actual operation. This is
5
‘‘As a rule, the more explicit the expectation, the more insecure it is’’ (Luhmann, 1995, p. 308).
176 Tore Bakken et al.
necessary, because systems that produce for their actual needs only become extremely
fragile in a changing environment. Thus, systems include in their operations the
conditions for potentialities. Systems operate amid possibilities, and its communica-
tions contain references to possibilities which cannot all be pursued at the same time:
There is always a core that is given and taken for granted which is surrounded by references to
other possibilities that cannot be pursued at the same time. Meaning, then, is actuality
surrounded by possibilities. The structure of meaning is the structure of this difference between
actuality and potentiality. Meaning is the link between the actual and the possible; it is not one or
the other. (Luhmann, 1990, p. 83)
6
Themes allow communication to produce difference. Communication inevitably produces difference, one
reason being that the object of the communication changes (the world moves on whether we like that or not;
thus communication about something is like shooting at a moving target), another reason being that what is
‘‘out there’’ can only be guessed at; it lies there as a half-hidden possibility.
An Autopoietic Understanding of ‘‘Innovative Organization’’ 177
7
Early systems theory perspectives of organizations were concerned with questions such as how members of
groups could maintain their respective roles in the group while being able to replace one another. Emery
(1977) suggested that redundancy may be built into systems in basically two different ways. One type of
redundancy, suggested Emery, is the redundancy of parts, meaning that if one part fails, another one takes
over. Another type of redundancy, he suggested, is the redundancy of functions, meaning that the function
may be fulfilled by different parts depending on the needs of the system.
178 Tore Bakken et al.
8
These themes correspond in principle to Thompson’s levels, which he related to different degrees of
openness.
An Autopoietic Understanding of ‘‘Innovative Organization’’ 179
7 Conclusion
which is the object of several works in innovation theory. Instead we have explored
how novelty may be absorbed as a function of different degrees of redundancy.
We have suggested that the degree of redundancy influences the degree of
complexity and surprise that the system may absorb. A high degree of redundancy
allows for unexpected outcomes to be absorbed and avoids disappointment. Increased
redundancy enables a broader range of novelty to be absorbed by the organization. It
may work as a ‘‘buffer’’ for themes in the organization that work with lower degrees
of redundancy. Parsimony of communication, on the other hand, reflects a narrower
range of redundancy, hence greater precision of expectations. Both these scenarios
have their advantages and disadvantages. Greater redundancy allows for a broader
range of possibilities to become actuality to the organization.
We have further assumed that what distinguishes different functions within an
organization is not so much what tasks they perform, but by what differentiates their
modes of communication, and especially how different functions incorporate different
degrees of redundancy in their communication. Using the examples of strategy,
administration, and product development, we have argued that for the two former
themes, their relative degree of redundancy is given by their functions, whereas
product development exhibits an adjustable degree of redundancy which may be
decided by the organization. We consider the degree of redundancy in communication
to influence the way in which novelty is engaged.
At the level of the individual theme, broadness comes at the expense of narrowness;
hence, increased redundancy makes it more likely that only possibilities that lie in the
vicinity of the day-to-day operations become of actuality. At the same time, extensive
redundancy carries with it less risk, because expectations are less precise than what is
the case with less redundancy. Conversely, narrowness entails greater risk of
expectations not being met, and thus engendering greater disappointment when they
are not met. However, when the organization is seen as a whole, it does not make
sense to say that broadness comes at the expense of narrowness. On the contrary, it is
quite possible that broadness (high redundancy) in one theme allow for narrowness
(low redundancy) in another theme.
References
Åkerstrøm Andersen, N. (2003). Polyphonic organizations. In: T. Hernes & T. Bakken (Eds),
Autopoietic organization theory (pp. 151–182). Liber, Oslo: Abstakt.
Bateson, G. (1972). Steps to an ecology of mind. Northwhale, NJ: Jason Aronson Inc.
Beckert, J. (1997). Grenzen des Marktes. Frankfurt: Campus Fachbuch.
Bower, J. L., & Christensen, C. M. (1995). Disruptive technologies: Catching the wave. Harvard
Business Review, Jan.–Feb., 43–53.
Callon, M., & Muniesa, F. (2005). Economic markets as calculative collective devices.
Organization Studies, 26(8), 1229–1250.
Christensen, C. M. (1997). The innovator’s dilemma. Cambridge, MA: Harvard Business School
Press.
An Autopoietic Understanding of ‘‘Innovative Organization’’ 181
Christensen, C. M., & Raynor, M. E. (2003). The innovator’s solution. Cambridge, MA:
Harvard Business School Press.
Cohen, W. M., & Levinthal, D. A. (1990). Absorptive capacity: A new perspective on learning
and innovation. Administrative Science Quarterly, 35, 128–152.
Eisenhardt, K. M., & Martin, J. (2000). Dynamic capabilities: What are they? Strategic
Management Journal, 21(10–11), 1105–1121.
Emery, F. E. (1977). Futures we are in. Leiden: Martinus Nijhoff.
Emery, F., & Trist, E. (1965). The causal texture of organizational environments. Human
Relations, 18, 21–32.
Fagerberg, J. (2006). Innovation: A guide to the literature. In: J. Fagerberg, D. C. Mowery &
R. R. Nelson (Eds), Oxford handbook of innovation. New York: Oxford University Press.
Giddens, A. (1996). Modernitetens konsekvenser. Oslo: Hans Reitzel.
Hargadon, A., & Douglas, Y. (2001). When innovations meet institutions: Edison and the
design of the electric light. Administrative Science Quarterly, 46, 476–501.
Hernes, T. (2008). Understanding organization as process — theory for a tangled world. London:
Routledge.
Joas, H. (1992) Die Kreativität des Handelns. Frankfurt am main, Suhrkamp.
Latour, B. (1999). Pandora’s hope — essays on the reality of science studies. Cambridge, MA:
Harvard University Press.
Leifer, R., et al. (2000). Radical innovation: How mature companies can outsmart upstarts.
Cambridge, MA: Harvard Business School Press.
Luhmann, N. (1987). Paradigmawechsel in der Systemtheorie-ein Paradigma für Fortschritt?
In: R. Herzog & R. Koselleck (Eds), Epochenschwelle und Epochenbewusstsein. München/
Munich: Wilhelm Fink Verlag.
Luhmann, N. (1990). Complexity and meaning. In: Essays in self-reference (pp. 80–85). New
York, NY: Columbia University Press.
Luhmann, N. (1995). Social systems. Palo Alto, CA: Stanford University Press.
Luhmann, N. (2000). Organisation und Entscheidung. Wiesbaden/Opladen: Westdeutscher
Verlag.
Luhmann, N. (2005). Vorbemerkungen zu einer Theorie sozialer Systeme. In: Soziologische
Aufklärung 3. VS Verlag für Sozialwissenschaften, Wiesbaden.
March, J. G. (1991). Exploration and exploitation in organizational learning. Organization
Science, 2(1), 71–87.
Morgan, G. (1986). Images of organization. Great Britain: Sage.
Parsons, T. (1951). The social system. London: Routledge and Kegan Paul.
Peters, T. J., & Waterman, R. H. (1981). In search of excellence. London: Harper and Row.
Stringer, R. (2000). How to manage radical innovation. California Management Review, 42(4),
70–88.
Thompson, V. A. (1965). Bureaucracy and innovation. Administrative Science Quarterly, 10(1),
1–20.
Tidd, J., Bessant, J., & Pavitt, K. (2001). Managing innovation integrating technological market
and organizational change (2nd ed.). Chichester: Wiley.
Tsoukas, H., & Chia, R. (2002). On organizational becoming: Rethinking organizational
change. Organization Science, 13(5), 567–582.
Van de Ven, A. H., Polley, D., Garud, R., & Venkatamaran, S. (1999). The innovation journey.
New York NY: Oxford University Press.
Von Foerster, H. (1993). Wissen und Gewissen: Versuch einer Brücke. Frankfurt: Suhrkamp.
182 Tore Bakken et al.
APPLICATIONS OF AUTOPOIESIS IN
ORGANIZATION PRACTICE
Chapter 10
Information in Organizations:
Rethinking the Autopoietic Account
Ian Beeson
environment (which will include other systems) act as mutual sources of perturbation
for one another. The changes in a living being which result from its interaction with
its environment are triggered by a disturbing agent in the environment, not determined
by it. It is always the structure of the disturbed system (the living being) that
determines the changes. There can therefore be no ‘‘instructive interactions’’ by
means of which something outside the system determines its behavior.
It is the organizational closure of living systems that produces their autonomy and
their individuality. Each individual has its own autonomous ontogeny (its own
separate development and history), which is neither controlled by its environment nor
determined by its class or species. This same closure dictates that there cannot be any
inputs or outputs of a system including information) that, for the system, have
independent, objective reality outside it.
Over time, provided there are no destructive interactions between the system and
the environment in which it realizes itself, the system will appear to an observer to
adapt to its environment. What is in fact happening, though, is a process of
‘‘structural drift’’ occurring as the system responds to successive perturbations in the
environment according to its structure at each moment. Though it might seem as if
the system’s behavior is controlled by the environment, this is a feature of the
observation, not of the operational reality.
What in the generalization of their argument Maturana and Varela (1987, p. 181)
call ‘‘third-order’’ couplings and unities arise as the natural result of the congruence
between the respective ontogenic drifts of higher organisms of the same species.
A third-order unity — a ‘‘social system’’ of animals or people — itself displays
autopoietic organization to the extent that the network of participating individual
organisms coproduce it through their reciprocal structural coupling. The reciprocal
coordination mutually triggered among the members of a social unity, they call
communication (1987, p. 193). So defined, communication is an aspect of social
behavior, emerging out of structural coupling and drift in social groups, and not a
separate or distinct mechanism suddenly appearing in evolutionary development.
Communication is then the basic behavior from which language (or ‘‘languaging’’)
emerges. Human linguistic behavior is a domain of reciprocal ontogenic structural
coupling which human beings establish and maintain together. Although from an
external perspective, an observer may be able to describe words as designators of
objects or situations in the world, the operational reality of our use of language with
one another reflects a structural coupling in which words are ‘‘ontogenically
established coordinations of behavior’’ (Maturana & Varela, 1987, p. 208).
Just as Maturana and Varela put communication before language, so they suggest
that mind and consciousness, far from being prior to language, arise out of it. As
they phrase it, ‘‘language is a condition sine qua non for the experience of what we
call mind’’ (1987, p. 231). Reversing the Cartesian cogito ergo sum, they place
thinking, in the sense of conscious thought, at the end of a chain which starts with
autopoiesis, then proceeds through structural coupling, communication, and language.
By putting being before thinking, the theory of autopoiesis offers an approach to
communication that has existential and ontological, rather than epistemological
foundations.
Information in Organizations: Rethinking the Autopoietic Account 187
signal
information encoder/ receiver/ information
source transmitter decoder destination
message message
noise
channel was that a channel capable of transmitting C bits per second could transmit
symbols from a source with an entropy of H bits per symbol at an average rate no
higher than C/H symbols per second. He also showed that, for any channel whose
capacity was greater than the noise it was subjected to, an encoding scheme could be
devised which would guarantee accurate delivery of messages.
Shannon’s theories lie at the foundation of communication engineering. He and his
interpreter, Weaver, were at pains to point out that a specialized notion of
‘‘information’’ was being used here, related not to any intention to convey meaning,
but rather to variety in the message set and uncertainty in the source. Weaver
postulated three levels of communication:
Weaver acknowledged that Shannon’s theories applied only to Level A, but then
went on to claim that ‘‘the theory of Level A is, at least to a significant degree, also a
theory of Levels B and C’’ (Shannon & Weaver, 1949, p. 6). Ritchie (1991) believes
this generalization of Weaver’s to be unfounded, and to be based at least in part on a
confusion between two meanings of uncertainty: the statistical uncertainty relating to
the set of symbols which can be generated by the source and the subjective or
cognitive uncertainty of an observer about what particular message was sent in a
noisy channel (pp. 53–55). Weaver’s treatment of Levels B and C is in fact quite
vague, and verges on the mystical. He wants the theory to be generalizable: ‘‘y the
mathematical theory is exceedingly general in scope, fundamental in the problems it
treats, and of classic simplicity and power in the results it reaches. y [T]he theory is
sufficiently imaginatively motivated so that it is dealing with the real inner core of the
communication problem — with those basic relationships which hold in general, no
matter what special form the actual case may take’’ (Shannon & Weaver, 1949, p. 25).
And he is willing to restrict language behavior to make the theory fit:
This idea that a communication system ought to try to deal with all possible messages, and that
the intelligent way to try is to base design on the statistical character of the source, is surely not
without significance for communication in general. Language must be designed (or developed)
with a view to the totality of things that man may wish to say; but not being able to accomplish
everything, it too should do as well as possible as often as possible. That is to say, it too should
deal with its task statistically. (p. 27)
the basic technical model in all its essentials — simply adding levels of semantic and
pragmatic source/destination, encoding/decoding, transmission/receipt and noise,
and assuming that analogous mechanisms are at work at each level. But, in fact, the
generalization of the technical model to semantic and pragmatic levels must be
contested, since it appears beset with flaws. Linguistic communication between people
is not the same as, and not very like, signal transmission between machines. It is not a
stochastic process.
The Shannon model persists in organizational theory and in management thinking
even though it does not give an adequate representation of human communication.
Shannon devised the model for signal transmission, but it has appeared to his
interpreters to be capable of more general application. Because this possibility of
generalization also seems to accord rather strongly with ideas about the detachability
of information from context and the desirability of controlled communication,
Shannon’s model has not been definitively curtailed. This is no doubt partly a
political matter, but there are also theoretical and technical issues surrounding
attempts to store information in databases and transmit it across networks.
Akmajian, Demers, and Harnish (1998, p. 65) list four problems with what they call
the Message-Model (the Shannon model generalized to talk exchanges):
1. it identifies the message to be sent with the literal meaning of the words uttered;
2. it depicts the process of encoding the meaning into sounds as linear and sequential;
3. it depicts the decoding process from sounds into meaning as linear and sequential;
4. it depicts the hearer’s recovery of the intended message as being identical with the
decoding of the meaning of the sentence.
may have different deep structures. Akmajian et al. give the example (p. 70): ‘‘I saw the
man with the telescope’’; that two different meanings are available (to both speaker
and hearer) shows that the meaning of a sentence cannot in general be derived
sequentially from the meaning of the individual words.
Proceeding to the pragmatic level (roughly equivalent to Weaver’s Level C —
‘‘effectiveness’’), Akmajian et al. list six inadequacies of the Message-Model (pp. 75–78):
as stochastic processes. They are not engaged in selecting symbols from a finite set of
symbols on a basis of probability. Successful communication depends on a good deal
more than the transmission and receipt of symbols or sounds — for instance,
common language, shared context, resolution of references, resolution of ambiguities,
identification and interpretation of non-literal meanings, communication of intention,
and identification of indirect meanings.
Sperber and Wilson (1986, 1998) have further developed the pragmatic analysis of
communication by building on Grice’s theory of implicature (Grice, 1975). Grice
draws a critical distinction between what is said by a speaker and what is implied.
What is said is closely related to the conventional meaning of the words uttered, and
can be worked out from an understanding of the language. What is implied is
sometimes also derivable from the words uttered and their conventional meanings,
but more generally, in a conversation, implications derive not from the words, but
rather from the working out of understandings between the participants in the course
of their exchanges. He pictures conversation as a cooperative venture.
Sperber and Wilson give an account of inferential communication derived
from Grice’s, but founded on the central notion that to communicate is to claim
someone’s intention, and hence to imply that the information communicated is
relevant (Sperber & Wilson, 1998, p. 82). Communication is achieved when a
speaker produces ostensive stimuli which provide evidence to the audience from
which the speaker’s informative intention can be inferred (p. 91). They proceed
toward a definition of relevance in which — other things being equal — an
assumption with greater contextual effects is more relevant than one with smaller
effects, and an assumption requiring a smaller processing effort is more relevant than
one requiring a larger effort (p. 96). The principle of relevance is simply this: that a
hearer treats it as axiomatic that the speaker has done his best to be maximally
relevant (p. 361).
In interpreting an utterance the hearer uses this principle as a guide, on the one hand towards
correct disambiguation and assignment of reference, and on the other in deciding whether
additional premises are needed [i.e., for indirect implications], and if so, what they are, or
whether a figurative interpretation was intended. (Wilson & Sperber, 1998, p. 361)
Sperber and Wilson do not propose that their version of an inferential model of
communication should be elevated into a general theory of communication (p. 91).
They even leave a limited role for the message model (which they call the ‘‘code
model’’): they suggest that a hearer recovers the linguistic meaning of an utterance by
a form of decoding, but then applies inferential processes to recognize the speaker’s
intentions (p. 82).
This pragmatic analysis of communication goes a long way toward defusing the
arguments leveled against information in the theory of autopoiesis. In this analysis,
information is no longer seen (as in the message model) as existing objectively and
externally to those using it. The emphasis instead, as Maturana and Varela would
wish, is on how a hearer responds to a communication, and how members of a
community use language to coordinate their behavior.
Information in Organizations: Rethinking the Autopoietic Account 195
Dell (1985), in his illuminating comparison of the work of Maturana and Bateson,
remarks that ontology is ‘‘the road not taken’’ in Bateson’s thinking and suggests that
the biological ontology delineated in Maturana’s work could provide a sound
foundation for the social and behavioral sciences.
As commented earlier, Luhmann’s significant reworking of autopoiesis followed an
epistemological rather than an ontological route. Ontology is the branch of
philosophy concerned with what things exist in the world. In Maturana and Varela’s
theory, the most fundamental behavior (comments Dell) is to exist, and the most
fundamental knowledge is to know how to exist. In autopoietic theory, the world is
populated by (structure-determined) living organisms. Dell explains the significance
of Maturana’s ontology thus:
Structure-determined living systems automatically become organized into interactional systems.
Whenever two or more structurally plastic living systems interact they will begin to co-evolve a
closed pattern of interaction. They will form a system. When a system is understood in terms of
structural coupling, it can be seen that there is no need to explain the system’s organization in
terms of homeostasis, systemic rules, or control hierarchies. y The system arises naturally from
the way its structurally plastic components fit together. Such a system results from, and is, the
structural coupling of its components. (Dell, 1985, pp. 13–14)
6 Information in Organizations
The theory of autopoiesis can be combined with the inferential model of
communication and the theory of information pickup to give an approach to
information in organizations that emphasizes action, interaction, and attention rather
than storage and transmission. In this approach, human beings are regarded as active
and attentive in their pursuit of information. Information is obtained by them
continuously as they move through their environment or as they interpret the
implications and resonances of what is communicated to them. They are constantly
interacting with one another, always on the lookout for information, and always
oriented toward working out its relevance.
The lacunae and ambiguities always present in the exchange of messages in
organizational settings create the conditions in which information problems are
bound to occur. The expansion of information systems across organizational
boundaries, because it attenuates the connections between communicating partners
Information in Organizations: Rethinking the Autopoietic Account 197
still further, will inevitably make matters worse. The solution cannot be found in
formal information or computer systems, however far extended, and however much
enhanced by the introduction of intelligent agents and similar software assistants to
track and prompt our data usage. Such ‘‘active’’ information systems will help people
marshal their data, but the derivation of meaning will perforce remain a human
prerogative.
In the familiar designs and expectations of current information systems, there is no
assumption that (nor any attempt to make) information always available, or always
relevant. Information is rather — under a rationale of efficiency or control —
confined and limited within reporting cycles, hierarchical regimes of disclosure, and
predefined accesses or queries. The question of relevance does not come up for two
reasons: the meaning and implication of any item is presumed to be fully contained
within the item itself (which in turn assumes a common context for originator and
recipient of the information) and whatever interpretation is made should in any case
be the standard one laid down in operating procedures. The ability of individuals to
find new information and make new interpretations is thus obstructed, often on
purpose. The expressive range of language is severely curtailed by confining
information within prearranged formats and conventional meanings; the entire
figurative dimension, for instance, is sacrificed in the reduction of information to
records and recipes. Information gets lost as interaction and coupling are blocked in
the mire of procedure and hierarchy.
Were managers and designers to recognize that individuals naturally and inevitably
interpret information and seek relevance in it (however obscure it is made), they could
then design systems which exploited this capability and encouraged interpretation.
For example, structured information sources could be expanded (either within the
system or in supplementary documentation) to include explanations or commentaries
on data items and algorithms or decision routines. This would recognize and exploit
the fact that individuals are predisposed to try to make sense of the information that
confronts them. If, further, the author or agent currently responsible for each data
item and routine were to be made known to users, the possibility would be opened of
a conversation between originator and user which could deepen relevance, refocus
attention, and perhaps suggest different or additional uses of information. Taking up
Sperber and Wilson’s suggestion that relevance can be plotted on a continuum, we
can imagine an information analysis of an organization which concentrated less on
the formal structure of items, and more on the contextual effects and interpretive
effort associated with each item. Sperber and Wilson have commented that a full
account of how people understand one another’s utterances — especially, but not
only, figurative ones — requires a theory of rhetoric, which is to say a theory of
argument and persuasion. If we are to make sense of and more effectively use the
information abundantly available in our organizations, we need to be able to judge
what is relevant, to make convincing interpretations, and to persuade others of its
import. This is so because meaning is not to be found in the symbols, and not in the
databases nor the diagrams, but is constructed by us as we engage with the world and
in the spaces of conversation which stretch between us.
198 Ian Beeson
7 Conclusion
References
Akmajian, A., Demers, R. A., & Harnish, R. M. (1998). Overcoming inadequacies in the
message-model of linguistic communication. In: Kasher, A. (Ed.), Pragmatics: Critical
concepts (Ch. 78, Vol. V, pp. 63–81).
Beeson, I. A. (2001). Implications of the theory of autopoiesis for the discipline and practice of
information systems. In: N. L. Russo, B. Fitzgerald & J. I. DeGross (Eds), Realigning
research and practice in information systems development (pp. 317–332). Norwell, MA/
Dordrecht: Kluwer/IFIP.
Dell, P. F. (1985). Understanding Bateson and Maturana: Toward a biological foundation for
the social sciences. Journal of Marital and Family Therapy, 11(1), 1–20.
Gibson, J. J. (1986). The ecological approach to visual perception. Hillsdale, NJ: Lawrence
Erlbaum Associates.
Grice, H. P. (1975). Logic and conversation. In: Kasher, A. (Ed.), Pragmatics: Critical concepts
(Ch. 54, Vol. IV, pp. 145–161).
Information in Organizations: Rethinking the Autopoietic Account 199
Kickert, W. J. M. (1993). Autopoiesis and the science of (public) administration: Essence, sense
and nonsense. Organization Studies, 14(2), 261–278.
Luhmann, N. (1995). Social systems. Palo Alto, CA: Stanford University Press.
Maturana, H. R., & Varela, F. J. (1980). Autopoiesis and cognition: The realization of the living.
Dordrecht: D. Reidel.
Maturana, H. R., & Varela, F. J. (1987). The tree of knowledge. Boston, MA: Shambhala.
Merleau-Ponty, M. (1992). Phenomenology of perception. London: Routledge.
Morgan, G. (1997). Images of organization. Thousand Oaks, CA: Sage Publications.
Ritchie, L. D. (1991). Communications concepts 2: Information. Newbury Park, CA: Sage.
Shannon, C. E., & Weaver, W. (1949). The mathematical theory of communication. Urbana-
Champaign, IL: University of Illinois Press.
Sperber, D., & Wilson, D. (1986). Relevance: Communication and cognition. Oxford: Blackwell.
Sperber, D., & Wilson, D. (1998). Précis of relevance: Communication and cognition. In:
Kasher, A. (Ed.), Pragmatics: Critical concepts (Ch. 79, Vol. V, pp. 82–115).
Wilson, D., & Sperber, D. (1998). On Grice’s theory of conversation. In: Kasher, A. (Ed.),
Pragmatics: Critical concepts (Ch. 63, Vol. IV, pp. 369–382).
Chapter 11
1 Introduction
The relation between organizational change and information systems has received
much attention in the information systems literature. Much of this research has
concentrated on the effects of organizational change on information systems and vice
versa. Such research has generated rich insights in the facilitating as well as
constraining role of information systems in the process of organizational change.
Many analyses assume a distinction between the system and its environment, the
organization. Information systems are seen as being relatively stable entities while the
environment is a source of change and uncertainty. Such a perspective stresses the need
for a continuous adaptation of the information system to its dynamic environment.
In this paper we challenge this conception of the relation between information
systems and organizational change by outlining a perspective that focuses on the
process of evolution itself that leads to the effects commonly researched in the
information systems literature. This reflects a wider interest within the study of
information systems and organizations that calls for a richer understanding of the
generative mechanisms through which information systems and organizations evolve.
Evolution, in this view, is not conceived as a trajectory of improvement, which leads
to a desired end-result through several phases but as a process of which we need to
explore its inner workings. We then might develop a richer insight as to how and why
information systems and organizations change and stabilize, irrespective of the effects
they produce.
In this paper we draw on the theory of autopoiesis, a recent biological theory
which sheds new light on the evolution of living systems and which might be relevant
for the evolution of information systems as well. In this theory the common relation
between a system and its environment is blurred. The environment only exists through
perception, and thus is part of the system. Change, in this view, is generated internally
and evolution is not a process of adaptation but of maintaining the system’s self-
identity. We attempt to apply the ideas of autopoiesis metaphorically to the evolution
of information systems, which we illustrate with an empirical study, and we present
some theoretical implications of this perspective.
The next section presents a short overview of autopoiesis with respect to the issues
that are relevant for our purposes. In the third section we will outline how we use
autopoiesis for the analysis of the evolution of information systems. The ideas that
are metaphorically derived from this theory are illustrated in the fourth section by
drawing on a case study of the evolution of an information system over 13 years.
Subsequently we analyze the case by drawing on autopoiesis and discuss our findings
in the light of contemporary social theory. In the sixth section we end the paper with
the conclusion that autopoiesis opens new and interesting lines of thinking about
change in and of information systems.
Maturana and Varela have formulated the theory of autopoiesis in the early 1970s as
an explanation for the nature of living systems (Maturana & Varela, 1980). The term
autopoiesis is adopted from Greek and means self-production. The theory is a new
approach to systems thinking. The central idea of autopoiesis is that living systems
produce themselves. The system’s components and processes jointly produce the same
components and processes, thus establishing an autonomous, self-producing entity
(Mingers, 1995). Autonomy of a system is the key feature of living beings and refers
to the ability to specify what is proper to it (Maturana & Varela, 1992). The
mechanism that makes living systems autonomous is autopoiesis.
The recognition of the autonomy of a living system implies that the traditional
distinction between a system and its environment is no longer valid because an
external observer makes such distinctions. Instead, autopoiesis poses that a living
system continuously constitutes its own boundaries, perceives its surroundings (which
Maturana and Varela call the medium) in its own ways, thereby constructing an
environment. In Varela’s words: ‘‘[W]e are becoming more and more interested in an
epistemology which is not concerned with the world-as-picture, but with the laying
down of a world’’ (Varela, 1984). In explaining the workings of the human brain, for
example, Maturana and Varela say that the brain produces images of reality which
are determined by how the brains themselves are structured. In other words, the
patterning of the brain determines the perception of the world. With those images,
interaction occurs that may lead to changes in the organization of the brain,
depending on the actual experience. In this sense, the environment is not ‘‘something
out there’’ but it is actively constructed by the system itself as part of its own
organization. Hence, the environment needs to be seen as part of the system.
Although a living system operates in a physical environment, the relation to that
environment and the interaction with it is determined internally. Thus, for example,
Autopoiesis and the Evolution of Information Systems 203
certain berries are poisonous for human beings. This is, however, not the intrinsic
property of the berries but dependent on the physical properties, i.e., the
organization, of the human being. For other living systems, certain birds for
instance, the berries may not be poisonous at all.
In maintaining autopoiesis the identity of a living system is of central importance,
and all activity is meant to preserve this identity. All interaction which the system is
engaged in is meant to reinforce or reproduce this identity. Patterns of interaction are
circular and part of the system’s organization. An important characteristic of
autopoietic systems is that they are organizationally closed systems, meaning that all
possible states of activity must always lead to or generate further activity within the
system (Mingers, 1995). Or, to put it differently, all activity must maintain autopoiesis
to prevent the system from disintegration. The environment, which is created by the
system itself, is therefore a projection of its own identity. The way the world is seen by
the system is determined by the system itself, instead of being a reflection of an
externally existing order. Living systems thus close in on themselves to maintain a
stable pattern of relationships that are self-referential. The interaction of the system
with its environment is always self-referential in the sense that it refers back to the
system’s identity in order to facilitate self-production, i.e., to maintain autopoiesis.
If living systems strive to maintain autopoiesis and relations with the environment
are determined internally, then systems can evolve and change only along with self-
generated changes in identity (Morgan, 1986). The theory of autopoiesis perceives the
evolution of living systems as a result of internally generated change. Rather than
suggesting that the system adapts to an environment or that the environment selects
the system that survives, autopoiesis places its emphasis on the way living systems
shape their own future.
Changes in the system are only triggered from outside. What the eventual change
will be and what in the environment can or cannot act as a trigger are determined by
the actual living system. The changes that an autopoietic system can undergo are
determined by the individual system so long as autopoiesis is maintained (Mingers,
1995). Living systems ‘‘are organized in such a way that their processes produce the
very components that are necessary for the continuance of the processes’’ (Mingers,
1989). Maintaining autopoiesis is not just the reproduction of the same characteristics
in similar circumstances, but rather the production of subsequent elements different
from previous ones. However, the state of the actual system at a given time will
determine the actual changes that the structure undergoes. In autopoiesis this is
known as structurally determined. The internal structure determines what changes are
possible to occur — only those that maintain autopoiesis — and thus how interaction
with the environment will trigger changes in the system.
have refuted the idea that social systems are autopoietic. Some have suggested
that social systems often portray characteristics similar to those autopoiesis explains
in living systems, such as autonomy and the persistence of identity in contexts of
massive change (Mingers, 1995). Opinions on the applicability of autopoiesis to
areas other than living systems, such as groups, organizations, or society, seem to
differ (Mingers, 1995; Brocklesby & Mingers, 2005; Huysman, Frisart, & Heng, 1995;
Seidl & Becker, 2006). Some have directly applied the theory to social systems or
have tried to slightly alter the theory of autopoiesis to fit the social system as
well. Influential has been the work of Niklas Luhmann on social systems as
being autopoietic and in which communication plays a central role in the process of
self-production (Luhmann, 1986; Seidl & Becker, 2006). The book by Von Krogh,
Roos, and Slocum (1995) brought the theory in relation to knowledge management
issues.
We share the concerns about the direct applicability of autopoiesis to social
systems and support Kickert’s remark that ‘‘y it is not so important whether a useful
idea is an accurate translation of the original natural scientific model, but rather
whether the idea is interesting and relevant y’’ (Kickert, 1993). In order to explore
the relevance of the theory of autopoiesis to the area of information systems we use
the theory as a metaphor. It has been successfully exercised and embraced in the fields
of organization theory and information systems (Morgan, 1986; Walsham, 1991).
The use of autopoiesis as a metaphor for information systems questions common
conceptualizations of information systems. Traditional views on information
systems are rooted in a mechanistic paradigm based on cybernetic systems thinking,
while the increased attention to social issues call for ways of thinking beyond this
‘‘dead paradigm for living systems’’ (Ray, 1993; Blonk, 2002). In this paper we
follow Kling’s conceptualization of an information system as a web of computing,
a perspective that explicitly includes the social, historical, and political dimensions
of the system besides the focal technology. Information systems ‘‘are not only
flexible information processing tools [but] their ‘‘shape,’’ the way they are used, the
leverage they provide, and the interests they serve depend upon the interplay of
stakeholders, resources, and social games within which they are deployed’’ (Kling,
1987). So, information systems are not just neutral entities, but they embody
procedures, routines, power structures, and so on. They pre-select actions, relations,
and possibilities. They embody ‘‘how things are done around here’’; they have an
identity.
In the next section we describe a case study of an information system that clearly
acquired such an identity. It embodies how things were done one way rather than
another. And its evolution shows a tendency to maintain itself, to keep up its identity
regardless of its dynamic surroundings. The case shows that it was not just the
hardware and the software, but also the social groups involved, the structures that
were created, the style of thinking, and the way of managing that preserved the system
as a whole. This whole socio-technical ensemble is what we attempt to analyze as if it
were autopoietic. A system that continuously constituted its own boundaries, seemed
to have acquired a high degree of autonomy, and was actively involved in reproducing
and thus maintaining itself.
Autopoiesis and the Evolution of Information Systems 205
The case tells us about a financial management information system at the Dutch
Railways, which was developed in 1981 and continued to exist until 1993. The life of
this system is described parallel to a massive process of change at the Dutch Railways
which, in this period, was transforming from a state-owned and open-end financed
corporation to a privatized and commercial business. Within the context of these
massive changes, the financial information system continued to exist despite the
‘‘match’’ with its organizational context was lost. Seen in retrospect, it raises the
question how a system that increasingly did not fit its environment anymore continued
to be supported and financed, and was even redeveloped. For this, we need to
understand the historical context of the organization and how it responded to change.
The Dutch Railways is an old organization, which traditionally has been a state-
owned company. Before approximately the 1980s the organization had been quite
stable even though in the 1970s the Dutch Railways saw an enormous expansion of
the organization, its infrastructure, and its activities. The organization had developed
quite a strong culture, which provided the members of the organization security and
stability. Employment was lifelong, and salaries and fringe benefits were good. The
organization itself was hierarchical, and administration and staff were centralized at
the head office. The regional units of the organization were mainly concerned with the
operational processes to keep the trains running, such as personnel scheduling and
maintenance work. The strong position of the unions and importance to acquire
status and resources (which was dependent on number of employees) made personnel
issues a very central focus. Financially, the organization was open-end financed,
which meant that all costs were accounted for by the Dutch government. Especially in
the period of expansion of the 1970s, expenses had grown enormously which in fact
started the process of restricting expenses, a process which would unfold into different
successive ways of financial management and eventually in the privatization of the
Dutch Railways in the 1990s.
Around 1980 the Dutch government decided to cut expenditure on the Dutch
Railways by introducing yearly budget limits. As a consequence, the organization was
restructured into a flatter organization with more regional units and decentralization
of responsibilities. The position of the regional accounting departments gained much
in importance; parts of the financial administration was now carried out regionally
and the head of the department became member of the regional management team. In
this context, the head of the central department of Planning and Control initiated a
project to develop a new method of registrating costs and providing management
information. This new accounting method (NAM) was meant to support the new
budget-based organization, but the motivation to develop this method and its
supportive system was more comprehensive; it also was an attempt to introduce a
certain style of management based on personal as well as professional grounds, as the
former head of the central department of Planning and Control states:
It was purely my initiative. It was my opinion we would have to do this and that we should do
this. There was no decent instrument for financial management available.
206 Marleen Huysman et al.
To realize the initiative, the head of the central department of Planning and
Control hired an accountant and consultants who designed the basic structure of the
system. A project group was started to further develop the system and to implement it
in every regional unit. Young professional assistants were hired and trained to
implement and use NAM. Every assistant was assigned the task to take care of the
implementation of the system in the regional unit in which they were given a
permanent position within the accounting department. The system was a revolution
because for the first time costs of operational processes were identified, registered, and
managed. The project group worked under the supervision of the central department
who could decide how the system was modified and further developed. In 1984 the
system can be said to be implemented in every regional unit, but several regions
showed quite some resistance. An important reason for this was that regional
management disliked the fact that NAM made decisions financially transparent, as a
former accountant notes:
The head [of the regional accounting department] wasn’t a very popular guy; somebody who was
always harping on the money. By then, it was not very common to talk about money.
Further, the system, which still was manual, appeared to involve much work of a
labor-intensive and simple calculative nature. A first attempt to automate the system
failed because of technological reasons. A spreadsheet appeared not to work on one
of the first IBM PCs when the data of the NAM system was entered. A new head of
the central department of Planning and Control, somebody who also was involved in
the development of NAM, asked the central information systems department to
design and build a computer-based system to support NAM. Although that system,
called NAMIS, took over much of the routine work, the underlying method of NAM
was not changed.
In 1988, not too long after the automated system was introduced, the budget
structure was changed into what was called ‘‘contract management.’’ Instead of the
Dutch government setting budget limits (which could be exceeded), the Dutch
Railways were now required to plan their expenditures in advance thereby estimating
the budget needed. The proposed budget was then recorded in a sort of contract
between the Dutch government and different levels of management. The context in
which the system was designed and operated had now changed quite significantly in
nature. NAMIS was designed to registrate and allocate costs and produce
management information based on the recorded data. It was not designed to support
the planning of expenditures a year ahead in order to determine the budgets needed.
NAMIS also suffered from some functional shortcomings and rigid features, which
had resulted in resistance and dissatisfaction with the system. Users had developed
extensive procedures and routines to cope with the system’s rigidity and restrictions in
order to do their work properly. This, then, was the background for the initiative to
rebuild NAMIS. The project team, which did an initial study in 1990, concluded that
NAM and its underlying logic should be maintained. The problems were identified in
obsolete computer equipment, functional shortcomings and mistakes, and inadapt-
ability to local requirements. Solving these problems by rebuilding the system was
seen as the way out.
Autopoiesis and the Evolution of Information Systems 207
Also in 1990 the organization, again, was restructured. The 15 regional units were
grouped together into 8 larger units and were given significantly more autonomy.
Each region thus formed its own management team, and several new managers and
controllers were hired, some of which came from outside the organization. Also, a
controller from corporate level replaced the head of the central department of
Planning and Control. Several newly formed regions started to develop their own
information systems according to their own views and needs, as a former controller
remarks:
New managers in the regions with new controllers, often not from the original organization.
Everywhere the wheel was re-invented, and everywhere different.
Shortly after the new organization was implemented, the new head of the central
department of Planning and Control had to decide whether or not to continue the
project of rebuilding NAMIS, history of which he was not familiar with. Several
regional controllers under his supervision (who had been involved in the development
and implementation of NAM) were in favor of the continuance. The decision was
taken to continue the project as the head of the central department of Planning and
Control comments:
Was the system bankrupt in peoples minds? Of some, yes. But surely not of everybody. [We
continued the project because] otherwise you have nothing, then you don’t know what happens
out there, what kind of costs are being made.
Parallel to the start of the new organization, the project to rebuild NAMIS
continued from early 1991 onwards. The project team consisted of information
systems professionals and a small group of users who were selected by the project
leader. Most of the participating users were already familiar with NAM for a number
of years. During its development the system was presented very attractively to the
organization — proposed future users could engage in a prize contest to give the
system a name, a logo for the system was designed, frequent newsletters were
distributed, and an expensive-looking manual for the NAM method and the new
information system were distributed among the users. In the meantime the ‘‘old’’
NAMIS was still supposed to be used in the regions until the new system was ready.
But the actual situation was very different. Several systems were being developed in
different regions, and even the central department of Planning and Control started
projects for a new financial management information system. These systems were
based on the new organization and the information requirements it imposed. However,
they also were alternatives for the NAM and its supportive information systems. One
region produced management information reports for the central departments using
their own systems to generate the information and a word processor to imitate the
layout of NAMIS reports.
The project to rebuild NAMIS suffered from a number of drawbacks: The formal
description of NAM had been lost and needed to be rewritten, there were performance
problems, and a conflict with the supplier of the system delayed the project. The system
was introduced in 1993, approximately two years after the reorganization and nearly
four years after the project was started. During the implementation and user training
208 Marleen Huysman et al.
the team noticed serious resistance. Two regional units refused to implement the
system and a third wanted to postpone the implementation for one year. Subsequently,
the general manager forced the regional units to implement the system. Even though
the system was technically implemented it was never successfully used. One and a half
year later the department formally responsible for maintaining the system didn’t have
a clue who were still using the system, and they were very surprised to hear that some
were still working with it.
Yet, years after the collapse of NAM and its computer-based versions, several
people still view the system as a good system that should have been used till today.
The problems are not in the system, but in its environment — the organization, they
say:
[If you watch the developments now] there’s nothing new. What we introduced in ‘84, the
organization might be ready to work with it. But it required two expensive information systems
to get there. What is now presented as new or innovative, is just a revival if you’ve been in it long
enough. (A former project member)
[NAM] could have been extended and changed into a system that still would have been used.
Then it would have been the current system. But we went through a different line of
developments; they blew it, the reorganization, no support from management, everybody wanted
their own system — and now again there is a trend toward a uniform system. The same result,
but just a different path. (Former controller 1)
Even though the computerized information system NAMIS was not used
throughout the organization, the underlying NAM method was still followed and
used throughout various regions, although the method was used in different ways and
to different levels. It was no longer a uniform method, and as such also reflected the
new decentralized organizational structure. The embedded centralized structure and
enforced uniformity of NAMIS had become out of sync with the new decentralized
organizational structure of the Dutch Railways. Still the heart of the system, the
NAM method, was still standing strong, and renewed itself during the years to come.
Over the next few years, several initiatives had been taken to develop computer-based
information systems to support different regional accounting methods, often based on
NAM to a certain extent. One of these information systems has even been used in
different regions. The controller who was involved in the development of this regional
information system was one of the early assistants hired to develop the NAM method,
now involved in the development of the information system that has made NAMIS
obsolete. But he himself sees a strong continuation:
The whole idea behind NAM is still alive and present in today’s systems, but it is experienced as
something totally new. It’s so crazy, so funny. People who have never known NAM end up with
the same sort of ideas. And that’s quite nice to notice. (Former controller 2)
reflect the system’s view and understanding of the organization, as well as its interests,
which apparently closely linked to those of the accountants and controllers. As time
went on, the system further actively organized its environment by introducing
computer-based information systems that embodied the method and took over much
of the administrative work. This is to be seen as a process of institutionalizing this
particular way of organizing so that the system could maintain itself.
The next issue which is raised by embracing the metaphor is what it is that is
evolving, what is the identity that is reproduced, and how? The identity of the system
determines how the system is organized, and the stability of the organization enables
the system to reproduce itself. Autopoiesis thus refers to a duality between the
structure of a system and what it aims to be. This is a recursive process in the sense
that changes in the system’s organization have to cohere with the system’s identity.
They are not separable, but evolve alongside. The process of organizing is the
realization of an identity; both a system’s identity and its organization including its
perception of the environment is the subject of evolution. Therefore, the system
creates its own conditions for evolution; it shapes its own future.
In the case, the identity of the system is the world of thought on which the system is
based, and which is embodied in the system. This self-image of the system was
continuously maintained and reinforced through self-referential processes. It contains
a set of values, and patterns of thinking, a preconceived structure of this particular
part of the organization of the Dutch Railways, and it is most clearly reflected in the
three statements at the end of the case description in the previous section that talk of
the ideas, the world of thought behind the system. Even though the computer-based
information system had failed and NAM was no longer used as a system, people
interpreted the world of thought behind the information system as still present and
relevant, although it was realized in different systems. The identity was further
emphasized and constructed through the prize contest to give the system a name and
the logo that gave the system a face. The system was recognizable. A system needs to
have such a cultural identity in order to deal with insecurity, uncertainty, and anxiety
that are inherent in social life (Gagliardi, 1986). A system may give people (who are
‘‘in’’ the system) a sense of identity — a framework of theory, values, and related
technology that enables people to make sense of their roles in the system (Schon,
1971; Gagliardi, 1986). This, at the same time, forms a condition for the maintenance
of the system’s very identity. The system in our case study clearly involved people who
continuously supported it, and were being involved in the maintenance and
redevelopment of the system. During data collection some people spoke of ‘‘die-
hards’’ who kept on supporting the system, and one of the project members, very
convinced, said: ‘‘It is my system.’’
The cultural identity and the system’s organization shape the construction of
opportunities and threats; it determines how a system perceives what is to be seen as
threats and opportunities. Problems, crucial developments, priorities, etc., in this
perspective, are closely linked to the identity a system wishes to maintain (Morgan,
1986). In the case of NAM, the strong tendency to maintain the cultural identity of the
system prevented it from incorporating the wider developments in the organization,
and shaped the conditions for failure. An interesting perspective on the strive for
Autopoiesis and the Evolution of Information Systems 211
stability in social systems and its resistance to change has been described by Schon in
his book Beyond the stable state, as what he calls ‘‘‘dynamic conservatism’ — that is to
say, a tendency to fight to remain the same.’’ Schon acknowledges that resistance is
inherent to social systems: ‘‘Resistance to change does not come from the stupidity or
venality of individuals within the system; it is a function of the system itself’’ (Schon,
1971). He thus points to a similar characteristic of social systems as autopoiesis points
to in living systems, which is the tendency to fight to remain the same.
The question is how social systems deal with threats and disruptive changes that
may cause destruction of the system. Schon discusses five strategies how social
systems exercise dynamic conservatism, i.e., how they deal with changes that disturb
the stable state, or to put it differently, how they engage in the process of maintaining
autopoiesis. First, the system tries to ignore the presence of a threat, and if it cannot
be ignored it launches a counter attack or a preventive attack before the threat has
materialized. If it does not succeed, it allows the threatened change a limited scope of
activity and keeps it isolated. The fourth strategy is to absorb agents of change and
turn to their own ends the energies originally directed toward change. And finally, if
even that appears not to work, the system changes, but it allows the least change
capable of neutralizing the intrusive process. The sequence of the first three strategies
we clearly recognize during the process of rebuilding NAMIS. The changes in the
organizational environment were ignored at first by keeping on using the same
system. When it appeared that the changes could no longer be ignored, developments
were attempted to be countered by redeveloping the NAMIS system. However, the
developments were interpreted as a reason for optimizing the existing system without
questioning the underlying method and assumptions, thus in such a way that the
system’s identity could be maintained. But the changes embodied in the new
organizational structure and new systems were of a different nature and could not be
ignored. The system appeared not able to adopt the fourth and fifth strategies, at least
not in time, and it was abolished.
6 Conclusion
In using the theory of autopoiesis as a metaphor to analyze the evolution of
information systems we were able to draw attention to the generative mechanisms, the
underlying dynamics that determine the continuity of a system. We have seen that a
system is able to create its own conditions to be successful, or alternatively for its
destruction. Autopoiesis, when used as a metaphor, may provide us with an
interesting perspective on the process of evolution of information systems. The
interesting insight this approach offers, and which was illustrated in the case, is that
the main aim of systems is to maintain their identity despite the changes in their
surroundings, in contrast to the common view of evolution as adaptation to external
changes. It draws attention to how the system is structured to view the world rather
than the way the world ‘‘is’’ (Walsham, 1991). This is not to say that such systems are
static but rather that their evolution is determined by the identity the system has
212 Marleen Huysman et al.
References
Blonk, H. (2002). Changing the order, ordering the change. The evolution of an information
system at Dutch Railways. Ph.D. thesis, Vrije Universiteit, Tinbergen Institute Research
Series no. 264, Thela thesis, Amsterdam.
Brocklesby, J., & Mingers, J. (2005). The use of the concept of autopoiesis in the theory of
viable systems. Systems Research and Behavioral Science, 22, 3–10.
Gagliardi, P. (1986). The creation and change of organizational cultures: A conceptual
framework. Organization Studies, 7(2), 117–134.
Huysman, M. H., Frisart, R., & Heng, M. S. (1995). Autopoiesis and organizational learning.
Implications for information and information systems. Proceedings of the Third SISnet
Conference, Berne, 18–19 September.
Kickert, W. J. M. (1993). Autopoiesis and the science of (public) administration: Essence, sense
and nonsense. Organization Studies, 14(2), 261–278.
Kling, R. (1987). Defining the boundaries of computing across complex organizations. In: R. J.
Boland, Jr. & R. A. Hirschheim (Eds), Critical issues in information systems research.
Chichester: Wiley.
Luhmann, N. (1986). The autopoiesis of social systems. In: F. Geyer & J. van der Zouwen
(Eds), Sociocybernetic paradoxes (pp. 172–192). London: Sage.
Maturana, H. R., & Varela, F. J. (1980). Autopoiesis and cognition: The realization of the living.
Dordrecht: Reidel.
Maturana, H. R., & Varela, F. J. (1992). The tree of knowledge. The biological roots of human
understanding. Boston, MA: Shambhala Publications Inc..
Mingers, J. (1989). An introduction to autopoiesis: Implications and applications. Systems
Practice, 2, 159–180.
Autopoiesis and the Evolution of Information Systems 213
1 Introduction
Social systems theory (Luhmann, 1984, 1995) closely embraces the concept of
autopoiesis which, originally, describes the recursive (self)-production of living systems
(Maturana & Varela, 1980). Following this, autopoietic organization theory (Bakken
& Hernes, 2003; Seidl & Becker, 2006) establishes a more specialized understanding of
autopoiesis in terms of organization studies. The transition from the biological to the
social realm, however, draws frequent critique. Some scholar suspiciously regard social
systems theory as antihumanistic (Blühdorn, 2000; Viskovatoff, 1999), for it neglects
individuals in favor of interactions, organizations, and societies. Others deconstruct
autopoietic organization theory with the argument that its definition of communica-
tion is ‘‘flawed with an unavoidable mental dimension, namely the component of
understanding’’ (Thyssen, 2003, p. 213).
The validity of the critique rests mainly with the level of abstraction that the
concept of autopoiesis maintains in both theories. In contrast to other schools of
thought (e.g., natural, rational, and open systems theory; cf. Hatch, 1997; Scott,
1998), there is little empirical leeway to back up ideas such as that individuals and
organizations are mutually exclusive autopoietic systems or, in more theoretical
terms, that individuals and organizations are separated by operation yet coupled by
observation. After all, there is no consciousness without communication as well as no
communication without consciousness (Luhmann, 2002).
In the following, I bridge the gap between theoretical contemplation and empirical
research with a model and simulation of autopoietic organizational knowledge,
learning, and memory. ‘‘Like voltage, current, and resistance, the terms knowledge,
learning, and memory must be defined in terms of each other’’ (Spender, 1996, p. 75).
Together, then, these terms characterize the fundamental dialectic between structures
and dynamics of organizations. The model draws heavily on social systems theory,
thereby complementing autopoietic organization theory. The simulation stands in the
tradition of mathematical approaches to organizational structures and dynamics such
as Cohen, March, and Olsen’s (1972) Garbage Can Model of Organizational Choice,
March’s (1991) Exploration and Exploitation in Organizational Learning, Carley’s
(1992) Organizational Learning and Personnel Turnover, to name but a few.
While the model is first and foremost a mathematical conceptualization of the
theory at hand, the simulation is a scientific inquiry ‘‘halfway between theory and
experiment’’ (Waldrop, 1992, p. 63), a blend of deduction and induction. It is not an
end in itself, however, but a means of manipulating the structures and dynamics of
organizations in order to observe corresponding organizational behaviors. Therefore,
the simulation findings allow for conclusions which are not necessarily evident from
the theory and model alone. In closing, I discuss some of the findings in terms of
future theoretical contemplation and empirical research.
2 Fundamentals
this distinction, systems are mutually environment to each other; psychic systems are
environment to social systems, and vice versa. (Surprisingly, already March and
Simon (1958, p. 2) write, ‘‘for most people formal organizations represent a major
part of the environment.’’) This circular exclusion deprives each system of any
immediate influence on respective others since neither one can determine system/
environment distinctions apart from its own.
Although systems cannot operate outside their boundary, they are by no means
isolated in their existence. Psychic and social systems surmount their operational
closure by observing other systems in their environment (Luhmann, 1986). In other
words, individuals incorporate observations of organization in their network of
consciousness, while organizations incorporate observations of consciousness in their
network of communication. By and large, the reciprocity of boundary-spanning
observations among psychic and social systems accounts for structural drifts (cf.
Weick (1976) and Orton and Weick (1990) on loosely coupled systems). Moreover,
the contingency of consciousness and communication forces individuals and
organizations to attribute these drifts to environmental irritations (disturbances,
perturbations, stimulations, etc.), mainly because they cannot think or communicate
about everything and all. Therefore, psychic and social systems are necessarily less
complex than their environment (Luhmann, 1995, p. 182).
In addition to the observation of other systems in the environment, individuals and
organizations incorporate observations of themselves in consciousness and commu-
nication, respectively. Indeed, (self-)observations are but operations to begin with.
The distinction between operations and observations merely presents the reentry of
past system/environment distinctions (i.e., previous consciousness or communication)
into present systems’ operations (Luhmann, 1995, p. 36; Spencer-Brown, 1979, p. 49).
Simply put, psychic and social systems uphold their autopoiesis by connecting past,
present, and future operations in observation.
Next to the system/environment and the operation/observation distinction, I must
briefly review the complementing (organizational) concepts of communication and
expectation. With a particular interest in organizations, the respective complementing
(individual) concepts of consciousness and expectation are left unaddressed. For the
most part, however, the following review holds true for both organizations and
individuals. Now, to ask with Luhmann (1992), What is Communication?
Just like life and consciousness, communication is an emergent reality, a state of affairs sui
generis. It arises through a synthesis of three different selections, namely, selection of information,
selection of utterance of this information, and a selective understanding or misunderstanding of
this utterance and its information.
None of these components can be present by itself. Only together can they create
communication. Only together — and that means only when their selectivity can be made
congruent. Therefore communication occurs only when a difference of utterance and information
is understood. That distinguishes it from mere perception of the behavior of others. In
understanding, communication grasps a distinction between the information value of its content
and the reasons for which the content was uttered. It can thereby emphasize one or the other side.
It can concern itself more with the information itself or with the expressive behavior. But it
always depends on the fact that both are experienced as selection and thereby distinguished.
(Luhmann, 1992, p. 252)
218 Steffen Blaschke
3 Theory
In the past 50 years, management science and organization theory cover individual and
organizational knowledge, learning, and memory in great detail (for a Trip Down
Memory Lane, see Walsh, 1995). The following contemplations are easily understood
as critique of mainstream theory, but this is rather circumstantial. Autopoietic
organization theory is a mere alternative next to established schools of thought; in
some parts, it breaks away from mainstream theory, in other parts, it complements
management science and organization theory. For more elaborate critique and further
reference to the concept of autopoiesis in terms of knowledge, learning, and memory,
turn to von Krogh and Roos (1995), Magalhaes (1996), and Blaschke (2008).
To begin with, I elaborate on knowledge. For lack of a better linguistic distinction,
then, I discuss learning and unlearning in terms of learning. Last but not least,
remembering and forgetting amount to the more generous term memory. Without
further ado, Figure 2 illustrates the autopoiesis of organizational knowledge, learning,
and memory. Knowledge is situated in the present organizational state of affairs,
learning anticipates future operations, and memory contemplates past ones.
3.1 Knowledge
one truth, just as there is not one environment (Luhmann, 1995, p. 181ff.; Weick,
1969, p. 63ff.) or one reality (Berger & Luckmann, 1966; von Foerster, 2003).
Nonetheless, a definition of knowledge as justified belief is considerably closer to
the argument of social systems theory that knowledge is generalized cognitive
expectation (Luhmann, 1995, p. 328). In other words, autopoietic organizational
knowledge is expectation which provides communication with genuine (thus justified,
thus generalized) system/environment distinctions. Recall that although expectation
eases the selection pressure of communication, it carries an inherent uncertainty with
respect to environmental dynamics (Section 2). Then again, knowledge as generalized
cognitive expectation brings a certainty to communication in that it indicates which
system/environment distinctions are genuine.
Let me give an example. As many others, my wife and I generally expect the sun to
rise in the East. Fortunately, our balcony faces in just that direction, and so the
decision to have our morning coffee out on the balcony is usually rewarded with a
little sunshine. The movement of the sun is common knowledge to many others
besides my wife and me, of course. There is rarely a need for us to explicitly make the
sunrise or sunset topic of our communication; still, it implicitly accompanies the
decision to have our morning coffee out on the balcony (cf. Polanyi, 1958, on the
matter of implicit and explicit knowledge).
The Autopoiesis of Organizational Knowledge, Learning, and Memory 221
3.2 Learning
(i.e., inert), and organizations must unlearn their own past before any partial
structural changes come about at all. These and other important concepts are beyond
the scope of my argument which, in turn, is simply that autopoietic learning is
essential to organizations, just as knowledge is, just as memory is.
3.3 Memory
4 Model
The above fundamentals of social systems theory and the introduction to knowledge,
learning, and memory in the light of autopoietic organization theory are rather
abstract, so far. The following model elaborates on these structures and dynamics of
organizations. It bridges the gap between theoretical contemplation and empirical
research by means of a mathematical conceptualization which, in turn, provides the
basis of a later simulation (Section 5).
Communication emerges from three different selections, namely, the selection of
information, the selection of utterance of this information, and the selective
understanding or misunderstanding of this utterance and its information (Luhmann,
1992). (Note that while Luhmann (1992) makes a case in point in always speaking of
both understanding and misunderstanding, I omit the concept of misunderstanding
from further discussion and illustration in order to keep the model as simple as
possible.) Information, utterance, and understanding, then, are best illustrated with a
binary value of 1 (for a selection) or 0 (for no selection). Hence, communication
emerges from the selection of information, utterance, and understanding if — and
only if — each one of these selections displays a value of 1. Think of the emergence of
communication as information utterance understanding. Obviously, if any one of
these selections displays a value of 0, no communication emerges to begin with.
Now, communication synthesizes a decision, topic, or theme in observation of
consciousness or expectation, that is, it observes a system/environment distinction.
Communication may bring forth a decision to have the morning coffee on the balcony
or in the kitchen, for example. Any such system/environment distinction generally
implies a particular quality (balcony or kitchen?) which is best illustrated with a ternary
value of 1 (for a particular quality of a decision, say, balcony), 1 (for a respective
other quality of the decision, say, kitchen), or 0 (for no quality, i.e., no decision). To
continue the example, the decision on where to have the morning coffee may display a
value of 1 in case consciousness or expectation indicate the balcony as a suitable
location, a value of 1 in case the kitchen is more preferable, or a value of 0 in case
communication brings forth no decision at all. Think of the synthesis of communica-
tion as consciousness or expectation information utterance understanding. Table 1
illustrates the emergence and synthesis of communication on m dimensions.
m 1 2 3 4 5 6 7 8 9
Consciousness 1 0 1 1 0 1 1 1 0
(Expectation) 0 1 1 0 1 0 0 1 1
Information 1 0 1 1 1 0 1 1 0
Utterance 0 1 1 1 1 1 0 1 0
Understanding 0 0 1 1 1 1 1 1 0
Communication 0 0 1 1 1 0 0 1 0
224 Steffen Blaschke
m 1 2 3 4 5 6 7 8 9
Consciousness 1 0 1 1 0 1 1 1 0
(Expectation) 0 1 1 0 1 0 0 1 1
Communication 0 0 1 1 1 0 0 1 0
Environment 1 1 1 1 1 1 1 1 1
communication brings forth a decision in line with the environment, still. The
knowledge that accompanies communication, however, disappoints in the light of the
task. Fortunately, the possibility of organizational learning and with a change in
expectation is at close hand.
Autopoietic organizational learning takes place in that expectation changes its quality
(e.g., from a value of 1, as illustrated on the eighth dimension, m ¼ 8) to the quality of
communication (i.e., to a value of 1). The probability of learning depends on a number
of factors (e.g., organizational inertia), though it generally increases with the continuous
disappointment of knowledge. That is to say, the longer expectation fails to provide
communication with genuine system/environment distinctions, the more likely it is to
change to whatever quality communication reflects. Now, learning is possible and
indeed probable even if expectation reflects no quality to begin with (i.e., the value is 0,
cf. m ¼ 4). No knowledge or, rather, non-knowledge disappoints in the light of the task,
whereupon communication generalizes expectation. As is the case on the fourth
dimension (m ¼ 4), expectation may only change its quality to that of communication,
that is, its value of 0 may only change to a value of 1, despite the fact that knowledge is
bound to disappoint. Therefore, learning is not necessarily advantageous.
Table 2 further illustrates autopoietic organizational memory. Indeed, memory
accompanies communication on all dimensions in that it either remembers or forgets
system/environment distinctions. Remembering specifically takes place on the third,
fourth, fifth, and eighth dimension (m ¼ 3, 4, 5, and 8), whereas forgetting accounts for
the remaining five dimensions (i.e., m ¼ 1, 2, 6, 7, and 9). Similar to learning, which
brings about partial structural changes in expectation, memory brings about partial
structural changes in communication, too. In contrast to learning, however,
remembering whatever is forgotten or forgetting whatever is remembered is always
possible, there is no need for knowledge to disappoint first. The probability of
remembering and forgetting depends on many the same factors that facilitate or impede
learning (e.g., organizational inertia). With respect to the environment, memory is likely
to partially change communication on dimensions 2 and 9 (m ¼ 2 and 9), that is, it may
remember system/environment distinctions which presently are forgotten.
5 Simulation
In addition, it provides the basis for (computational) simulation. Standing half way
between theory and practice, simulation is a third way of scientific inquiry. In general,
it takes three major steps: (1) the deductive development of a theoretical model, (2)
the implementation of the model in an experimental design, and (3) the inductive
analysis of the simulation results (Axelrod, 1997). Section 4 already outlines the first
step. Due to page constraints, I omit the implementation of the above model of
autopoietic organizational knowledge, learning, and memory altogether. My recent
work on the Structures and Dynamics of Autopoietic Organizations (Blaschke, 2008)
contemplates this second step in detail. In the remainder of this section, I jump right
to the third step and discuss some of the simulation results.
Modeling operations and observations of psychic and social systems as binary and
ternary vectors allows for a mathematical approach to the structures and dynamics of
organizations (i.e., knowledge, learning, and memory). Equation (1), for example,
denotes autopoietic organizational knowledge, OrgKnw(t), at time t,
(
1X m 0; Comi Envi 0
OrgKnwðtÞ ¼ Comi ðtÞ Envi ðtÞ (1)
m i¼1 1; Comi Envi 40
Comi(t) returns the value the ternary vector for communication holds in dimension i,
and Envi(t) returns the value the binary vector for the environment holds in
dimension i. The index variable i runs from 1 to m, then. If communication fails to
bring forth a decision (Comi ¼ 0) or the decision is different from the element of the
task it aims at (Comi6¼Envi), then the equation yields a value of 0; in case of a decision
in line with the task, it yields a value of 1. The sum of positive values (i.e.,
communication in line with the environment) over the number of dimensions (m)
accounts for autopoietic organizational knowledge, for instance, 3/9 or 0.3, as
illustrated in Table 2.
While knowledge reflects the structure of communication, learning and memory
reflect the dynamics of expectation and communication. Autopoietic organizational
learning takes place with the partial structural change in expectation, that is, with a
change in value from 0 to 1 or 1, from 1 to 1, or from 1 to 1. (In contrast, a
change in value from 1 or 1 to 0 denotes a rare case of amnesia.) Likewise,
autopoietic organizational memory takes place with the partial structural change in
communication, that is, with a change in value from 0 to 1 or 1 (remembering) or,
the other way around, from 1 or 1 to 0 (forgetting). The production and
reproduction of communication with a value of 1 or 1 pertains to remembering, too,
just as forgetting retains a communication with a value of 0. At this, autopoietic
organizational memory preserves some system/environment distinctions, while it
interrupts others (Luhmann, 1996).
Figure 3 displays autopoietic organizational knowledge, learning, and memory or,
rather, forgetting. The quantitative results of the simulation depend on initial
conditions, implementation specifications, and the like, but the qualitative results are
insensitive to these matters. Organizational knowledge at a level of 0.5 indicates that
50% of all decisions made are in line with the environment, learning at the same level
The Autopoiesis of Organizational Knowledge, Learning, and Memory 227
6 Conclusion
Social systems theory and autopoietic organization theory largely deny empirical
research on individuals and organizations, as the critique readily points out (Blühdorn,
2000; Viskovatoff, 1999; Thyssen, 2003). The model and simulation of organizational
structures and dynamics, however, address issues of knowledge, learning, and memory
beyond mere theoretical contemplation. They conceptualize the autopoiesis of
knowledge, learning, and memory as essential to organizations themselves and, at
the same time, allow for a better understanding (knowledge?) of organizations per se.
Mathematical approaches to the structures and dynamics of organizations are
quite common in management science and organization theory (e.g., March, 1991;
Carley, 1992; Cohen et al., 1972; Harrison & Carroll, 1991). Practice makes good use
of many such models, too. Organizations communicate their intellectual property in
balance sheets (Sveiby, 1997), for example. Unfortunately, any such static account of
knowledge is a snapshot at best and, more likely, produces a distorted picture
altogether, not the least because organizations frequently know more than they are
able to recall in a single year’s end survey.
A dynamic account of knowledge is a more appropriate way to the analysis of
autopoietic organizations, then. If organizations continuously keep record of their
decisions and match them with the problems, tasks, or the general business they face,
then they may report this production and reproduction of knowledge at any given time.
The evaluation of decisions is almost natural to organizations, particularly to for-profit
corporations. Indeed, organizations frequently employ decision support systems,
project management tools, and business planning software as a means to just this end.
Alas, they rarely rely on a rigorous mathematical conceptualization of theory.
Communication about knowledge is strenuous, but monitoring is ever the easier
with new information technologies. For instance, collaborative authoring systems
(e.g., wikis) provide a complete history of all changes made to documents which, of
course, comprise decisions, if they are identified as such. Empirical research may tap
into this resource of computer-mediated communication and process the data
according to the model of autopoietic organizational knowledge, learning, and
memory. Additional simulation may serve a number of purposes, for instance, the
discovery of knowledge with respect to new environmental challenges or the
prediction of learning and memory in the light of different future scenarios.
The Autopoiesis of Organizational Knowledge, Learning, and Memory 229
Let me finish by returning to the above example on last time. Each and every
morning, my wife and I face a task we only refer to as ‘‘coffee?’’ As simple as the task
is, it actually requires a number of decisions, namely, who will get up first to actually
make the coffee, where to have it, and the like. At this, we rely on our knowledge about
who is best to operate the coffee machine, about the sunrise (remember, we like to have
our morning coffee out on the balcony), and many more system/environment
distinctions. Knowledge, learning, and memory are implicit to our decisions; we hardly
ever talk about it. If we were to assess the knowledge required to successfully
accomplish the task (‘‘coffee?’’), we could simply document our decisions and how they
turn out. ‘‘Honey, it’s your turn to make coffee’’ is another frequent sentence I hear
from my wife, and the next time around I will put down somewhere that I made her
coffee, indeed. As flattered as I am to hear my wife praise my coffee, it only generalizes
the expectation that she favors me making her coffee. But we both know that.
References
Allee, V. (2003). The future of knowledge: Increasing prosperity through value networks.
Amsterdam: Butterworth-Heinemann.
Anand, V., Manz, C. C., & Glick, W. H. (1998). An organizational memory approach to
information management. Academy of Management Review, 23(4), 796–809.
Argote, L. (1999). Organizational learning: Creating, retaining and transferring knowledge.
Boston, MA: Kluwer Academic.
Argyris, C. (1976). Single-loop and double-loop models in research on decision making.
Administrative Science Quarterly, 21(3), 363–377.
Axelrod, R. (1997). Advancing the art of simulation in the social sciences. In: R. Conte, R.
Hegselmann & P. Terna (Eds), Simulating social phenomena, lecture notes in economics and
mathematical systems (pp. 21–40). Berlin: Springer.
Baecker, D. (2003). Organisation und Management. Frankfurt am Main: Suhrkamp.
Bakken, T. & T. Hernes. (Eds). (2003). Autopoietic organization theory: Drawing on Niklas
Luhmann’s social systems perspective. Oslo: Copenhagen Business School Press.
Bateson, G. (1987). Steps to an ecology of mind: Collected essays in anthropology, psychatry,
evolution, and epistemology. Northvale, NJ: Aronson.
Berger, P., & Luckmann, T. (1966). The social construction of reality: A treatise in the sociology
of knowledge. Garden City, NY: Doubleday.
Blaschke, S. (2008). Structures and dynamics of autopoietic organizations: Theory and simulation.
Wiesbaden: Gabler.
Blaschke, S., & Schoeneborn, D. (2006). The forgotten function of forgetting: Re-visiting
exploration and exploitation in organizational learning. Soziale Systeme, 12(1), 100–120.
Blühdorn, I. (2000). An offer one might prefer to refuse: The systems theoretical legacy of
Niklas Luhmann. European Journal of Social Theory, 3(3), 339–354.
Carley, K. M. (1992). Organizational learning and personnel turnover. Organization Science,
3(1), 20–47.
Cohen, M. D., March, J. G., & Olsen, J. P. (1972). A garbage can model of organizational
choice. Administrative Science Quarterly, 17(1), 1–25.
Davenport, T. H., & Prusak, L. (1998). Working knowledge. Cambridge, MA: Harvard
University Press.
230 Steffen Blaschke
1 Introduction
Over the more than 15 years since the term knowledge management gained popular
currency in organizations, considerable debate has ensued about how it may be
conceptualized. Indeed, within organizational research, there would appear to be at
least three broad streams of study relating to knowledge and its management (Sveiby,
1994; von Krogh & Roos, 1996), with many subthemes emerging under each as the
field has developed. Considering each stream from the perspective of the
epistemological assumptions it embraces, the first conceives of knowledge as an
object. This stream is often associated with information science and information
markets. von Krogh and Roos (1996) refer to this stream as the ‘‘Information
Processing epistemology.’’ The key concern of this stream is with the codification of
‘‘knowledge’’ into units of information that can be easily moved, sold, or attributed
value in some form.
The second stream of research views knowledge more as a process and is concerned
with the behavioral aspects of organizational life and their effect on the retention and
transfer of knowledge throughout the organization. von Krogh and Roos (1996)
referred to this as the ‘‘Network Epistemology.’’ The key concern in this stream is
with the different ways of connecting people within the organization, with
information systems often playing a central role as supposedly the most cost-effective
means by which this can be achieved. Epistemologically, as with the first stream, a
reified view of knowledge is adopted. Partly due to this, a further splintering of
approaches has become evident:
that of the individual, where knowledge is seen as the property of individual people
(Polanyi, 1958, 1967), and
that of the organization, where the organization itself is viewed as having knowledge
(Walsh & Ungson, 1991; Weick & Roberts, 1993).
In terms of bringing together complexity and KM, this third stream is significant
for one simple reason: it moves the discussion of knowledge away from the reified
perspectives common to the other streams of KM research to a perspective which is
more aligned to the approach discussed by Cilliers (2000b).
Knowledge begins with cognition, and our cognitive capacity is inherent in our
biology. Any theory of KM and consequently any application of complexity needs to
be consistent with our understanding of the biology of cognition. If this is accepted,
then one of the first things we need to rethink is the widespread adoption within KM
of the representationalist view of cognition. The idea that we capture understanding
of the world in our brains — using some form of symbolic representation as with
digital machines — is no longer defensible; we know our brains simply do not work
this way. Continuing to use such a metaphor supports the continued idea that KM
can be associated with the capturing of salient representations about the real world
and the idea that the more ‘‘accurate’’ the representation the better the knowledge.
This is inconsistent with what biology tells us about how we ‘‘know’’ anything. Once
this idea is rejected, with it falls the idea that human action is guided primarily by a
rational weighing of facts about the real world and/or on the calculation of some
‘‘utility’’ by each individual. This assumption is of course the foundation stone of
most management and organization theory.
Within KM, complexity theory is primarily argued to be relevant to understanding
macro organizational dynamics. What we are arguing here is that the macro-level
dynamics of organizations emerge from the activities of micro-level human agents.
Complexity therefore has further relevance in that those human agents are themselves
macro phenomena — emerging from the interaction of microbiological agents —
cells. To build a bridge between complexity and KM, we ideally need a set if internally
consistent theories that allow us to traverse this cascade of micro–macro
boundaries — from our molecular foundations to the highest order social
phenomena. At the same time we need to reject the convenient way in which
sociologists and organization theorists have avoided becoming entangled in the
micro–macro problem by quarantining different levels of analysis. While convenient
236 Robert Kay and Chris Goldspink
it means that we can never really explain the emergent nature of social phenomena —
we can only provide incomplete descriptions. In this context, the field of complexity is
providing us with the tools needed to move beyond this situation.
Our goal in this chapter is to set a model of organization which has the following
attributes:
different implications for the viability of that person in his/her environment, given
his/her history. Although the nervous system is operationally closed it is plastic, its
structure changes over time, and it is this quality that allows for changes in behavior
and subsequently what we describe as learning (Mingers, 1991). Therefore as the
nervous systems structure changes, so too will the potential range of behaviors that its
structural determinacy makes possible. The term used for this history of structural
change is ontogeny (Maturana & Varela, 1992).
Where there is a history of recurrent interaction between two individuals, a
structural congruence can develop — they become structurally coupled. We have
argued (Goldspink & Kay, 2003, 2004) that structural coupling is the mechanism by
which all social structures emerge including what we refer to as organizations. Thus,
structural coupling constitutes the building block for organizational knowledge.
This description of autopoietic theory should only be considered as a cursory
introduction to some of the major concepts. The significance of these ideas, however,
becomes apparent when they are applied to the notions of cognition, knowledge, and
organizations, as they define the process by which the individual comes to know of
their environment and orient themselves within it.
Varela, Thompson, and Rosch (1992) have developed a theory of cognition consistent
with the autopoietic nature of living systems. They state that cognition takes place
whenever an organism behaves in a manner consistent with its maintenance and
without loss of identity, i.e., without loss of any of its defining characteristics. This
theory challenges the most commonly accepted view of human cognition — that of
cognitivism or representationalism. Specifically they state:
The central intuition behind cognitivism is that intelligence — human intelligence included — so
resembles computation in its essential characteristics that cognition can actually be defined as
computations of symbolic representations (Varela et al., 1992, p. 40; our emphasis].
Thus, representations are defined teleologically they are intentional, and are
‘‘about something for the system’’ (Varela et al., 1992, p. 44). Cognitivism constructs
a duality: the environment is experienced as a facticity and acted upon directly, but is
also conceived and symbolically represented in the mind. Mind and behavior are
linked as hypothesis and experiment. This way of understanding human cognition is
being increasingly criticized from within biology and other disciplines such as artificial
intelligence (AI) and by complexity theorists (Cilliers, 1998, 2000a; Stacey, 2001).
Both AI and complexity theories have given greater impetus to connectionist
theories of cognition (Cilliers, 1998). Here emergent structure or pattern arises from
massively interconnected webs of active agents. Applied to the brain, Varela et al.
point out:
The brain is thus a highly cooperative system: the dense interconnections amongst its
components entail that eventually everything going on will be a function of what all the other
components are doing. (1992, p. 94)
238 Robert Kay and Chris Goldspink
Maturana and Varela have provided a biological grounding point for a complex
systems understanding of knowledge in organizations. In this understanding,
knowledge is defined as a process of bringing forth ‘‘a’’ world. That world is the
lived experience of the individual as he/she responds to his/her environment. Within
this conceptualization, the notions of ‘‘doing,’’ ‘‘being,’’ and ‘‘knowing’’ are all bound
into the single notion of knowledge and all are subject to structure-determined
processes of change. The range of behaviors available will depend on the individual’s
history of interaction with others — his/her ontogeny. As the nervous system’s
structure changes in response to the environment, so too will the potential range of
behaviors that its structural determinacy makes possible. As an observer we might call
someone ‘‘more knowledgeable’’ if, after observing them, we notice that they generate
a wider range of behaviors or are more successful at satisficing the constraints they
confront in that environment. Over time a human being may extend the behavioral
repertoire he/she can generate and this we call learning — the gaining of knowledge.
Knowledge may therefore be considered as the range of actual and potential
behaviors that an individual may generate to respond to and remain viable within any
given environment. This is however a judgment made by an observer and is a
comment on that observer’s assessment of the quality of the responses the subject
generates. The observer is noticing a macro (emergent) pattern and ascribing to it
certain qualities. This brings us to how we may understand the relationship between
individual learning (as described above) and an organization’s capacity to survive in
its environment.
We have so far considered the pathway by which order is generated bottom-up —
from micro to macro in a way that reconciles it with biological constraints and
describes the particular mechanisms that operate with biological agents — humans.
But what of the top-down, macro and micro processes; how can these be resolved
within this framework? Within autopoietic theory this top-down process is explained
through the dynamics of the individual’s relationship with his/her environment. The
ontogeny of each individual, while unique to that individual, is also a product of his/
her interactions with others — it is a co-evolved structure which is the product of
structural coupling. Each agent is constrained to interact in a limited way not by the
other agents in a direct causal sense but by virtue of its structural determinacy. The
individual adjusts his/her behavioral repertoire as a function of his/her interactions
with those other agents (his/her nervous system is now geared to produce a learned
limited subset of responses). In this sense the collective aspect of knowledge is ‘‘in’’
each agent — it is reflected in the constraints embedded in his/her nervous system.
What we have described so far could be applied to any social animal. There is
however another pathway between macro and micro that is unique to humans.1
1
It may apply to other organisms which have evolved a capacity for self-awareness, such as some apes and
cetaceans and possibly elephants, but the evidence is not yet clear on this, perhaps due to our inability to
create a common language with them.
240 Robert Kay and Chris Goldspink
the observer is a participant in the process (i.e., is one of the agents of the network
which is giving rise to the observed pattern) or
the observer was external, but becomes a part of that system by communicating his/
her observations such that those being observed respond by changing their
behavior.
are not entities in a distinguishing sense (Kay, 2001). Indeed this point is emphasized
in the ongoing debates within the literature on autopoietic theory regarding the
ontological nature of organizations. In our view, to consider organizations as
knowledgeable in themselves constitutes an inappropriate reification which denies the
basis for their emergence in and through the structural coupling of the humans which
give rise to them. Organizations are distinctions not distinguishers. This is to say
organizations cannot ‘‘have’’ knowledge.
That organizations can be distinguished by an observer to have changed their
patterns, exhibiting variation, selection, and retention processes (Aldrich, 1999) in a
purposeful way is to confuse outcome with process. Complexity theory can aid the
development of our understanding of the process by which these changes occur.
However, it can only do this if we are mindful of what the general principles of
complex systems translate to within biological and, more specifically, human systems.
The reflexive and recursive nature of this process, as we have seen above, requires a
clear epistemology accounting. Without this accounting it is easy to become confused
and to see knowledge as a cause rather than the endpoint of a process.
5 Conclusion
Autopoietic theory combined with complexity theory has been demonstrated here to
have profound implications for our thinking about knowledge and organizations and
hence KM. When applied to the domain of social action, the resulting theory implies
the need to adopt a radically different view of the origins and nature of knowledge.
This is because when we consider that organizations are complex systems of a
particular class — ones comprised of human (biological reflexive) agents, we are
drawn to a very different understanding of the origins of social structure and hence
the nature of organizations as well as the nature of cognition. As well as causing us to
rethink the implications of complexity for knowledge at an individual level, this
approach requires us to question what we understand about the constitutive nature of
an ‘‘organization.’’ We are compelled to consider the organization as a distinction
made by an observer. As a distinction (and not a thing), it cannot itself make
distinctions and so cannot ‘‘have’’ knowledge.
Here we have presented a view of knowledge where what the observer identifies as
‘‘knowledge’’ is an attribution based on the observation of coordinated behaviors at
some level of the organization. Knowledge does not cause anything — it is not a basis
for, nor the origin of coordinated behavior; it is an attribution that denotes (for the
observer) the presence of some attribute in the quality of interaction being observed.
References
Aldrich, H. (1999). Organizations evolving. London: SAGE.
242 Robert Kay and Chris Goldspink
1 Introduction
systems are discrete, autonomous systems whose identity remains constant through
the continual maintenance of their organization. Evidently a ‘‘systems’’ based theory,
autopoiesis lends itself nicely to being applied to other domains. Organizational
learning, which is often considered a ‘‘soft’’ aspect of knowledge management, will
undoubtedly benefit from autopoietic theories detailing learning, adaptation, and
evolution. Epistemological debates can be settled using the scientific ideas autopoiesis
proposes about knowledge and its structure. The application of autopoiesis to
knowledge management also brings with it the added bonus of giving knowledge
management, a sound theoretical foundation, subsequently confirming the need for
an interdisciplinary approach to the current problems in knowledge management.
However, autopoiesis cannot be randomly applied to knowledge management or
organizational learning; a rigorous and coordinated method is needed, and this is
provided by a matching methodology. Matching is a two-step process, comprised of
theoretical discourse and inscription. Theoretical discourse is the unification of two
theories through frequent dialogue. The output is a new lexis, unique to the merging
of the two domains. The lexis then feeds into the inscription process, which captures
and makes an object out of the knowledge previously constructed through the
theoretical discourse. Matching is different from positivist and interpretivist
approaches to research due to this inscription process. Essentially, knowledge is
presented in such a way that it can help other theory building attempts (in this case,
knowledge management).
This chapter will, through the process of inscription, give third era knowledge
management initiatives a theoretical foundation, such that knowledge management
will carry greater scientific weighting.
interactions and transformations continuously regenerate and realize the network of processes
(relations) that produced them; and (ii) 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. (Maturana & Varela, 1980)
As such, the key feature of an autopoietic entity is the constant maintenance of its
organization, or network of processes that realize the entity as living. There are four
consequences of an entity being autopoietic: autonomy, individuality, organizational
closure, and self-specification of boundaries (Maturana & Varela, 1980). Autonomy
is the ability of an entity to specify its own laws and the behavior it exhibits
(Maturana & Varela, 1998). The view that living entities are autonomous also
contributes to the individuality of living entities. Maintaining their organization as
autopoietic, living entities are also actively maintaining their identity (Maturana &
Varela, 1998). Organizational closure is an essential feature of autopoietic entities, if
they are going to remain living; if they did not maintain their autopoietic
organization, they would disintegrate and die. However, just because a system is
organizationally closed does not mean it cannot receive physical inputs (Mingers,
1995). An autopoietic entity is also able to specify its own boundaries. In the case of a
cell, the internal dynamics produce the necessary components for the boundary, while
at the same time, the boundary contains the processes of self-production (Maturana
& Varela, 1998).
As outlined by Varela, Maturana, and Uribe (1974), there are also six principles of
autopoiesis, which are presented in Figure 1. These six principles were developed to
determine if a given entity is autopoietic, and also used to create a computer model of
autopoiesis in a chemical scenario (McMullin, 2004).
Knowledge management is a discipline that has emerged from the move toward a
knowledge economy (Green, 2005) where knowledge is the key asset in an
organization. Knowledge management, the culmination of many different disciplines,
can be defined as:
the effective learning processes associated with exploration, exploitation and sharing of human
knowledge (tacit and explicit) that use appropriate technology and cultural environments to
enhance an organization’s intellectual capital and performance. (Jashapara, 2004)
focuses on activities that help the individual to learn, or solve problems on their
own and is generally accepted as following the Lewinian experiential model (Kolb,
1984). The Lewinian model of learning is a four-stage cycle, which places great
importance on experience. The cycle starts with a concrete experience, which gives an
opportunity for observation and reflection. These observations and reflections are
then formed into generalizations or theories, which, in the final stage, are tested in
new situations (Kolb, 1984). The emphasis of this cycle is that learning is a continual
process based on a person’s experience and testing of abstract concepts they develop
as a response. Based on experience, it would appear that Kolb’s philosophy is the
guiding ethos in statements such as ‘‘learn by doing’’ and ‘‘practice makes perfect,’’
showing that there has to be some element of practical experience involved in
learning.
Team learning is when individuals ‘‘solve problems by drawing on the strengths of
other members in a team’’ (Yeo, 2005). The third process, organizational learning, is
somewhat different from the first two processes, since the focus is on external
resources. The main objective in organizational learning is to ‘‘develop new principles,
positions, aims, roles, and identity in preparing the organization for the dynamic
changes of the external environment’’ (Yeo, 2005). These three processes are
important in organizational learning, since they show firstly that organizational
learning occurs when individuals learn, and secondly that individuals are perhaps the
most important feature in organizational learning. This means a direct link can be
made from individual learning, through team learning, to organizational learning.
For the purposes of this research, organizational learning is a concept taken to exist
under the wider discipline of knowledge management.
The shift toward the knowledge-based organization that occurred in the 1990s
(Johanessen, Olaisen, & Olsen, 1999) instigated a paradigmatic shift in the workings
and management of organizations. A focus on internal motivation, managing
relationships, and widespread idea generation became important, and subsequently,
knowledge management became increasingly important. However, it quickly became
apparent that knowledge management was growing from several disciplines
(Jashapara, 2004), and as such lacked its own theoretical foundation.
The argument for applying autopoiesis to knowledge management has already
been presented (Johanessen et al., 1999; Scholl et al., 2004), outlining the shift in
organizational philosophy and management style, and the need for an interdisci-
plinary approach. However, a review of current applications of autopoiesis reveals
some significant variations in the approach used and results gained. Current
applications of autopoiesis to knowledge management include creating an epistemol-
ogy, treating organizations as first-order autopoietic entities, and using organizational
learning as an alternative, better suited approach.
248 Paul Parboteeah et al.
Epistemological research in the field of autopoiesis typically takes one of two paths: one,
assuming knowledge is autopoietic itself and another, suggesting knowledge is an
emergent property of second-order autopoietic systems. This difference can also be
traced back to the debate concerning whether autopoiesis can exist outside the molecular
domain. Authors proposing knowledge itself is autopoietic (Hall, 2005) believe that
autopoiesis can be applied to conceptual and other physical domains and ultimately
that knowledge is living. Authors proposing that knowledge is an emergent property
(Abou-Zeid, 2007) believe that knowledge is embodied in the knower, and subsequently
cannot be separated from them. Any activity in the field of knowledge management
should start from an autopoietic definition of knowledge (Limone & Bastias, 2006). The
rationale being that since organizations are cognitive systems, and are third-order
autopoietic systems, any knowledge management effort should entail a cognitive aspect.
It is possible that knowledge is a biological entity (Hall, 2005), and as such inherits
the characteristics of any biological being, typically a system and a control
mechanism. In which case two types of knowledge would appear to exist: embodied
and encoded (Hall, 2005). Embodied knowledge, also known as tacit knowledge, is
that which the autopoietic system would normally gain through its activities. Encoded
knowledge, or ‘‘control information,’’ is knowledge encoded into the systems
structure, such that it is used for that system’s survival. Hall’s (2005) concept of
control information seems to bear a striking resemblance to that of DNA. This
appears to be a reappearance of the idea put forward by Luisi (2003) that autopoiesis
provides the biologic for the operation of the autopoietic entity.
An alternative view of autopoietic knowledge arises from the perspective that
autopoiesis cannot exist outside of the molecular domain. This view proposes that
knowledge is embodied in the knower, and cannot be stored, transferred, or externally
manipulated (Abou-Zeid, 2007; Biggiero, 2007). Knowledge of this form is always
private, and that only information or data can be stored, transferred, or manipulated.
With this as an epistemological base, it becomes difficult to see how knowledge can be
managed. From this viewpoint, it would appear all that can be done is try and
support people learning and acquiring knowledge by themselves. With this in mind, it
is possible to create a knowledge management support system (Abou-Zeid, 2007). The
design of a knowledge management support system should feature two parts: one for
the actual system and one for the procedures of designing the system, or ‘‘meta-
design.’’ Such an approach would ensure that the principles of autopoiesis were
inherent in the design of the system.
A less explored aspect of autopoietic knowledge is the notion that knowing is a
process intertwined with the process of living. Knowing can be defined as leading to
‘‘effective action, that is, operating effectively in the domain of existence of living
beings’’ (Maturana & Varela, 1998). The essence of this definition is that knowledge is
the key to effective action, and that perhaps through the process of living, and acting,
that knowledge may be admitted. An option that does not appear to have been
explored in the literature is whether observation of, and participation in, effective
action leads to the admittance of knowledge, whatever the form of knowledge may be.
Autopoiesis as the Foundation for Knowledge Management 249
However, trying to follow a line of research could result in numerous problems, such
as trying to define effective action, trying to evaluate whether any knowledge had
been admitted, and whether that knowledge was the correct knowledge. It would seem
prudent to end with Biggiero’s (2007) statement that ‘‘explicit knowledge is an
oxymoron.’’ In other words, all knowledge is embodied within the knower, and
subsequently, knowledge management systems trying to directly manage knowledge
will fail. The position taken in this research will be akin to Abou-Zeid’s (2007) that
knowledge cannot be stored, manipulated, or transferred: it is embodied in the
knower, along with Biggiero’s (2007) view that all knowledge is private and only data
or information can be transferred.
such as the spontaneous emergence of autopoietic entities are not used. Whilst not
using autopoiesis in its true sense, or even using autopoiesis in the context of second-/
third-order autopoiesis, the model proposed appears to work in its context. This
could be because only aspects of autopoiesis needed are used, with the rest discarded.
This resistance not to use all of autopoiesis just because it is possible has benefited the
model being developed. However, because only parts of autopoiesis have been used,
and other parts of the model do not obey autopoiesis, the final model should not be
called autopoietic, as has been done (Thannhuber et al., 2001).
monitors itself through a feedback loop, which effectively defines the system as self-
regulating. Second-order cybernetics arose when researchers began applying
cybernetic principles to the role of the observer, recognizing that the very act of
observing a system would change the system under observation. Autopoiesis is
undoubtedly a second-order cybernetic theory, with its continual focus on the
observer and its impact on the system under study.
Maula (2006) defines the organization as a living system, and proposes a ‘‘living
composition’’ as an enabling infrastructure. However, the underlying problem in
Maula’s (2006) model is that it also assumes organizations are first-order autopoietic
entities, and subsequently falls into the problems described earlier. Maula also
discusses two knowledge flows: ‘‘sensing’’ and ‘‘memory’’ (2006). The problem of
objectifying knowledge in this way assumes that knowledge takes on a form that it
was not meant to, namely that it can exist outside the knower. While Maula (2006)
does consider learning as a process, it is unfortunate that an assumption is made that
organizations are autopoietic. Despite also using Varela et al.’s (1974) checklist for
identifying autopoietic systems, the incorrect conclusion that organizations are
autopoietic is used. In light of these weaknesses, the living composition model
presented is still a good representation of organizations and how they might learn and
adapt to new situations. Perhaps, the model could be redesigned such that it
recognizes that only data and information can flow between different people/entities,
and that organizations are not first-order autopoietic entities.
It is also evident from the literature that research has not looked at applying the
scientific principles of autopoiesis to a preexisting model of organizational learning.
However, this approach falls into the trap of not starting from an autopoietic
definition of knowledge. The result of this process would be a list of criteria for
making an existing model of organizational learning, or even knowledge manage-
ment, autopoietic. The models to which autopoiesis had been applied would then
need testing to ensure the changes made a positive impact. No impact, or a negative
impact, would obviously require a profound restructuring of the research.
Jackson (2007) starts from the premise that current research within knowledge
management is lacking a foundation, and is filled with lots of disagreements.
However, after an introduction to autopoiesis, the research simply presents
comparisons between autopoiesis and different aspects of knowledge management
and organizational learning, arguing that autopoiesis in its entirety is too complicated
to be useful in an organizational setting. However, the resulting metaphorical analysis
finds that aspects of autopoiesis that used were far too simple to be applied to
organizational learning. Whilst this paper did follow the social constructivist
approach (Jackson, 2007), it failed to recognize that organizations could be viewed
as cognitive systems, or third-order autopoietic systems, instead focusing solely on
first-order autopoiesis. Jackson (2007) does realize that a problem exists in viewing
organizations as first-order autopoietic entities because truly autopoietic entities are
purposeless, and the same cannot be said for organizations. Viewing organizations as
third-order autopoietic entities would have removed the boundary problems
(Jackson, 2007) and allowed research to focus on how third-order autopoiesis can
create a consensual domain, and allow for languaging to occur, and ultimately
Autopoiesis as the Foundation for Knowledge Management 253
increasing knowledge sharing within the organization. With this in mind, it should be
possible to develop the concept of organizational learning to include concepts such as
embodied knowledge and enacted cognition.
interpretations may exist for the same reality. As such, ‘‘understanding becomes a
part of valid knowledge’’ (Cornford & Smithson, 1996).
It is apparent that positivism or interpretivism neither solely suitable for applying
autopoiesis to knowledge management. However, an integration of ideas from both
perspectives would be ideal, and this is possible using matching. Matching is a new
methodology developed by von Krogh, Roos, and Slocum (1996), and is used for the
integration of two or more theories. Often described as unifying languages and
relationships, matching is a two-step process: theoretical discourse and inscription.
Theoretical discourse is the frequent dialogue about the theories, from which a new
language emerges and through which the theories unite. Following on from which is
inscription, which can be defined as ‘‘the process of making and presenting knowledge
from the first stage, such that it can inform other theory building attempts’’ (von
Krogh et al., 1996).
Numerous models could have been chosen for this research, such as Buckler (1996),
Matthews (1999), or Örtenblad (2004), but Kim’s (1993) model of organizational
learning was selected (Figure 2). While Buckler’s (1996) model does contain elements
of feedback, it is not an inherently circular model, like that of Kim (1993). Buckler’s
(1996) emphasis is also on organizational learning, virtually ignoring the process of
individual learning. Matthews (1999), on the other hand, has undeniably created a
circular model, but again does not detail how individual learning occurs, or indeed
that it occurs in a separate cycle to organizational learning. Örtenblad (2004) uses the
slightly different perspective of the learning organization, and has a circular model,
but unlike Kim’s (1993) model, has two inputs to the cycle, which, in autopoietic
terms, is undesirable, since external influences cannot determine change that occurs
within an autopoietic entity.
Kim’s (1993) model of organizational learning starts on the employee level, and
defines individual learning as based on the experiential learning model. The cycle
starts with a concrete experience, on which an observation may or may not be made.
If an observation is made, then the individual will assess that observation (either
consciously or subconsciously) to create, or design, generalizations or abstractions of
that situation. Finally, the individual will test or implement the generalization in the
real world, hence creating another experience and starting the cycle again. From
individual learning, Kim (1993) adds the notions of single- and double-loop learning
to the model. The main distinction made by Kim (1993) is that only double-loop
learning uses memory, and single-loop learning links straight from the Observe–
Assess–Design–Implement (OADI) cycle to individual action. The link to memory by
double-loop learning infers the presence of mental models, which take to form of
frameworks and routines. Together, these frameworks and routines model a person’s
view of the world, in turn affecting any abstractions/generalizations they design and
implement.
Autopoiesis as the Foundation for Knowledge Management 255
Moving to the organizational level, Kim (1993) identifies shared mental models as
the key to access the organizational memory. Shared mental models can directly lead
to organizational action, but can also affect individual action through a person’s
individual mental models. These two leads for shared mental models are double-loop
learning cycles because, as identified earlier, any use of a memory store infers double-
loop learning occurring. The model goes on to show how individual and
organizational action occurs, with the resulting environmental response that is
observed by the original OADI cycle at the start of the model.
The matching process used to develop the model of knowledge in this paper took
place over the course of several meetings between a Ph.D. student and the supervisory
team. All potential terms to be used in the model were discussed and definitions of
words were explored to resolve any conflicts, for instance, whether the term
‘‘observation’’ was purely related to sight, or all senses. The concept of environmental
responses was also discussed, in light of the notion of perturbations in autopoietic
256 Paul Parboteeah et al.
entities. The issue of whether organizations could be autopoietic, and have a memory,
or even knowledge, was discussed. Applications of the model were also explored to
ensure terminology being used was not inherently restrictive. The second stage of the
process involved modification of the model. After the model was modified, it was
subject to numerous reviews prior to being finalized.
As a result of the matching process, four changes were made to the original model of
organizational learning. First, an extra ‘‘Observe’’ stage was added to the OADI loop,
between the ‘‘Implement’’ and original ‘‘Observe’’ stages. This extra ‘‘Observe’’ stage
also makes the OADI loop comparable to the cell wall–cell metabolism argument
presented by Maturana and Varela (1998). They propose that neither can come first in
the production of the cell, but rather they must develop together to ensure the
successful creation of a cell. The two ‘‘Observe’’ stages also recognize the different
types of observation that need to occur. The first is a general observation/problem
detection process, whilst the ‘‘Observation’’ stage after the ‘‘Implement’’ stage is more
a reflection process.
The second change was to change ‘‘Environmental Response’’ to ‘‘Observed
Environmental Response.’’ This is because the environment can never be part of the
system it contains (Maturana & Varela, 1998). If the model kept the ‘‘Environmental
Response,’’ then it would imply that the system is not distinct from the environment,
breaking one of the autopoietic principles developed by Varela et al. (1974).
The third change to the model was to rename ‘‘single-loop learning’’ as
‘‘autopoietic learning.’’ Single-loop learning can be defined as learning that involves
the detection and correction of error (Argyris & Schön, 1996), which essentially
means if an error occurs, the person changes their actions so the error does not
happen again. One feature of autopoietic entities is that ‘‘the environment only
triggers structural changes in the autopoietic unities (it does not specify or direct
them)’’ (Maturana & Varela, 1998), and this is akin to single-loop learning because
the consequences do not determine the change that will happen, it merely triggers it.
This kind of learning can be renamed autopoietic learning, defined as a random
change of behavior when current actions do not have the desired effect or outcome.
The key feature of autopoietic learning is that when an undesired outcome occurs, the
person undertakes an unconscious attempt to produce the desired effect. This is
characterized by a random change of behavior without any analysis of what went
wrong in the first instance.
The final change to the model was to rename ‘‘double-loop learning’’ as
‘‘allopoietic learning.’’ Double-loop learning is when the consequences of an action
cause the person to look back at their ‘‘governing variable’’ (Smith, 2001), or
‘‘individual mental models’’ (Kim, 1993), and determine frameworks or routines that
caused the consequences so that they can be changed. This is not similar to an
autopoietic entity because change is not only being triggered, but also determined.
Subsequently, if a machine is not autopoietic, it is an allopoietic machine which has,
Autopoiesis as the Foundation for Knowledge Management 257
It would appear that the abstract nature of the model could hinder developing a
suitable survey. Concepts such as mental models are by their very nature abstract, and
trying to identify and evaluate them could prove very difficult. It could be easier, and
more useful, to look at the effect of the presence of mental models. For instance,
instead of asking respondents about their own frameworks and routines, the
questions could ask about the respondents’ personal experience and how often they
use assumptions.
The model is also circular, which itself is suited to autopoiesis. Testing the circular
nature could mean the questions become self-checking. In other words, as questions
develop around one part of the model, by the time they move around the model and
end up in the same place, the questions should be testing what they were testing
originally. The circular nature of the model could also cause problems because, if
elements are related in a circular nature, it could make it difficult for questions to
make an entrance into the loop to start testing. However, an alternative approach to
viewing the model as circular is to view the model as numerous authors’ works
combined into one model. This would aid testing because each authors’ section could
be independently tested.
The very nature of organizational learning itself will undoubtedly have an impact
on the method used to test the model. As already identified, the abstract nature of
concepts involved means the survey will either have to make explicit what the
concepts mean, potentially reducing their meaning or leave the respondent to
interpret the concepts and risk their incorrect interpretation. A balance is clearly
required; the survey would interpret the more abstract and vague concepts, while
leaving the respondents to interpret the more well-known concepts.
258 Paul Parboteeah et al.
There is also a problem with the German word ‘‘Weltanschauung,’’ of which there
is no direct English translation. In the original model, ‘‘Weltanschauung’’ is
interpreted as the organization’s view of the world, or even just the organization’s
view point (Kim, 1993). A problem then arises if a native German answers the survey,
because they could understand the full extent of the term, and so provide an answer
radically different from other respondents. The best approach would therefore be to
put the interpretation of the word into the survey, and not allow the respondents to
interpret ‘‘Weltanschauung’’ for themselves.
Also when dealing with abstract concepts in surveys, it could be very easy to give
away intended answers in the question itself. It would be better for the respondents to
offer the answers themselves, which is why an interview, which allows respondents to
elaborate when they feel necessary, is probably better.
Autopoiesis as the Foundation for Knowledge Management 259
5 Conclusion
This chapter has shown that autopoiesis can be adapted for use as an underlying
paradigm for knowledge management in organizations. By using the principles
developed by Varela et al. (1974), and following the matching methodology
(von Krogh et al., 1996), a new model of organizational learning was developed.
Further work is necessary to test the model and consider whether the model could
more accurately reflect work scenarios. Of course, the model as it currently stands
could be tested in several organizations to determine if the findings this chapter
presents are reproducible.
The research presented in this chapter would seem to suggest that autopoiesis
could be used as a backdrop to knowledge management in organizations. In doing so,
autopoiesis also appears to introduce a multidisciplinary approach to knowledge
management by introducing biological thinking. Perhaps as an aside, it would be
interesting to introduce Luhmann’s (1995) concept of social autopoiesis to add a
social dimension to knowledge management. Either way, this chapter has provided an
insight into how autopoiesis can benefit knowledge management, and followed the
notion through to developing a model ready for testing.
References
Abou-Zeid, E. (2007). Towards a design theory of autopoietic knowledge management support
systems. In: M. Khosrow-Pour (Ed.), Managing worldwide operations and communications
with information technology (pp. 614–617). Vancouver, Canada: IGI Publishing.
Argyris, C., & Schön, D. (1996). Organizational learning II: Theory, method and practice.
Reading, MA: Addison Wesley.
Biggiero, L. (2007). Organizations as cognitive systems: What do they process and deliver? http://
Mpra.Ub.Uni-Muenchen.De/3137: Munich Personal Repee Archive.
Buckler, B. (1996). A learning process model to achieve continuous improvement and
innovation. The Learning Organization, 3(3), 31–39.
Cornford, T., & Smithson, S. (1996). Project research in information systems: A student’s guide.
Basingstoke: Macmillan Press.
Curado, C. (2006). Organizational learning and organizational design. The Learning
Organization, 13(1), 25–48.
Green, A. (2005). A framework of intangible valuation areas. In: M. Stankosky (Ed.), Creating
the discipline of knowledge management: The latest in university research (pp. 189–208).
Amsterdam: Elsevier Butterworth Heinemann.
Hall, W. P. (2005). Biological nature of knowledge in the learning organization. The Learning
Organization, 12(2), 169–188.
Hasan, H., & Gould, E. (2003). Activity-based knowledge management systems. Journal of
Information & Knowledge Management, 2(2), 107–115.
Holsapple, C., & Joshi, K. (1997). Knowledge management: A threefold framework. Kentucky
Initiative for Knowledge Management, Paper no. 104.
Jackson, T. W. (2007). Applying autopoiesis to knowledge management in organizations.
Journal of Knowledge Management, 11(3), 78–91.
260 Paul Parboteeah et al.
Varela, F. J., Maturana, H. R., & Uribe, R. (1974). Autopoiesis: The organization of living
systems, its characterization and a model. Biosystems, 5(4), 187–196.
von Krogh, G., Roos, J., & Slocum, K. (1996). An essay on corporate epistemology. In: G. von
Krogh & J. Roos (Eds), Managing knowledge: Perspectives on cooperation and competition.
London: Sage.
Wiig, K. (2002). New generation knowledge management: What may we expect? Arlington, TX:
Knowledge Research Institute.
Yeo, R. K. (2005). Revisiting the roots of learning organization: A synthesis of the learning
organization literature. The Learning Organization, 12(4), 368–382.
Chapter 15
1 Introduction
‘‘bullwhip effect’’ (Forrester, 1961; Lee, Padmanabhan, & Whang, 1997). The fact
that several living systems that feature suchlike characteristics have been examined by
using the concept of autopoiesis (Varela, 1979), as well as other applications like
economic systems (Zeleny, 1997) or even war as an autopoietic system (Matuszek,
2007), leads to the question if the idea of autopoiesis can offer some contributions to
an optimal design of logistics systems. Therefore, autopoiesis might, on the one hand,
function as a metaphor to describe logistics systems and its characteristics, which
though would include that there are insurmountable differences between them
(Drosdowski, Müller, Scholze-Stubenrecht, & Wermke, 1985). This would reflect an
insufficient perspective on the reality of possible prospective technologies and today’s
logistics research. According to Wycisk et al. (2008) (quod vide, Surana et al., 2005;
Choi, Dooley, & Rungtusanatham, 2001; Pathak, Day, Nair, Sawaya, & Kristal,
2007), logistics systems can be described as complex adaptive systems (CAS), which
include characteristics like the system’s self-creation and self-reference. This, in turn,
has been factored explicitly into the autopoietic theory (Maturana & Varela, 1980)
and applied to social systems by Luhmann (Luhmann, 1984; Krause, 2005).
Therefore, the questions arise, on the other hand, if characteristics of autopoietic
systems can be implemented in logistics systems, such as ISN, in order to cope with
increasing complexity and dynamics and what instruments are necessary to do so. The
theoretic instrument used in this paper, to show respective possibilities, is the concept
of autonomous cooperation, which basically states the ability of a system to react to
changes in the environment based on own decisions and means of the system’s single
components. The wider concept of autonomous cooperation is self-organization,
which, in turn, has its roots in different concepts, from which one is the autopoietic
theory. Therefore, it can be said that autonomous cooperation reflects the idea of
autopoiesis and self-organization (e.g., Maturana & Varela, 1980; Haken, 1973;
Prigogine, 1969) and allows a system to create ordered structures autonomously
(Manz & Sims, 1980). Therefore, the concept of autonomous cooperation will be
applied to CALS in order to examine its contributions to implement autopoietic
characteristics and therefore to increase the system’s ability to deal with complexity
and dynamics. Figure 1 illustrates the coherences between the autopoietic theory and
ISN, which will be examined in detail in the following sections.
The question this paper is going to answer is: Is the concept of autonomous
cooperation a reasonable way to implement autopoietic characteristics into logistics
systems in order to enable them respectively to increase their ability to deal with
complexity and dynamics in CALS like ISN? In order to examine this question, the
paper is going to proceed as follows. Section 2 represents the tendency from linear
supply chains to ISN. This will be used to reveal the increasing sensitivity of ISN due
to external events. Section 3 introduces the concept of CALS and connects it with the
properties of autopoietic systems. The underlying concept of CAS will be described to
examine its essential characteristics. One of these properties is autonomous
cooperation. Due to the assumption that autonomous cooperation is able to
contribute to logistics systems’ robustness, it will undergo a deeper examination in the
following section of this article. Based on these properties, the connection between
CAS and ISN will be observed. Afterwards, the concept of autopoietic systems will be
introduced and connected with the findings of the previous examination. In doing so,
the autopoiesis concept will be applied to logistical problems whereas the focus will be
Autonomous Cooperation 265
their direct business partners but as well indirectly to other logistic agents in other
supply systems, other countries, and other cultures (Tapscott, 1999; Hülsmann,
Scholz-Reiter, Freitag, Wycisk, & de Beer, 2006). This results from the simultaneous
integration of one agent in different supply networks and the agent’s multi-
dimensional interrelations in combination with the system’s immanent openness to
other systems in its environment (Tapscott, 1999). Therefore, the originally phrased
definition of supply chain management that integrates all of the company’s activities
and that is based on Porters value chain (Porter, 1980, 1999) has to be amplified by a
view that focuses on the networks lying behind and connecting these value chains with
others (Hülsmann & Grapp, 2005). Terming this a network instead of a chain is more
precise due to its more wide meaning that includes not only material flow and
information flow, but also their coordination between every single logistic actor in the
network (Hülsmann & Grapp, 2005). In other words, a tendency from linear supply
chains to ISN can be observed. However, Surana et al. came to the same conclusion
but integrate the two terms by defining a supply chain as ‘‘a complex network with an
overwhelming number of interactions and inter-dependencies among different
entities, processes and resources’’ (Surana et al., 2005, p. 4235).
In the course of globalization, the number of actors as system elements in logistics
systems increases. On the one hand, this leads to an increasing quantity of
information the information system of an ISN has to handle. Therefore, a necessity
for an increasing information-processing capacity can be observed in ISN (Hülsmann
et al., 2007).
On the other hand, the more actors are linked to each other, the more potential
problem sources exist. Being linked to many actors in a network means being
dependent on many actors, which are in turn accident sensitive to different external
events. Furthermore, this means that the system’s conditions change in increasingly
smaller intervals (Hülsmann & Berry, 2004). Therefore, new problems can occur from
different directions with which companies might have never had to deal before. This
leads to an increasing number of potential problems and therefore to a higher
sensitivity of the whole supply network. Resulting from this, the butterfly effect can
appear, which states that minor changes in a complex system can lead to completely
different conditions of the system in the future (Lorenz, 1972).
In consequence, the management of ISN is increasingly confronted with the
challenge to adapt to environmental changes.
3.1.1 Complex adaptive systems. Many kinds of systems, from natural to artificial,
can be characterized as complex, for example, ecologies, social systems, or
Autonomous Cooperation 267
communication networks (Surana et al., 2005). Thereby, the term complexity does
focus not only on the number of elements in the system, but particularly on the
quantity of relationships between the elements and between the elements and the
environment (Dörner, 2001; Malik, 2000). Due to that, the more elements exist and
the more these elements are linked to each other in any kind of relationship, the more
complex the regarded system is. A biological cell, for example, is a typical complex
system that consists of many proteins that send signals to each other, though they
have multiple relationships to other proteins in the cell (Holland, 2006).
Furthermore, systems are increasingly confronted with environmental changes,
which lead to increasingly dynamic circumstances in which the elements in a
system operate and in which the system is situated (Hicks, Gullett, Phillips, &
Slaughter, 1975; Hülsmann & Berry, 2004). Dynamic occurs when involved elements
change themselves or their relationships to other linked elements in the network
(Hülsmann et al., 2007). Therefore, dynamics can be described as ‘‘the rate
of modification of a system over a specific period of time’’ (Windt & Hülsmann,
2007, p. 35).
Due to the above-mentioned aspects, it can be assumed that a system’s ability to
adapt to environmental changes by its own means has an increasing relevance.
Furthermore, it can be assumed that the concept of CAS can provide this ability (e.g.,
Surana et al., 2005; Choi et al., 2001; Holland, 2002; Wycisk et al., 2008), which leads
to the necessity to examine these systems and their properties more precisely.
Its roots can be found, on the one hand, in biology where systems with living
entities were analyzed (Gell-Mann, 2002) and, on the other hand, in the theories of
complexity and chaos (Mason, 2007). These different approaches have in common
that they all deal with complex systems containing and constituted by a large number
of elements, which are linked to each other in a complex structure. (Mason, 2007;
Gell-Mann, 2002).
According to Wycisk et al. (2008), the following properties, mentioned before by
Kauffman (1993) and Holland (2002), are essential for CAS and its elements as well
as for their behavior: (1) Heterogeneity, (2) Interaction, (3) Autonomy, (4) Ability to
learn, (5) Melting Zone, (6) Self-organization, and (7) Coevolution. The elements
within a CAS in turn can be called agents which may represent, for example, an
individual, a team, or an organization (Choi et al., 2001; Surana et al., 2005; Holland,
2002).
First of all, the agents in a CAS are heterogeneous. This means that they distinguish
themselves from each other through different properties, functions, and rules
(Holland, 2002). In consequence, this results in differentiated behavior within the
system because every agent follows ‘‘individual goals under different constraints and
different action patterns’’ (Wycisk et al., 2008, p. 111).
The agents act as well autonomously to a certain degree since their actions are not
totally determined by other entities, for example, from a higher level in a hierarchy
(Kauffman, 1993; Holland, 2002). This results from the existence of the single agents’
own rules concerning their behavior (Mason, 2007); therefore, they are able to make
decisions autonomously without the need of any supervision; in other words, the
system is characterized by decentralized decision-making (Windt & Hülsmann, 2007;
268 Michael Hülsmann et al.
Probst, 1987). Decisions can be made decentrally if the system provides adequate
single elements with necessary resources for it (e.g., relevant information). In contrast
to totally hierarchical systems, elements in heterarchical structures have the
authorization to make decisions about different action-alternatives autonomously
without having to call an entity from an higher level in an hierarchical structure
(Windt & Hülsmann, 2007). That in turn means that the management of a system does
not have to absorb the complexity that accompanies every single decision situation in
the system. The internal and external complexity, which the management of a
system has to absorb, decreases because it can be distributed among its multiple
elements (Hülsmann & Grapp, 2005; Hülsmann & Wycisk, 2005). Furthermore,
autonomy implies the system’s ability to arrange its structure by own means (self-
formation), to supervise itself (self-control), and to develop without any impact
from the system’s environment (self-development) (Probst, 1987). However, these
mentioned characteristics lead as well to a nonpredictability about the system’s
future states, though it is nondeterministic (Flämig, 1998). This aspect follows the
characteristics autonomy — in logical consequence and therewith the decentralized
decision-making, which are the reasons for the nonpredictability of the single
elements’ behavior. This means that multiple possible future states of the system exist
(Haken, 1983).
Due to the agents’ heterogeneity, the agents are well equipped with heterogeneous
resources — for example, information — which motivates them to engage in exchanges
among each other. Therefore, CAS are characterized by interactions between
agents as the system’s elements (Holland, 2002, 2006; Wycisk et al., 2008). In order
to do so, agents have to communicate with each other, which in turn means that
agents react on other agents’ actions. In the course of a direct exchange of informa-
tion between them, they do not have to put up with a detour over a hierarchical
higher entity. In consequence, the necessary time for decision-making processes
abbreviates due to the elements’ possibilities to communicate directly with each
other (Hülsmann et al., 2008). Insofar this can lead to synergetic effects which
in turn can result in reaching a qualitative higher level of the whole system
(Haken, 1983).
Finally, a CAS is characterized by its agents’ ability to learn (Holland, 2002;
McKelvey, Wycisk, & Hülsmann, 2009). According to Holland, elements ‘‘y modify
their rules as experience accumulates, searching for improvements’’ (Holland, 2002,
p. 25). Thereby CAS achieve the possibility to react on environmental changes, in
other words the ability to adopt. In consequence, CAS can be regarded as intelligent
systems, whereas the intelligence ‘‘y may be located in its smart parts (y) and their
connectivity’’ (McKelvey et al., 2009, p. 7).
Besides the properties of CAS, there is a need to examine its behavior (Wycisk
et al., 2008). First of all, it has coevolutionary characteristics resulting from the agents’
autonomy — and interaction — properties as well as their ability to learn. On the one
hand, agents respond to other agents’ actions; on the other hand, the system is
capable of reacting to environmental changes and in turn can shape its environment
through actions or responses, which have influences on it (Kauffman, 1993; Choi
et al., 2001).
Autonomous Cooperation 269
(4) Self-reference. The terms complexity and autonomy were described above and
constitute essential characteristics of CAS. The existence of autonomously acting
agents, which interact with each other and whose behavior is not determined by other
entities, enables a self-organizing system to evolve by own means (Windt &
Hülsmann, 2007), which in turn leads to unpredictability of future system states
(Haken, 1987; Prigogine, 1996).
Redundancy means that there is no difference between organization and execution
in self-organizing systems (Probst, 1987). Furthermore, the elements are equipped
with similar or the same assets and abilities, which means that no functions exist in
the system that can be executed by just one element (Wycisk, 2006). Insofar, the
system comprises no elements with a permanent dominant impact on the system’s
development; in other words, the elements have similar degrees of influence on it
(Probst & Mercier, 1992). This aspect shapes the system’s heterarchical structure
(Hülsmann & Wycisk, 2006). This leads to an increasing flexibility of the system
because if one element is for any reason no longer able to execute its function, this
function can be taken over by another element. Beside this and as mentioned before,
the complexity the management of a system has to absorb decreases due to the
absence of an entity on an higher level in an hierarchical structure that supervises the
single elements’ functions (Hülsmann et al., 2008).
Whereas self-organizing systems are, on the one hand, open to absorb information
and resources, which enables them to adapt to environmental changes (Varela, 1979;
Malik, 2000), on the other hand, they are operationally closed (Windt & Hülsmann,
2007). This results from their self-reference, which describes the system’s ability to
build its own borders by own means. It offers the elements the information they need
to decide autonomously, which is the basis for their actions. Therefore, the system’s
behavior starts with the existence of this characteristic, as well as the ability to
measure environmental changes or to realize possibilities for internal synergies
(Probst, 1992). In consequence, self-reference enables the system to distinguish itself
from its environment (Luhmann, 1984).
3.1.3 Complex adaptive logistics systems. As shown before, a tendency from linear
supply chains to ISN can be observed (Surana et al., 2005, Hülsmann et al., 2007;
Mason, 2007). Several authors were arguing that ISN in turn can be regarded as CAS
272 Michael Hülsmann et al.
(e.g., Surana et al., 2005; Choi et al., 2001; Pathak et al., 2007; Wycisk et al., 2008),
which leads to the term CALS. According to Pathak et al., it is self-evident to identify
an ISN as a CAS because ‘‘organizations exhibit adaptivity and can exist in a complex
environment with a myriad relationships and interactions’’ (Pathak et al., 2007,
p. 550).
Complexity in ISN results from the ‘‘large amount of involved organizations and
relations between these organizations,’’ which can be circumscribed by the already
mentioned term hyper-linking (Hülsmann et al., 2007). To illustrate this complexity,
the example of a multinational textile company in Hong Kong will be regarded. In a
highly globalized world, the value chains of companies are not limited to one country,
not to mention to one organization. This company can have its customers,
its different retailers, as well as its different production locations and suppliers
for different parts of their clothes widespread over the world (Natarajan, 1999). This
leads to the system’s inherent complexity and requires, therefore, an ability to cope
with it.
Furthermore, increasing dynamics in ISN can be observed, which is a result of the
already mentioned phenomena hyper-linking (Tapscott, 1999), hyper-competition
(D’Aveni & Gunther, 1994), and hyper-turbulence (Monge, 1995). Dynamics emerge
when, for example, the contracts with retailers or with extern suppliers expire or
organizations linked to the exemplified textile company are not able anymore for any
reason to deliver resources. The textile company can as well search for better or
cheaper possibilities to get the needed parts like yarn or zippers or for better and
cheaper ways to deliver the clothes to the retailers as well as to search for new retailers
Autonomous Cooperation 273
and in new markets. However, changes in the ISN can arise from multiple different
directions, which reveals the inherent dynamic in ISN.
If organizations do not adapt to these kind of changes, they can lock-in into a
suboptimal situation in which they are no longer able to respond adequately to the
resources requests of their environment, which in turn can lead to negative effects
on the continuity of the organization (Schreyögg, Sydow, & Koch, 2003). In
consequence, this leads to the organizations’ necessity to adapt to environmental
changes, which means to adopt the ability to act flexible without losing stability,
though to reach a balance (Hülsmann et al., 2008).
Surana et al. (2005) call a supply network ‘‘[y] highly non-linear, [that] shows
complex multi-scale behavior, has a structure spanning several scales, and evolves and
self-organizes through a complex interplay of its structure and function’’ (Surana
et al., 2005). Therewith, the most essential CAS properties can be found in ISN. In the
course of describing these properties, Pathak et al. (2007) as well as Wycisk et al.
(2008) confirm the parallels to supply networks but formulated a few limitations to
them, for example, that CAS properties best fit to living systems whereas logistics
systems are only partly alive, for example, managers (Wycisk et al., 2008). But the
validity of this limitation is decreasing due to the current development of new
technologies that enable nonliving parts of systems to decide autonomously and
though to approximate to the acting rules of human parts in the system and become
a kind of intelligent. According to Scholz-Reiter, Windt, and Freitag (2004), a
paradigm shift in supply networks can be observed that leads from centralized
planning and control of nonintelligent entities to decentralized planning and control of
intelligent entities in the network. In consequence, with consideration of these
developments, ISN can be regarded as CALS (Wycisk et al., 2008; Pathak et al., 2007).
1980). Furthermore, the single elements are distinct; thus, they are heterogeneous, and
they are able to act without the need of confirming it by another entity, which means
that they are autonomous (Maturana & Varela, 1980) and that they act according to a
set of specific behavioral rules (Zeleny, 1997, 2001). The autonomy of the elements
within a system results from its circularity which means that there is no central
reference point in the system from which it develops. According to Flämig (1998), this
is a central aspect in the autopoiesis concept. Therewith, the system’s elements build
the network of production processes themselves; in other words, they constitute it and
simultaneously build its borders. The organization of such a system is called an
autopoietic organization, in other words, a system that organizes itself by its own
means (Maturana & Varela, 1980; Maturana, 1999).
Beside other authors that transferred this approach to new areas of application
(Flämig, 1998 and Luhmann (1984) uses it for certain questions in social sciences,
which have relevance for the examination of organizations. The recursive self-
production process is illustrated by communication between single actors in a closed
social system (e.g., an enterprise). Therewith, communication functions as the creator
of social systems. According to Luhmann (2006), communication does only exist as
social systems and in social systems. Due to the impossibility to divide communica-
tion, a smaller element does not exist in a social system. Luhmann (1990) describes a
chain effect in which every communication produces another following communica-
tion without the need of an external entity to trigger the following one; therewith this
can be called a self-organizing system.
system evolves, develops, and is being controlled by its own elements, just like a
biological cell that creates its own elements by own means (Flämig, 1998); in other
words, it is characterized by its recursive self-production.
These properties lead in logical consequence to the elements’ ability to learn and to
coevolution (Kauffman, 1993; Choi et al., 2001). When environmental changes occur,
the single elements have to change their rules to keep the system alive due to the
absence of an outside impact on the system. This means that the elements in an
autopoietic system as well as in a CALS react to the other elements’ actions and to
environmental changes. Changes of the elements lead in turn to changes in the
system’s structures, whereas some of these changes can have influences on the
environment of the autopoietic system (Kauffman, 1993; Choi et al., 2001). Finally
autopoietic systems must have in analogy to CALS a kind of melting zone due to a
system’s impossibility to be completely self-organized as well as being completely
organized by an external entity; rather it is situated in a continuum between these two
extremity pools (Wycisk, 2006).
In summary, it can be said that CALS exhibits the same essential properties as
autopoietic systems, which in consequence means that they can be regarded as one of
that ilk.
As shown above, CALS (e.g., ISN) can be regarded as autopoietic systems, whereas
these systems can comprise multiple organizations (e.g., production locations,
retailers, suppliers) with multiple elements (e.g., managers, nonliving items equipped
with technology that enables them to render decisions autonomously). Due to their
self-organizing property, which is an essential characteristic of both concepts, they are
characterized by their possibility to adopt autonomously cooperating processes
(Windt & Hülsmann, 2007).
As mentioned before, the environment of CALS is characterized by increasing
complexity and dynamics (Wycisk et al., 2008). These are in turn causes for external
risks (Hülsmann et al., 2007), which can be described as an impossibility to forecast
future developments. This impossibility in turn has its causes in a lack of information,
which would be necessary to decide under secure circumstances (Rosenkranz &
Missler-Behr, 2005). Caused by increasing complexity and dynamics, the informa-
tional basis for secure decision-making is deteriorating (Hülsmann et al., 2007). If the
system is no longer able to handle the incoming information, which means that it is
not able to make rational decisions, the system can become locked-in (Schreyögg
et al., 2003; Hülsmann & Wycisk, 2005). This leads to a necessity to enlarge the
capacity to handle the information the system is confronted with (Hülsmann et al.,
2007). The ability to cope with these external risks implies the system’s ability to
adapt to environmental changes that are in turn causative for the appearance of
external dynamics (Hülsmann et al., 2007).
276 Michael Hülsmann et al.
Autonomous cooperation has been discussed as a possible enabler for coping with
these challenges and therewith to increase the system’s robustness (Hülsmann et al.,
2006). Hence, the question arises whether the inherent characteristics of autonomous
cooperation can contribute to that. These possible contributions, shown in Figure 5,
are described in the following text and will be illustrated by the already mentioned
CALS example of a multinational textile company, which behaves like an autopoietic
system.
Decentralized decision-making: Due to the delegation of decisions to the single
elements of the system (e.g., production locations), the decision-making capacity of the
whole ISN increases (Hülsmann et al., 2007). In consequence, the logistics system’s
ability to handle information and therefore to adapt to environmental changes can be
enlarged. For example, if one location of the textile company stops its production, a
decision must be made which other location has to continue this work. If the
headquarter would have to ask first all locations if they are able to do the same work
and then to find the best alternative between all these locations, the decision-making
process could take longer than in a case where every location stays in direct contact to
the others and is able to render this decision independently. Therefore, the
headquarters can be disburdened from its necessity to handle all information about
the single system’s elements and the decision-making capacity of the whole ISN can
be enlarged.
Autonomy: If the elements can decide on their own, they are, as shown above,
autonomous (Windt & Hülsmann, 2007; Probst, 1987), which means that the single
elements are responsible for the system’s design in which they exist. For this reason,
the development of the system and therewith its direction is as well controlled by its
Autonomous Cooperation 277
single elements (Probst, 1987). This can lead to a superior system structure concerning
its ability to absorb complexity and to handle dynamics, in other words, to adapt to
environmental changes (Hülsmann et al., 2007). For example, if the production
locations can choose by own means what kind of clothes they produce and to which
retailers in which countries they deliver their products, without having to ask the
headquarter first, the headquarter would have less decisions to make and therefore
less complexity to absorb. Furthermore, if the elements in a logistics system are able
to decide, the production locations can as well decide on their own to open up another
one (e.g., to increase the production capacity) or to close one of them. Consequently,
the production- and delivery-structure of the whole ISN would evolve and develop as
well as if being controlled by its own production locations (Flämig, 1998; Probst,
1987). This can lead to a superior production structure (which location produces
which product and which should be downsized respectively enlarged), concerning its
ability to handle dynamics caused by environmental changes (e.g., changes in the
demand structure), compared to a structure that is totally controlled by a
headquarter. In this aspect, the resulting recursive self-production of the system
through its elements becomes apparent (Maturana & Varela, 1980).
Interaction: Because the elements in the autopoietic CALS are able to interact
respectively to communicate directly with each other (Holland, 2002, 2006; Wycisk
et al., 2008; Maturana & Varela, 1980; Luhmann, 2006), a more target-oriented
exchange of information can result. The elements exchange only their needed portion
of information, so that they need less capacity to handle them (Hülsmann et al.,
2007). Picking up the mentioned example, the single production location that stays in
direct contact with the other locations does not have to pick up a detour over the
headquarter to get some needed information. Instead, they can directly ask other
elements in the ISN (e.g., retailers or other production locations).
uncertainty about the question whether the system will be able to handle
environmental changes and enable it to react to changes in the system structure can
be reduced (Hülsmann et al., 2007). This implies an increasing ability to cope with risks
resulting from dynamics. In the mentioned example of a textile company,
nondeterminism would exist when the production locations, for example, were not
bounded to a certain production plan for a fixed time period. Changes in demands on
the market could not be taken into account if they were bounded. Therefore, this
nondeterminism enables the elements to change plans and the system to react to
environmental changes, which can occur as risks for the whole ISN.
4 Empirical Test
In the previous research, it has been identified theoretically that CALS can be seen as
autopoietic systems because the main characteristics of both are similar or even
identical. According to this, an empirical test of the implications that autonomous
cooperation has on the management of CALS could allow to draw conclusions about
autonomous cooperation in autopoietic systems as well. To measure the effects of
autonomous cooperation on CALS, a simulation and measurement system has been
developed that allows analyzing the effects of different autonomous cooperation
methods on the robustness of production networks with different levels of complexity
and different levels of external dynamics. In earlier work of the authors, the
simulation model has been used to analyze effects of autonomous cooperation on ISN
(Hülsmann et al., 2006, 2007). A similar approach will be used to analyze effects of
autonomous cooperation on CALS. Figure 6 shows the simulation model that has
been implemented as a discrete event simulation. The scenario shows a matrix-like
network of different production stages that are interlinked and that are able to
exchange information, resources, and orders to perform a multistage production
process. On one stage, the facilities are able to perform resembling production steps.
Each order has a specific processing plan, i.e., a list of processing steps that have to be
undertaken to produce goods. In the model, the orders are not directed by a
centralized control entity but have the ability to render decisions on their next
processing step autonomously by using different concepts of autonomous coopera-
tion. Depending on the different autonomous control methods, the overall system
shows altered behavior and dynamics.
This model comprises the opportunity to evaluate the system’s ability to cope with
different levels of complexity as well as different amounts of external dynamics. The
complexity can be varied by using different numbers of production facilities or
different kinds of orders and products. The orders enter the system at the sources.
Here, the external dynamics can be varied by using different functions that define the
arrival rate of different kinds of orders. By implementing different autonomous
control methods, the ability of autonomous cooperation to influence a system’s
Autonomous Cooperation 279
robustness can be analyzed. Therefore, the system’s performance will be measured for
different autonomous control methods with varying levels of complexity and external
dynamics.
In the following, the applied autonomous control methods will be described. The first
method, called queue length estimator (QUE), compares the current buffer level at all
parallel processing units that are able to perform the next production step. The buffer
content is not counted in number of parts but the parts are rated in estimated
processing time and the actual buffer levels are calculated as the sum of the estimated
processing time on the respective machine. When a part has to render the decision
about its next processing step, it compares the current buffer level, i.e., the estimated
waiting time until processing, and chooses the buffer with the shortest waiting time
(Scholz-Reiter, Freitag, de Beer, & Jagalski, 2005).
The pheromone method (PHE) does not use information about estimated waiting
time, i.e., information about future events, but uses data from past events. This
method is inspired by the behavior of foraging ants that leave a pheromone trail on
their way to the food. Following ants use the pheromone trail with the highest
280 Michael Hülsmann et al.
concentration of pheromone to find the shortest path to food. In the simulation this
behavior is imitated in a way that whenever a good leaves a processing unit, i.e., after
a processing step is accomplished, the good leaves information about the duration of
processing and waiting time at the respective processing unit. The following parts use
the data stored at the machine to render the decision about the next production step.
The parts compare the mean throughput times from parts of the same type and
choose the machine with the lowest mean duration of waiting and processing. The
amounts of data sets that are stored define the up-to-datedness of the information.
This number of data sets can be used to tune the pheromone method. The
replacement of older data sets resembles the evaporation of the pheromone in reality
(Scholz-Reiter, Freitag, de Beer, & Jagalski, 2006).
The due-date method (DUE) is a two-step method. When the parts leave a
processing unit they use the queue length estimator to choose the subsequent
processing unit with the lowest buffer level. The second step is performed by the
processing units. The due dates of the parts within the buffer are compared and the
part with the most urgent due date is chosen to be the next product to be processed
(Scholz-Reiter, Freitag, de Beer, & Jagalski, 2007).
The following simulation analyzes the overall system’s ability to cope with rising
structural complexity and rising external dynamics using different autonomous
control methods. At each source, the arrival rate is set as a periodically fluctuating
function. The logistical goal achievement is measured using the key figure throughput
time for different levels of complexity and different autonomous control methods.
Therefore, the simulation model is able to represent the main characteristics of
CALS. The agents within the model have different characteristics, for example,
different due dates or different production steps; therefore, they can be assumed to be
heterogeneous. Additionally, interaction between the system’s elements is implemen-
ted, due to the agents’ ability to communicate with each other (e.g., the different
orders communicate with the processing facilities; furthermore, the order agents
communicate with each other using the stigmergy concept, i.e., communication via
the environment by leaving information for following agents). As well the agents are
able to act autonomously since the orders are able to render their decisions concerning
the next processing step. Up to now the learning ability within the simulation model is
limited due to the fact that the agents are modeled relatively simple so that they
themselves do not have any learning abilities. In contradiction to that, the methods of
autonomous control enable the system to react flexibly to changes and to learn about
changes in the system’s structure; for example, if a machine breaks down the control
methods enable the agents to avoid this machine in the production plan and therefore
enable the overall system to learn how to react on unexpected changes. The melting
zone is part of the analytical aim of the simulation. As different grades of autonomous
control are compared to each other via different control methods, they can be seen as
different grades in a continuum between 100% decentralized decision-making and
100% centralized decision-making (Hülsmann & Grapp, 2006). Therefore, the
different methods of autonomous control might represent different locations in the
melting zone. Finally, self-organization is implemented into the simulation model by
applying the ability to the agents to organize their processing autonomously.
Autonomous Cooperation 281
Figure 7 shows the results, i.e., the mean throughput times for the three different
autonomous control methods in dependence of the system’s complexity. To the right
of the figure, the system’s complexity is increased by enlarging the amount of
processing units as well as the number of sources. Furthermore, the minimal
throughput time, which is rising with increasing complexity, is shown. It can be
observed that the curves for the due-date method and the queue length estimator
show almost the same results. The due-date method shows a slightly worse
performance because of sequence reordering, while the pheromone method shows
inferior goal achievement. The first two curves are almost parallel to the minimal
throughput time and can be fitted by linear functions, which are shown in the inset of
Figure 7. This means that a constant logistical goal achievement is gained during
rising complexity. The pheromone method shows an inferior behavior, which is
proved by the fact that the curve can be fitted by a second-degree polynomial. In this
scenario, the dynamic is too high and the boundary conditions change faster than the
pheromones are updated.
Therefore, the pheromone method is not able to adapt to changing conditions and
this effect seems to cause more problems the more complex the scenario gets. With
rising complexity of the model the pheromone shows declining performance, which is
caused by the fact that the pheromone method is not able to use the higher amount of
degrees of freedom during frequently changing boundary conditions.
In a second simulation, external dynamics is varied to determine the system’s
robustness, i.e., the system’s ability to cope with external dynamics without being
30
QUE
DUE
25 PHE
Min TPT
Simulation runs
20
TPT [h]
15
80
QUE
70 DUE
PHE
Simulation runs
60
50
TPT [h]
40
30
20
10
0.3 0.35 0.4 0.45
Mean Arrival rate [1/h]
Figure 8: Logistical goal achievement for different mean arrival rates and multiple
autonomous control methods.
unstable. In this simulation, the system is called unstable if one of the system’s
parameters increases without restraint. To determine this boundary of stability, the
mean arrival rate at all sources has been increased and the highest possible arrival rate
before the system starts to be unstable is measured. Figure 8 shows the results. The
queue length estimator shows the highest robustness. The model shows stable
behavior until a mean arrival rate of 0.43 parts per hour is reached. The other two
methods show unstable behavior at much lower workload. They begin to destabilize
at 0.35 respectively 0.36 parts per hour. This is caused by reordering in case of the
due-date method and the above-mentioned inertia of the pheromone method
respectively.
The arrows in Figure 8 highlight the interesting measurement points in this
simulation study from the dynamics perspective. The system shows altering phases of
worse and improved behavior although the external dynamic is continuously
enlarged. This is caused by the fact that the system shows different characteristics
of internal dynamics at the different parameter constellations, which cause different
performance rates. This strong interrelation between certain parameter constellations,
dynamics, and performance is typical of complex systems and especially those with
elements of autonomous cooperation. Those systems tend to show chaotic-like
dynamics including extreme events, and their behavior strongly depends on initial
conditions.
It has been shown that different autonomous cooperation methods, i.e., different
levels of autonomous cooperation cause different robustness and internal dynamics
which result in different performances. This has been shown for different levels of
complexity and external dynamics. The key finding of the simulation study is that
Autonomous Cooperation 283
The question this article is aiming to answer is whether the concept of autonomous
cooperation can be seen as a reasonable way to implement autopoietic characteristics
into logistics systems and therewith to increase their ability to deal with complexity
and dynamics.
For the management of organizations and information systems (e.g., in logistical
contexts), the analysis in this article was able to show that the quantity of information
which organizations and their information systems have to process increases. This is
due to the tendency from linear supply chains to ISN, which has been confirmed
during this analysis (Surana et al., 2005, Hülsmann et al., 2007; Mason, 2007).
Therefore, it has been outlined that the environment of actors within logistics
processes is characterized by increasing complexity and dynamics (Hülsmann &
Berry, 2004). This leads to one of its inherent problem sources: Due to a larger
number of involved actors and to a larger number of relationships between these
actors in ISN, it can be stated that they are more accident sensitive than linear supply
chains. In consequence, new challenges, concerning the ability of systems to cope with
increasing complexity and dynamics and therefore to handle an increasing quantity of
information, arise (Hülsmann et al., 2007).
Furthermore, it has been shown that the concept of autonomous cooperation
could be an appropriate instrument to implement autopoietic characteristics like self-
creation into logistics systems. This contributes to its robustness on each of its levels,
for example, decision system, execution system, as well as the information system
(Hülsmann & Grapp, 2006). It has been shown that ISN can be regarded as CAS as
well and therefore the possibility arises to use CALS and its underlying theories as a
framework to analyze their structures as well as their inherent complexity and
dynamics. CALS’ inherent self-organizing characteristic leads to apparent similarities
to the autopoiesis concept (e.g., Holland, 1995; Choi et al., 2001; Surana et al., 2005;
Maturana & Varela, 1980; Maturana, 1999). Therewith, autopoietic systems have
potentials for implementing autonomously cooperating processes and therefore
increase the information-processing capacity (Windt & Hülsmann, 2007). Hence,
regarding an ISN as a system with autopoietic behavior can contribute to enable the
organizations within an ISN and the information systems within the organizations to
handle the increasing quantity of information (Hülsmann et al., 2007) caused by
284 Michael Hülsmann et al.
increasing complexity and dynamics and therefore to keep its ability to make rational
decisions (Hülsmann & Wycisk, 2005).
The simulation model confirmed the theoretical assumption that autonomous
cooperation is a possible instrument to enable an organization, an information
system, or CALS as autopoietic systems to cope with the discussed challenges. By
examining different methods of autonomous cooperation, it has been pointed out that
different systems require different methods and different degrees of autonomous
cooperation to get the best resulting system performance. Therefore, in the context of
information systems, it has to be evaluated which autonomous cooperation methods
and which degree of autonomous cooperation have to be chosen to generate the
desired effects on the system’s ability to handle the increasing quantity of information
and therefore the highest robustness and, in summary, the best performance.
Further research requirements result, on the one hand, from the question which
degree of the single constitutive attributes of autonomous cooperation (e.g.,
autonomy, interaction, etc.), on which level, and which combination between the
degrees and the levels should be aspired, especially in autopoietic systems (e.g.,
CALS). On the other hand, it is still unclear which impacts the single degrees and the
single attributes have on each other. Beside this, problems in measuring the degree of
the single attributes of autonomous cooperation can be observed that have not been
under closing research up to now. Furthermore, the transfer of the autonomous
cooperation idea as well as the findings from research in autopoietic systems has not
been completely transferred into a practical logistics context yet (Hülsmann et al.,
2008).
Acknowledgment
This research was supported by the German Research Foundation (DFG) as part
of the Collaborative Research Centre 637 ‘‘Autonomous Cooperating Logistic
Processes — A Paradigm Shift and its Limitations.’’
References
Bea, F. X., & Göbel, E. (1999). Organisation, Theorie und Gestaltung. Stuttgart, Germany:
Lucius & Lucius.
Bick, H. (1973). Population dynamics of protozoa associated with the decay of organic
materials in fresh water. American Zoologist, 13(1), 149–160.
Choi, T. Y., Dooley, K. J., & Rungtusanatham, M. (2001). Supply networks and complex
adaptive systems: Control versus emergence. Journal of Operations Management, 19(3), 351–366.
D’Aveni, R., & Gunther, R. (1994). Hypercompetition: Managing the dynamics of strategic
maneuvering. New York: The Free Press.
Dörner, D. (2001). Die Logik des Misslingens (14th ed.). Hamburg, Germany: Rowohlt.
Drosdowski, G. (1989). Duden Herkunftswörterbuch dictionary (2nd ed.). Mannheim, Germany:
Dudenverlag.
Autonomous Cooperation 285
Drosdowski, G., Müller, W., Scholze-Stubenrecht, W., & Wermke, M. (Eds) (1985). Duden —
Bedeutungswörterbuch (Bd. 10, 2nd ed.). Mannheim: Dudenverlag
Flämig, M. (1998). Naturwissenschaftliche Weltbilder in Managementtheorien: Chaostheorie,
Selbstorganisation, Autopoiesis. Frankfurt am Main.
Forrester, J. W. (1961). Industrial dynamics. New York: Wiley.
Gell-Mann, M. (2002). What is complexity? In: A. Q. Curzio & M. Fortis (Eds), Complexity
and industrial clusters (pp. 13–24). Heidelberg: Physica-Verlag.
Grapp, J., Wycisk, C., Dursun, M., & Hülsmann, M. (2005). Ideengeschichtliche Entwicklung
der Selbstorganisation — Die Diffusion eines interdisziplinären Forschungskonzeptes. Reihe:
Forschungsbeiträge zum Strategischen Management, Bd. 8, Bremen.
Haken, H. (1973). Synergetics: cooperative phenomena in multi-component systems.
Symposium on Synergetics, 30 April to 6 May 1972, SchloX Elmau, Stuttgart.
Haken, H. (1983). Erfolgsgeheimnisse der Natur: Synergetik: die Lehre vom Zusammenwirken
(3rd ed.). Stuttgart: Dt. Verl.-Anst.
Haken, H. (1987). Die Selbstorganisation der Information in biologischen Systemen aus Sicht
der Synergetik. In: B.-O. Küppers (Ed.), Ordnung aus dem Chaos. Prinzipien der
Selbstorganisation und Evolution des Lebens (pp. 127–156). München.
Hicks, H. G., Gullett, C. R., Phillips, S. M., & Slaughter, W. S. (1975). Organizations: Theory
and behaviour. New York: McGraw-Hill.
Holland, J. H. (1995). Hidden order. Perseus Books: Cambridge, MA.
Holland, J. H. (2002). Complex adaptive systems and spontaneous emergence. In: A. Q. Curzio
& M. Fortis (Eds), Complexity and industrial clusters (pp. 25–34). Heidelberg: Physica-
Verlag.
Holland, J. H. (2006). Studying complex adaptive systems. Journal of Systems Science and
Complexity, 19(1), 1–8.
Hülsmann, M., & Berry, A. (2004). Strategic management dilemma — its necessity in a world of
diversity and change. SAM/IFSAM VII World Congress on Management in a World of
Diversity and Change, Göteborg, Sweden.
Hülsmann, M., Scholz-Reiter, B., Freitag, M., Wycisk, C., & de Beer, C. (2006). Autonomous
cooperation as a method to cope with complexity and dynamics? — A simulation based
analyses and measurement concept approach. In: Proceedings of the International Conference
on Complex Systems (ICCS 2006). Boston, MA, USA. Avaliable at: http://necsi.org/events/
iccs6/viewpaper.php?d ¼ 339 (8 pages).
Hülsmann, M., Scholz-Reiter, B., Austerschulte, L., de Beer, C., & Grapp, J. (2007).
Autonomous cooperation — A capable way to cope with external risks in international
supply networks? In: K. S. Pawar, C. S. Lalwani, J. C. de Carvalho & M. Muffatto (Eds),
Proceedings of the 12th International Symposium on Logistics (12th ISL) (pp. 243–249).
Loughborough, UK, 2005.
Hülsmann, M., Grapp, J., & Li, Y. (2008). Strategic adaptivity in global supply chains
competitive advantage by autonomous cooperation. International Journal of Production
Economics, Special Edition, 114(1), 14–26.
Hülsmann, M., & Grapp, J. (2005). Autonomous cooperation in International Supply
Networks — The need for a shift from centralized planning to decentralized decision making
in logistic processes. The 10th International Symposium on Logistics (10th ISL) —
Innovations in global supply chain networks. 3–5 July 2005. Lissabon, Portugal. In: K. S.
Pawar, C. S. Lalwani, J. C. de Carvalho & M. Muffatto (Eds), Conference Proceedings
(pp. 243–249). Loughborough, UK.
Hülsmann, M., & Grapp, J. (2006, July 9–11). Monitoring of autonomous co-operating logistic
processes in International Supply Networks. The 11th International Symposium on Logistics
286 Michael Hülsmann et al.
Natarajan, R. N. (1999). Logistics, strategy, and supply chain: Making the right connections
in the information age. In: M. Muffato& K. S. Pawar (Eds), Logistics in the information age,
Proceedings of 4th International Symposium on Logistics. Florence, Italy (11–14 July 1999).
Okino, N. (1993). Bionic manufacturing systems. In: J. Peklenik (Ed.), Flexible manufacturing
systems, past, present, future (pp. 73–95). Ljubljana, Slovenia: Faculty of Mechanical
Engineering.
Pathak, S. D., Day, J. M., Nair, A., Sawaya, W. J., & Kristal, M. M. (2007). Complexity and
adaptivity in supply networks: Building supply network theory using a complex adaptive
systems perspective. Decision Sciences, 38(4), 547–580.
Porter, M. E. (1980). Competitive strategy: Techniques for analyzing industries and competitors.
New York: Free Press.
Porter, M. E. (1999). Wettbewerbsvorteile-Spitzenleistungen erreichen und behaupten (5th ed.).
Frankfurt am Main: Campus-Verlag.
Prigogine, I. (1969). Structure, dissipation and life. In: M. Marois (Ed.), Theoretical physics and
biology (pp. 23–52). Amsterdam: North-Holland.
Prigogine, I. (1996). The end of certainty: Time, chaos, and the new laws of Nature. New York:
The Free Press.
Probst, G. J. B. (1987). Selbstorganisation: Ordnungsprozesse in sozialen Systemen aus
ganzheitlicher Sicht. Berlin: Parey.
Probst, G. J. B. (1992). Selbstorganisation. In: E. Frese (Ed.), Handwörterbuch der Organisation
(pp. 2255–2269). Stuttgart: Schäffer-Poeschel.
Probst, G. J. B., & Mercier, J.-Y. (1992). Organisation: Strukturen, Lenkungsinstrumente und
Entwicklungsperspektiven. Landsberg/Lech: Verl. Moderne Industrie.
Rosenkranz, F., & Missler-Behr, M. (2005). Unternehmensrisiken erkennen und managen.
Berlin: Springer.
Sanchez, R. (1993). Strategic flexibility, firm organization, and managerial work in dynamic
markets: A strategic options perspective. Advances in Strategic Management, 9, 251–291.
Scholz-Reiter, B., Windt, K., & Freitag, M. (2004). Autonomous logistic processes: new
demands and first approaches. In: L. Monostori (Ed.), Proceedings of the 37th CIRP
International Seminar on Manufacturing Systems (pp. 357–362). Budapest, Hungary.
Scholz-Reiter, B., Freitag, M., de Beer, C., & Jagalski, T. (2005). Modelling and Analysis of
Autonomous Shop Floor Control. Proceedings of 38th CIRP International Seminar on
Manufacturing Systems. Universidade Federal de Santa Catarina, Florianopolis, BR.
Scholz-Reiter, B., Freitag, M., de Beer, C., & Jagalski, T. (2006). Modelling and simulation of a
pheromon based shop floor control. In: P. Cunha & P. Maropoulos (Eds), Proceedings of the
3rd International CIRP Sponsored Conference on Digital Enterprise Technology-DET2006.
University of Setubal, Setubal.
Scholz-Reiter, B., Freitag, M., de Beer, C., & Jagalski, T. (2007). Analysing the dynamics
caused by autonomously controlled logistic objects. In: H. A. El Maraghy & M. F. Zaeh
(Eds), Proceedings of 2nd International Conference on Changeable, Agile, Reconfigurable and
Virtual Production (CARV 2007) (pp. 273–280). University of Windsor, Windsor.
Schreyögg, G., Sydow, J., & Koch, J. (2003). Organisatorische Pfade — Von der
Pfadabhängigkeit zur Pfadkreation? In: G. Schreyögg & J. Sydow (Eds), Strategische
Prozesse und Pfade, Managementforschung 13. Wiesbaden.
Surana, A., Kumara, S., Greaves, M., & Raghavan, U. N. (2005). Supply-chain networks:
A complex adaptive systems perspective. International Journal of Production Research,
43(20), 4235–4265.
Tapscott, D. (1999). Creating value in the network economy. Boston: Harvard Business School
Press.
288 Michael Hülsmann et al.
Ueda, K. (1993). A genetic approach toward future manufacturing systems. In: J. Peklenik
(Ed.), Flexible manufacturing systems, past, present, future (pp. 73–95). Ljubljana, Slovenia:
Faculty of Mechanical Engineering.
Varela, F. J. (1979). Principles of biological autonomy. New York: Elsevier North Holland.
von Foerster, H. (1960). On self-organizing systems and their environment. In: M. C. Yovits &
S. Cameron (Eds), Self organizing systems. London: Pergamon Press.
Windt, K., & Hülsmann, M. (2007). Changing paradigms in logistics — understanding the shift
from conventional control to autonomous cooperation and control. In: M. Hülsmann &
K. Windt (Eds), Understanding autonomous cooperation & control — the impact of autonomy
on management, information, communication, and material flow. Berlin: Springer.
Wycisk, C. (2006). Qualifikationsanforderungen der Selbstorganisation — Eine Profilentwick-
lung. In: M. Hülsmann (Ed.), Forschungsbeiträge zum strategischen management (Bd. 10).
Bremen, Germany.
Wycisk, C., McKelvey, B., & Hülsmann, M. (2008). Smart parts-supply networks as complex
adaptive systems: Analysis and implications. The International Journal of Physical
Distribution and Logistics Management (IJPDLM), 38(2).
Zeleny, M. (1997). Autopoiesis and self-sustainability in economic systems. Human Systems
Management, 16(4), 251–263.
Zeleny, M. (2001). Autopoiesis (self-production) in SME networks. Human Systems Manage-
ment, 20, 201–207.
Chapter 16
1 Introduction
At present, education often takes place in an organized setting. From the end of the
18th century onwards, the educational system has unmistakably become differen-
tiated — into the nonorganized family and the organized school or university. This
evolution is connected with the growing complexity of modern society and with
evolutions in other social subsystems, such as politics and the economy. The family
context normally creates numerous moments of casual education, but it can hardly
provide adequate support for lengthy and complex processes of learning. Formal
organizations are able to specify and preserve the criteria necessary to steer these
complex processes in the right direction. Accordingly, the introduction of compulsory
schooling — in Western Europe during the long 19th century, reaching from Prussia
(1764) to Belgium (1914) — has strengthened the role of organized education. How
has this fact, viz. that education now takes place in an organized setting, influenced
the nature of educational interaction?
I want to tackle this complex question with the help of a systems-theoretical
framework, inspired by the German sociologist Niklas Luhmann. Throughout his
whole academic career, Luhmann (1927–1998) has paid particular attention to the
systemic characteristics of organizations. In his voluminous Organisation und
Entscheidung [Organization and Decision], which was published posthumously in
2000, the basic concepts of his approach are comprehensively presented. This
publication provides us with a detailed overview of Luhmann’s theory of organized
social systems. The specification of these abstract concepts — e.g., with regard to the
study of firms, hospitals, schools, or universities — remains, however, largely
unknown territory (see Baecker, 1999; Seidl, 2005). Departing from Luhmann’s
writings on organizational theory, as well as from some of his shorter articles on
education, this chapter focuses on the analysis of educational interaction in organized
social systems.
Systems-theoretical analyses of social systems continue to be criticized for being
uncritical, for accepting the social status quo (instead of questioning its rationale).
Critics often refer to the notions of ‘‘equilibrium,’’ ‘‘homeostasis,’’ or ‘‘pattern
maintenance.’’ The application of the concept of ‘‘autopoiesis’’ or ‘‘self-reproduc-
tion’’ to the study of human behavior has not stopped, but even fuelled this line of
critique (e.g., Habermas, 1998; Blühdorn, 2000). In contrast to this common point of
view, the following analysis intends to illustrate the critical potential of systems
theory. I will suggest that the analysis of the autopoiesis of educational organizations
sheds light on the hidden mechanisms of this social system. To underpin my
argument, I will first highlight recent developments in the field of systems theory and
indicate their relevance for organization theory. In the second part of the chapter,
education in school organizations will be analyzed from this systems-theoretical point
of view.
1
For example, Ludwig von Bertalanffy, the leading author in the field of general system theory, wrote: ‘‘The
change of entropy in closed systems is always positive, order is continually destroyed. In open systems,
however, we have not only production of entropy due to irreversible processes, but also import of entropy
which may well be negative. This is the case in the living organism which imports complex molecules high in
free energy. Thus, living systems, maintaining themselves in a steady state, can avoid the increase of
entropy, and may even develop towards states of increased order and organization’’ (von Bertalanffy,
[1955]1988, p. 41).
292 Raf Vanderstraeten
As mentioned, the German sociologist Niklas Luhmann has thoroughly explored the
possibilities of an up-to-date systems theory of organizations. As part of his
2
The meaning of this term can be illustrated as follows for the domain of living organisms: ‘‘The cell y is a
complex production system, producing and synthesizing macromolecules of proteins, lipids, and enzymes,
among others; it consists of about 105 macromolecules on the average. The entire macromolecular
population of a given cell is renewed about 104 times during its lifetime. Throughout this staggering
turnover of matter, the cell maintains its distinctiveness, cohesiveness, and relative autonomy. It produces
myriads of components, yet it does not produce only something else — it produces itself. A cell maintains its
identity and distinctiveness during its lifespan. The maintenance of unity and wholeness, while the
components themselves are being continuously or periodically disassembled and rebuilt, created and
decimated, produced and consumed, is called ‘autopoiesis’’’ (Zeleny, 1981, pp. 4–5). For a discussion of the
relevance of this concept in theories of socialization and purposeful education, see Vanderstraeten (2000).
Autopoiesis of Decisions in School Organizations 293
ambitious attempt to draft a new ‘‘grand theory’’ for the social sciences, Luhmann
has characterized organizations (just as social systems in general) as autonomous
systems which produce their own operations and which distinguish themselves in this
process of self-production from their environment. Following Luhmann, organiza-
tions construct themselves by means of decisions. The specific meaning of these
decisions depends upon the decision context of the organizations themselves. New
decisions are connected with or built upon previous decisions. Organizations
continue, specify, correct or forget their own history. ‘‘Modern organizations tend
more and more to justify themselves by their own decision-making histories, in which
the values of the surrounding societal system are no longer treated as obvious y
Each step in the definition offers data for the next, thus constituting irreversible
history where every change follows because its only rationality lies in its relation to
the present state’’ (Luhmann, 1976, p. 102). Thus, if a selection committee nominates
someone for appointment, after having evaluated the different candidates for the
position, it is this decision which will structure the further course of decision-making.
Afterwards, this person can be appointed or not be appointed — but it has now
become a decision for or against this particular candidate. The organization cannot
appoint another candidate without declaring itself against the nominated candidate.
The initial decision sets the stage, although it does not determine subsequent decision-
making within the organization. Organizations have to consult their own history of
decision-making; they have to observe themselves. It is in this sense that they operate
in a self-referential way. They individualize themselves by means of their network of
decisions, by means of their form of self-organization (cf. Luhmann, 2000).3
Organized social systems become differentiated from their environment. They are
caught in a ‘‘structural drift’’ when new decisions orient themselves to previous ones.
This way, they incessantly create and recreate the difference between themselves and
their environment. In every organization, decisions refer to previous decisions and
enable new ones. Organizations (re)produce themselves. Organizations are, in this
particular sense, autopoietic systems. This implies that the environment is merely able
to ‘‘irritate’’ or ‘‘disturb’’ the system’s internal mode of decision-making. Luhmann
writes: ‘‘In the context of autopoietic reproduction, the environment functions as
irritation, as disturbance, as noise, and it only becomes meaningful for the system,
when it can be related to the networks of decisions of the system’’ (1992a, p. 173).
This point of view is based on classical definitions of information (when interpreted
carefully). After the Second World War, in the context of general systems theory and
cybernetics, Claude Shannon already indicated that ‘‘information’’ is not an inherent
3
Here, the concept of organization is used in its common (sociological) sense. This usage is different from
that of Humberto Maturana. Maturana distinguishes between the structure and the organization of
composite units (e.g., living organisms). Organization refers to lasting relationships in a composite unity;
structure, by contrast, is the particular instantiation that a composite unity enacts at a particular moment.
For example, when a human individual is born, she has one kind of structure; when she enters puberty, she
has another; if she contracts a disease, she has still another. But throughout her lifetime, her organization
remains the same: that which is characteristic of a living human. With regard to the study of dynamic social
systems, this distinction is in my opinion not very helpful (see also Mingers, 1995).
294 Raf Vanderstraeten
3 Organizing Education
In the preceding sections, it has been argued that organizations are social systems of a
particular kind. Organized social systems consist of decisions, and they reproduce
4
An uncomplicated example from W. R. Ashby’s An Introduction to Cybernetics might clarify this point:
‘‘Two soldiers are taken prisoner by two enemy countries A and B, one by each, and their two wives later
each receive the brief message ‘I am well.’ It is known, however, that country A allows the prisoner a choice
from ‘I am well - I am slightly ill - I am seriously ill,’ while country B allows only the message ‘I am well,’
meaning ‘I am alive.’ The two wives will certainly be aware that though each has received the same phrase,
the informations that they have received are by no means identical’’ (1964, p. 124). Following this
perspective, there is an immediate relation between information and selection. Thus, information can be
related to uncertainty and entropy. Information creates order; the processing of information counteracts the
natural evolution towards maximum entropy (Vanderstraeten, 2001b).
Autopoiesis of Decisions in School Organizations 295
these decisions in a network of such decisions. How does this type of systemic order
influence the activities which take place in these systems? And more particularly: How
does organization affect education? What are the characteristics of the autopoiesis of
educational organizations?
The order which develops within schools is an ersatz order, an organizational order.
Strict control of the effects of educational interventions on the ‘‘inner world’’ of
students is impossible (see Kupferberg, 1996). Instead, the system orients itself to
decisions which are made within the system. At school, it becomes important to be a
‘‘good’’ student (when questions have to be answered, proficiency tests have to be
made, or when something is undertaken on one’s own initiative). Presently, it should
not come as a surprise that this way of working generates particular side effects.
Unintended consequences already appear when the intention to educate comes to play
itself a role in the educational interaction. Thus, the person who is educated may
perceive the intention of her educator and hence gain the freedom to thwart this
Autopoiesis of Decisions in School Organizations 297
intention. She creates the opportunity to react to the intention as such and to look for
other possibilities. For example, she may behave seemingly obedient, try to do it her
own way, react to marginal events, or become a dedicated rebel (without noticing how
much she is after all a dedicated follower of fashion). The occurrence of this kind of
reactions cannot be eliminated. In view of the complexity of the situation, teachers
will hardly be able to get a firm grip on what takes place in the heads of their students.
It has been argued that these side effects can overtake the effects of the intentional
activities and profoundly impress the students. They might be more influential than
the carefully planned activities of teachers (cf., Vanderstraeten, 2000).
It is not just the pedagogical intention which may provoke positive or negative
reactions among students. To the contrary, the particularities of organized
educational interaction also mark students. Organized interaction in classrooms is
characterized by extreme expectations (in comparison with the expectations in other
spheres of social behavior). Think of the asymmetrical relation between the teacher
and her students, the enforced discipline, the imperative to sit still and preserve
silence, the limited tolerance for diverging opinions, the tyranny of schedules and
curricula. Students are mostly only able to face the teacher and not each other. The
classroom is isolated from the world; this world only enters in the prefigured form of
‘‘learning material.’’ A period is ended when the bell has rung and not when the
students’ attention slackens or when the teacher’s efforts have been successful (see
Hammersley, 1990, pp. 101–113; Dreeben, 2000). Hitherto, the effects of these
atypical structures have most of all been discussed in terms of the ‘‘hidden
curriculum.’’ Mainstream research has emphasized the reproduction of power
relationships in schools — its consequence being that progressive teachers have
become demoralized (see Burbules, 1993, pp. 131–142). From a contemporary
systems-theoretical point of view, however, it seems useful to reframe the question
about the socialization effects of organizations. Departing from the autonomy of
organizations, one gains a more solid basis for research about the effects of decision-
making structures. One might, for example, consider the hypothesis that particular
organizational structures (such as the schemes of good/wrong or succeed/fail, as well
as arrangements about entrance regulation or graduation) impose themselves in such
striking ways that they dominate the entire socialization process in schools and bread
off other decision-making structures.
Unintended side effects are not uniquely characteristic of educational organiza-
tions. But schools appear to be particularly vulnerable. Even in comparison with
other ‘‘people processing’’ organizations, they find themselves in a difficult position.
In the ‘‘people processing’’ organizations, in which professional help is offered, clients
often participate as a consequence of biographic crises and do not need to be urged on
to cooperate. They are longing for help and often ready to pay a lot of money for it
(Vanderstraeten, 1999). Because children have to go to school (at least until the
compulsory school age), this source of commitment and dedication mostly fails.
In fact, one finds numerous forms of opting-out behavior in educational institutions.
Students constantly observe the activities of their teacher. Observing whether one is
being observed, or is temporarily out of the teacher’s sight, pretending that one listens
attentively, hiding behind the back of a classmate, looking as if all the subject material
298 Raf Vanderstraeten
4 Conclusion
The foregoing discussion has provided a global overview of the effects of the
organized school setting on educational interaction. To conclude this paper, some
further remarks can be made which might direct future research into new directions.
(a) It has been convincingly demonstrated that the relation between individuals and
society currently takes the form of a lifelong career (e.g., Beck, 1986). Integration
within society is less dependent upon an individual’s natural characteristics, as it
is upon particular accomplishments in the course of her career. The modern world
is a world of self-management. Moreover, career planning highly relies on
educational organizations such as schools and universities (Kurtz, 2001, 2005).
Education contributes to career development — not that much via activities
which change the ‘‘inner world’’ of students (but cannot prove this influence), but
via its selection mechanisms and the degrees and certificates which it grants. What
Autopoiesis of Decisions in School Organizations 299
does or does not happen at the beginning of one’s career can have far-reaching
results. What has been achieved, mostly effects what becomes possible.
Expectations are formulated on the basis of the trajectory which has been
traveled. In the educational system, there are not many alternatives available.
School careers appear as an almost automatic succession of sequences (as long as
everything is ‘‘normal’’). Students pass from course to course, from year to year,
from degree to degree. The standardization of school careers does not only
facilitate but also provoke comparisons. It generates a high pressure to perform in
accordance with the organization’s criteria. School success has become an
important item for students and their parents. Its impact on individual careers
might explain the rapid educational expansion of the last decades of the 20th
century (see Collins, 1979; Vanderstraeten, 1999, 2000).
(b) Educational interventions automatically elicit different forms of selection.
Attempts at ‘‘people processing’’ and educating inevitably produce the distinction
between good and wrong, between conform and deviant behavior. Educational
organizations extend and amplify these forms of selection. It is this situation
which seems to inspire the numerous plans and projects for innovation which
constantly call into question the habits established within the educational world.
Apparently, the educational establishment experiences this selection as a
Fremdkörper, as a function that society imposes upon school organizations and
that goes to the detriment of authentic education. Educators incessantly stress the
fundamental equality of all people; they reject selection mechanism which only
focus on performance criteria and distinguish students on the basis of ‘‘super-
ficial’’ characteristics such as test results. In other words, educational ideals
continue to clash with the logic of school education. But why does the field of
education identify itself with reform, with ‘‘reforming again, again, and again’’
(Cuban, 1990)? For a reorganization cannot put the organizational logic of
schools out of action. That is one lesson to be learnt from the failures of reform
efforts in school education. Therefore, one might presuppose that the pedagogical
rhetoric serves some latent functions, such as the legitimization of the educational
establishment itself.
(c) From the end of the 18th century, an educational establishment has emerged in
several countries in Europe. It started, for example, with the construction of
several ‘‘showcase’’ schools — such as those founded at the end of the
Enlightenment by so-called philanthropists. In the same context, it was also
argued that teachers themselves could not be carriers of a careful reflection on the
conditions of education, and certainly not of a systematic reform movement. This
establishment has expanded in the 19th and 20th century. It has left its mark on
the evolution of the educational system. This group not only concentrates on the
tasks and problems of teachers, but also develops its own dynamics. It has an
interest in legitimizing its own existence. Instructional problems may not
disappear because they get solved. Moreover, the activities of this group increase
the contingency with which teachers are confronted. Whatever, for example, the
phrase ‘‘key qualifications’’ may express and whatever it may encompass: what it
includes is declared as something that can be decided. ‘‘Decision’’ also means
300 Raf Vanderstraeten
These remarks further illustrate the critical potential of systems theory. They
indicate its potential for exploring the basic mechanisms which structure the field of
organized education. Seen against this background, the widespread reproaches about
the conservatism of social systems theory are totally undeserved. But this does not
mean that systems theory can promise and produce a better world. The preceding
analyses do not provide a straightforward path to reform proposals, which stress the
value of particular ideals. Critique is useful, when the expectations from which it
departs are themselves controlled by research outcomes and when its ideals can be
realized in our social world.
The analyses in this chapter have departed from Luhmann’s idea that social
systems can be described as specific types of autopoietic systems. Like in the case of
living systems, autopoiesis or self-referential closure here does not mean that a social
system is not be affected by its environment. But it does mean that a social system can
react to its environment only in accordance with its own mode of operation. A social
system, such as an organization, is closed with respect to the meaningful content of
communicative acts; meaning here can be actualized only by circulation in the
network of ongoing decisions. In comparison with open-systems theory, this
autopoietic view of social systems provokes remarkable shifts — for example, from
an interest in planning and control to an interest in autonomy and environmental
sensitivity, and from structural stability to dynamic stability. As indicated in my
preceding remarks and examples, these shifts make it necessary to analyze the
mechanisms that are used to establish and maintain boundaries between system and
environment.
References
Aldrich, H. E. (1979). Organizations and environments. Englewood Cliffs, NJ: Prentice-Hall.
Arminen, I. (2005). Institutional interaction. Aldershot, UK: Ashgate.
Ashby, W. R. (1964). An introduction to cybernetics. London, UK: Chapman and Hall.
Baecker, D. (1999). Organisation als System. Frankfurt am Main, Germany: Suhrkamp.
Baecker, D. (2001). Why systems?. Theory, Culture and Society, 18, 59–74.
Bateson, G. (1972). Steps to an ecology of mind. New York: Ballantine.
Beck, U. (1986). Risikogesellschaft: auf dem Weg in eine andere Moderne. Frankfurt am Main,
Germany: Suhrkamp.
Autopoiesis of Decisions in School Organizations 301
Blühdorn, I. (2000). An offer one might prefer to refuse: The systems theoretical legacy of
Niklas Luhmann. European Journal of Social Theory, 3, 339–354.
Bourdieu, P. (1986). The forms of capital. In: J. G. Richardson (Ed.), Handbook of theory and
research for the sociology of education. New York, NY: Greenwood Press.
Buckley, W. (1967). Sociology and modern systems theory. Englewood Cliffs, NJ:
Prentice-Hall.
Buckley, W. (1998). Society: A complex adaptive system. New York, NY: Gordon and Breach.
Burbules, N. C. (1993). Dialogue in teaching: Theory and practice. New York, NY: Teachers
College Press.
Burns, T., & Stalker, G. M. (1961). The management of innovation. London, UK: Tavistock.
Collins, R. (1979). The credential society: An historical sociology of education and stratification.
New York, NY: Academic Press.
Cuban, L. (1990). Reforming again, again, and again. Educational Researcher, 19(1), 3–13.
Dreeben, R. (2000). Structural effects in education. In: M. T. Hallinan (Ed.), Handbook of the
sociology of education (pp. 107–135). New York, NY: Kluwer-Plenum.
Emery, F. E., & Trist, E. L. (1972). Towards a social ecology: Contextual appreciation of the
future in the present. London, UK: Plenum.
Farrell, J. P. (2000). Why is educational reform so difficult? Similar descriptions, different
prescriptions, failed explanations. Curriculum Inquiry, 30, 83–103.
Filer, A., & Pollard, A. (2000). The social world of pupil assessment. London, UK: Continuum
Books.
Habermas, J. (1998). The inclusion of the other. Cambridge, MA: MIT Press.
Hammersley, M. (1990). Classroom ethnography. Milton Keynes, UK: Open University Press.
Illich, I. D. (1983). Deschooling society. New York, NY: Harper and Row.
Jackson, Ph. W. (1990). Life in classrooms. New York, NY: Teachers College Press.
Kalthoff, H., & Kelle, H. (2000). Pragmatik schulischer Ordnung. Zur Bedeutung von ‘Regeln’
im Schulalltag. Zeitschrift für Pädagogik, 46, 691–710.
Katz, D., & Kahn, R. L. (1966). The social psychology of organizations. New York, NY: Wiley.
Katz, D., & Kahn, R. L. (1978). The social psychology of organizations (2nd ed.). New York,
NY: Wiley.
Kupferberg, F. (1996). The reality of teaching: Bringing disorder back into social theory and
the sociology of education. British Journal of Sociology of Education, 17, 227–247.
Kurtz, T. (2001). Form, strukturelle Kopplung und Gesellschaft. Systemtheoretische
Anmerkungen zu einer Soziologie des Berufs. Zeitschrift für Soziologie, 30, 135–156.
Kurtz, T. (2005). Die Berufsform der Gesellschaft. Weilerswist, Germany: Velbrück.
Lawrence, P. R., & Lorsch, J. W. (1967). Organization and environment: Managing
differentiation and integration. Boston, MA: Harvard Business School Press.
Luhmann, N. (1976). A general theory of organized social systems. In: G. Hofstede &
M. S. Kassem (Eds), European contributions to organization theory (pp. 96–113). Assen, The
Netherlands: Van Gorcum.
Luhmann, N. (1990). Die Homogenisierung des Anfangs: Zur Ausdifferenzierung der
Schulerziehung. In: N. Luhmann & K. E. Schorr (Eds), Zwischen Anfang und Ende. Fragen
an die Pädagogik (pp. 73–111). Frankfurt am Main, Germany: Suhrkamp.
Luhmann, N. (1992a). Organisation. In: W. Küpper & G. Ortmann (Eds), Mikropolitik.
Rationalität, Macht und Spiele in Organisationen, 2. Auflage (pp. 165–185). Opladen,
Germany: Westdeutscher Verlag.
Luhmann, N. (1992b). System und Absicht der Erziehung. In: N. Luhmann & K. E. Schorr
(Eds), Zwischen Absicht und Person. Fragen an die Pädagogik (pp. 102–124). Frankfurt am
Main, Germany: Suhrkamp.
302 Raf Vanderstraeten
Luhmann, N. (1995). Social systems ((Original print in 1984)). Palo Alto, CA: Stanford
University Press.
Luhmann, N. (1997). Die Gesellschaft der Gesellschaft. Frankfurt am Main, Germany:
Suhrkamp.
Luhmann, N. (2000). Organisation und Entscheidung. Opladen, Germany: Westdeutscher
Verlag.
Maclure, M., & Walker, B. M. (2000). Disenchanted evenings: The social organization of talk
in parent-teacher consultations in UK secondary schools. British Journal of Sociology of
Education, 21, 5–25.
Margolis, E. (2001). The hidden curriculum in higher education. New York, NY: Routledge.
Maturana, H. R. (1981). Autopoiesis. In: M. Zeleny (Ed.), Autopoiesis. A theory of living
organization (pp. 21–33). New York, NY: North Holland.
Maturana, H. R., & Varela, F. J. (1980). Autopoiesis and cognition. The realization of the living.
Dordrecht, The Netherlands: Reidel.
Mingers, J. (1995). Self-producing systems: Implications and applications of autopoiesis.
New York, NY: Plenum.
Seidl, D. (2005). Organizational identity and self-transformation: An autopoietic perspective.
Burlington, VT: Ashgate.
Shannon, C. E., & Weaver, W. (1949). The mathematical theory of communication. Urbana, IL:
University of Illinois Press.
Stern, R. N., & Barley, S. R. (1996). Organizations and social systems: Organization theory’s
neglected mandate. Administrative Science Quarterly, 41, 146–162.
Stichweh, R. (1997). Professions in modern society. International Review of Sociology, 7,
95–102.
Thompson, J. D. (1967). Organizations in action. Social science bases of administrative theory.
New York, NY: McGraw-Hill.
Vanderstraeten, R. (1999). Educational expansion in Belgium: A sociological analysis using
systems theory. Journal of Education Policy, 14, 507–522.
Vanderstraeten, R. (2000). Autopoiesis and socialization. On Luhmann’s reconceptualization
of communication and socialization. British Journal of Sociology, 51, 581–598.
Vanderstraeten, R. (2001a). The school class as an interaction order. British Journal of
Sociology of Education, 22, 267–277.
Vanderstraeten, R. (2001b). Observing systems: A cybernetic perspective on system/
environment relations. Journal for the Theory of Social Behavior, 31, 297–311.
Vanderstraeten, R. (2005). System and environment: Notes on the autopoiesis of modern
society. Systems Research and Behavioral Science, 22, 471–481.
Vanderstraeten, R. (2007). Professions in organizations, professional work in education. British
Journal of Sociology of Education, 28, 621–635.
Varela, F. J. (1979). Principles of biological autonomy. New York, NY: North Holland.
Von Bertalanffy, L. (1988). General system theory: Foundations, development, applications
((Original print in 1955)). New York, NY: Braziller.
Winograd, T., & Flores, F. (1986). Understanding computers and cognition. Norwood, NJ:
Ablex.
Woods, P. (1990). The happiest days? How pupils cope with school. London, UK: Falmer.
Zeleny, M. (1981). What is autopoiesis?. In: M. Zeleny (Ed.), Autopoiesis. A theory of living
organization (pp. 4–17). New York, NY: North Holland.