Beth Dempster, School of Planning, University of Waterloo,
Waterloo, Ontario, Canada, N2L 3G1
mbldemps@fes.uwaterloo.ca
January 1999

Abstract

The intent of this paper is to consider the notion of post-normal science from a new systems perspective that is based on a distinction between autopoietic and sympoietic systems. These "self"-producing systems - with and without self-defined boundaries, respectively - have significantly different characteristics that make their comparison useful. Autopoietic systems are closed, centrally controlled, development oriented, and efficient, whereas sympoietic systems are open-but-not-too-open, evolutionary, and adaptive. When applied to science, the system descriptions have heuristic potential. The characteristics of autopoietic systems reflect Kuhn's description of "normal science," those of sympoietic systems reflect Funtowicz and Ravetz's description of post-normal science. Such comparison enables recognition of the different (and situational) advantages and disadvantages of the two modes of inquiry. The system descriptions provide an opportunity for considering the suitability of epistemological and methodological paradigms and raises questions regarding how to cope with, and encourage, change. By recognizing the sympoietic character of the many environmental-social challenges we currently face, inappropriate aspects of conventional science can be illustrated. In some situations, we need to develop approaches that carry more sympoietic characteristics. Yet the evolutionary, self-organizing tendencies in such complex systems and the interdependencies between the two system types present significant challenges.

"The world is considerably less tidy than we thought."^*1 This recognition - expressed here by an ecologist - is a growing concern among many scientists who have used conventional approaches of investigation to comprehend the complexity of the world. Reductionist methodologies, hard-systems analyses, and the scientific method have provided useful, but limited, understanding. Citing a variety of concerns, in particular the value laden context of science and the uncertainties inherent in complex systems, Funtowicz and Ravetz^*2 argue for a need to reframe the notion of 'science' toward a more extensive and inclusive process. Calling the notion post-normal science, they describe it as complementary to, but different from, conventional science. Drawing distinctions between applied science, professional consultancy, and post-normal science, Funtowicz and Ravetz illustrate that these three approaches are appropriate in different situations. In particular, post-normal science is relevant in situations involving high risks, uncertainty, and divergent values. Such situations require that we learn to extend the peer community, to consider extended 'facts,' and to address issues of quality in the development and application of scientific knowledge.

The intent of this paper is to consider the notion of post-normal science from a new systems perspective that is based on a distinction between autopoietic^*3 and sympoietic^*4 systems. These two types of self-producing system - with and without self-defined boundaries, respectively - carry significantly different characteristics. We have tended to view systems from the autopoietic perspective. The distinction, and especially the description of sympoietic systems, leads to considerations that are quite different from those arising from many traditional system conceptions. Understanding the difference between autopoietic and sympoietic systems is useful for understanding the advantages, characteristics, and interdependence of 'normal' and 'post-normal' science, and for considering their appropriate application. In particular, application of the systems concepts to science leads to two questions.

1. Given their characteristics, how can we investigate situations that carry sympoietic characteristics?
2. Given some of the advantages of sympoietic systems, how can we make science more sympoietic?

Coupled with other considerations relevant to science, the characterizations suggest potential directions to pursue in order to encourage post-normal science. In return, application of the poietic systems heuristic helps explain the system characterizations and illustrates the value of making a distinction between autopoietic and sympoietic systems.

This paper covers many key issues that are open to debate. For the purpose of discussion, I assume particular positions.

The underlying motivation for my work is a concern for sustainability. I interpret this notion broadly to mean continuation of natural conditions conducive to human persistence at some level above pure subsistence.^*5 I believe that current interactions between natural (biophysical) and social systems jeopardize such conditions. The ideas presented in this paper target toward the normative objective of coping with the difficulties we face in attempting to achieve sustainability.

The arguments presented here must be placed within the cultural context of 'western' science - a Euro-American cultural perspective. The tendencies I describe, including reference to 'our' application of the autopoietic lens, or that 'we' must learn to expand the peer community, refers to people embedded within this culture.

I perceive a conflict between two statements made by Funtowicz and Ravetz. 1) Post-normal science is compatible with science. 2) Post-normal science is based upon a new epistemology. The epistemology they outline conflicts with the positivist approach assumed crucial to many scientists and hence is not compatible with conventional science. Although I refer to this issue briefly below, addressing it adequately requires delving into the definition of science - a discussion beyond the scope of this paper.

Although I tend to agree with the underlying premises put forward by Funtowicz and Ravetz, I question use of the term 'post-normal science' on two counts. 1) Is the descriptor 'post-normal' appropriate, since 'post' so often connotes 'beyond' in the temporal sense? 2) Is the approach that Funtowicz and Ravetz describe still 'science'? For the purpose of this paper I accept the name to express what is described, leaving further discussion to future papers.

I rely heavily on Kuhn's interpretation of the progress of science, yet I acknowledge that it has not received universal acceptance. Many critics are dissatisfied with his implication that science is irrational and subjective.^*6 Although I believe the concepts introduced in this paper provide relevant explanations for addressing these criticisms, I do not focus my discussion on such a critique.

I make a caveat regarding the systems concepts discussed in this paper. The intent here is to provide a general introduction to a new system concept - sympoiesis - that has heuristic potential relevant to identifying and describing system characteristics that are typically ignored. These concepts and characteristics are more completely described elsewhere.^*7

A major difficulty that arises when discussing transdisciplinary issues is the definition and interpretation of terminology. For example, my use of 'boundary' more closely reflects the common definition of 'barrier' rather than the systems-theoretic reference to boundary conditions. My use of 'information' is more loosely defined than in some disciplines. Recognition of such problematic definitional issues is important for understanding the concepts as discussed here.

As heuristics, systems enable understanding of the world by describing patterns, relations, and behaviours. The variety of system concepts that exist enable different conceptualizations, ranging from the mechanical, control oriented hard systems view to recognition of chaos and complexity. In particular, 'new systems thinking,' which incorporates recent developments regarding complex self-organizing systems, provides new possibilities for understanding. Moving beyond hard systems thinking by describing such new heuristics enables the recognition of different patterns, relations, and behaviours. The new possibilities are especially relevant for conceptualizing the large complex natural and social systems of concern in the complex, uncertain, conflict ridden situations we currently face.

The system concepts introduced here arise from a concern that delineating boundaries - the typical approach for identifying systems - is often inappropriate. We have had a tendency to view systems as bound, homeostatic, predictable, controllable entities - reflecting application of the autopoietic heuristic. Such characterizations fail in many situations. The concept of sympoiesis emphasizes a need to relinquish boundary definition and to focus instead on linkages, relations and self-organizing factors. In addition, the distinction between autopoiesis and sympoiesis draws attention to system characteristics that have often been neglected and draws causal linkages among sets of characteristics.

It is important to emphasize two key points relevant to system concepts. First, the following descriptions represent abstract system characterizations. They are human constructs designed to make the world easier to understand. What we 'see' will depend on the heuristic we use. The two system types described (autopoietic and sympoietic systems) are caricatures, in a sense - extremes intended to emphasize two different sets of characteristics. Any 'real' system is unlikely to match the caricatures, but rather to carry a mixture of the characteristics described by the abstractions. Second, by emphasizing a need to relinquish boundaries I do not intend to argue against their delineation as useful conceptual devices in appropriate circumstances. Boundaries are also heuristics. However, I believe that they are often interpreted as barriers, and as such exclude from consideration many influences that are critical for understanding complex systems.

Autopoietic systems are self-producing systems with self-defined boundaries.^*9 The cell is a primary example. Fundamental characteristics include their autonomous and unitary nature, the ability to "regenerate and realize the network of processes (relations) that produced them," and the ability to "constitute and specify" boundaries through "preferential interactions within the network."^*10 As I use the term with respect to poietic systems, boundaries must be understood as barriers formed by systems to restrict inputs and outputs according to system specifications and control.

Although autopoiesis was initially introduced in biology to address the characterization of living systems, applications are commonly found in many other disciplines including sociology, law, and family therapy.^*11 The interpretation of autopoiesis described in this paper primarily results from making a distinction between autopoietic and sympoietic systems. In consequence, although based on the literature, my interpretation will differ in some respects. Key sources contributing to my understanding of autopoiesis include works by Maturana, Varela, Bednarz, Fleischaker, Mingers and Capra.^*12 Although some authors question the validity of applying autopoiesis to social systems, I concur that such extension is valid - and useful.^*13 However, following Bednarz,^*14 I note that it is essential to define social systems in the social domain. For example, social system components are cognitive images of physical entities rather than the physical entities themselves, or socially defined roles, rather than the biophysical individuals holding these roles.^*15 (Although the interdependence of biophysical and social domains must be recognized.)

Many complex systems, both biophysical and social, do not match autopoietic characteristics. Although the notion of self-production is useful, the emphasis on self-produced boundaries is inappropriate in many circumstances (e.g. many ecosystems and economic systems). As a contrast to autopoietic systems, then, sympoietic (collectively-producing) systems can be used to characterize systems that do not generate boundaries or barriers.^*16 Sympoietic systems recurringly produce a self-similar pattern of relations through continued complex interactions among their many different components. 'System-hood' depends on these continuing interactions rather than on boundary production. These systems control inputs and outputs through other means. The concept emphasizes linkages, feedback, cooperation, and synergistic behaviour. These systems rely on shared information, shared control, and ongoing self-organization.

To understand the conception and significance of self-producing systems, it is essential to understand some definitions specific to these conceptualizations. Originating from the work of Maturana and Varela, I have expanded on these definitions to some extent.^*17

• The pattern of organization^*18 of a system is the relations among components that define a system as a specific type of system (e.g. car vs. motorcycle).
• The structure of a system is the actual relations and components that constitute a particular system in a particular domain (e.g. actual car existing in physical domain). Note that this definition differs from the standard systems-theoretic definition of structure, but fits with the definition applied in biology. On a simplistic level a system's pattern of organization is the blueprint, its structure is the manifestation of that blueprint.
  Any pattern of organization can be manifest in many different structures (e.g. Subaru, Volkswagen).
• Poiesis refers to the recursive self-producing process in which a system's pattern of organization generates a particular structure which in turn generates a self-similar pattern of organization that has the ability to generate a self-similar system structure and so on (Figure 1).^*19 Reproduction refers to the production of another system - poiesis refers to continual production of the same system.

The benefit of distinguishing between pattern of organization and structure, and of conceptualizing poiesis, is the possibility for distinguishing between different types of exchange a system may have with its environment. In particular, it allows a distinction between structural exchange - energy and material, and organizational exchange - information. Note that my use of information refers specifically to information relevant to a system's pattern of organization. The distinctions between pattern of organization and structure enable description of the following two characteristics which are crucial to understanding poietic systems.

• Organizational closure refers to the degree of self-containment a system has with respect to its pattern of organization. A system can be organizationally closed, open or ajar. A system which organizes the relations among its components in such a manner as to ensure the continuation of its own pattern of organization is organizationally closed^*20 (e.g. organism). A system which relies completely on external sources for determining its pattern of organization, such as a human made artifact, is organizationally open (e.g. car). A system which relies on external sources, yet limits inputs in a self-determined manner, is organizationally ajar.^*21 An ecosystem which limits the introduction of new species is an example of the latter.
• Structural coupling refers to the interdependencies the structure of a system generates - with its environment - in the domain in which it exists, through its actual manifestation in that domain. For example, food is a structural requirement of biophysical systems. A system's need to obtain specific food from a specific environment reflects structural coupling. Continued poiesis depends on a structure's ability to obtain requirements from, and respond suitably to, its environment. Both poietic systems are structurally coupled.

Conceptualization of sympoietic systems arises from awareness of the boundary problematic, coupled with definition of these poietic system characteristics. The differences in these defining characteristics generate other significant differences and behaviours between the two system types, including different advantages and disadvantages (Table 1).^*22 Perhaps the most significant consequence of conceptualizing the distinction between the two system types is the ability to link the characteristics of each type together - to provide a theoretical basis for grouping each set of characteristics. Although any 'real' system interpreted through these heuristics will not accurately fit either set of characteristics, an interpretation that mixes 'opposite' characteristics (e.g. centralized control and adaptive) would raise questions regarding the appropriateness of the interpretation.

As illustrated in Figure 1, the critical difference between autopoietic and sympoietic systems - related to the difference regarding boundaries - is their degree of organizational closure. Autopoietic systems are organizationally closed. They generate barriers to limit external interference, separating themselves from their environments in a self-determined manner. This allows them to maintain organizational closure by controlling inputs and outputs. The distinction between organization and structure makes it possible to recognize that the autonomy of autopoietic systems is not equivalent to independence. These systems require structural exchange in order to maintain autopoiesis and will eventually become structurally coupled to specific environments, requiring specific structural inputs. Autopoietic systems, therefore, are homeostatic systems geared toward self maintenance and efficiency. Organizational closure enables them to keep out undesirable information while keeping essential information. This gives them the ability to build up complex information about their pattern of organization - a 'memory' or history of past interactions (e.g. in genes).^*23 Since this organizational information is limited to what is contained in the system, however, autopoietic systems have a limited capacity for coping with uncertainty in their environment. Their ability to adapt by changing their pattern of organization, and consequently their structure, is restricted. Since environments continually change, autopoietic systems will eventually reach a point at which adaptation is no longer possible in both a developmental and an evolutionary sense.

In contrast, sympoietic systems are organizationally ajar. Lacking self-defined boundaries, sympoietic systems are open to a continual flux of information, regulating it through internal structural coupling. This means information with the potential to influence their pattern of organization must be contained in a suitable structure in order to be incorporated into the system. As an example, new organizational information embodied in an exotic species will only be incorporated into an ecosystem if the exotic has a suitable structure - if it cannot find appropriate food, it will not survive. It is this dynamic, yet restricted, flux of information that allows the systems to be organizationally ajar. This allows them to capitalize on new information relevant to their pattern of organization as the very source of opportunities for adapting to change in their environment. Sympoietic systems evolve continuously, then, by adapting to changing conditions and by generating new ones. Historical and successional change in ecosystems are examples. In consequence, sympoietic systems are evolutionary and unpredictable with potentially infinite trajectories and the possibility for dramatic and surprising change. These are characteristics that reduce their efficiency - such continual change is not without cost. 'Stability' in a sympoietic system is a balance maintained by dynamic tension. The mechanism for 'equilibrium' is not homeostasis, but ongoing self-organization. These systems are generated and maintained through recursive interaction among many interacting global-local influences augmented by positive feedback.

In contrast to the centralized information of autopoietic systems, sympoietic systems carry information distributed among their components precluding the possibility for centralized control and limiting the potential for developing complex patterns of organization. Note, however, that the complex information in sympoietic systems is typically contained in autopoietic systems, which form the components of sympoietic systems. While the autopoietic systems depend on sympoietic systems for structural exchange (sympoietic systems effectively form the environment of autopoietic systems) sympoietic systems rely on autopoietic systems for the ability to build up complex patterns of organization and structures. Neither system is more important or independent since they are nested within each other. In addition, each system type has advantages and disadvantages, although the latter must be recognized as contextual judgements. For example, organizational closure is an advantage with respect to building complexity, but it is a disadvantage with respect to the ability for accommodating surprising environmental changes.

In the following section, I apply the distinction between these two types of systems to Kuhn's^*24 description of the progress of western science. Such application provides an example of the two system types in the social domain, aids understanding of Kuhn's interpretation, and aids understanding of post-normal science.

Kuhn's^*25 basic thesis, garnered from a combined philosophical and historical analysis, was that science progresses through cycles: "normal science" followed by revolution followed again by normal science and then again revolution. Kuhn described "pre-paradigm" periods in which random fact-gathering predominates; periods when "all of the facts that could possibly pertain to the development of a given science are likely to seem equally relevant."^*26 This stage - observable in behaviours characteristic of immature sciences - reflects the open, amorphous, characteristics and dynamic flow of information in sympoietic systems. As time progresses, individuals work on particular problems and phenomena, gradually developing methods, models, and tools. Specific communities develop, each one adhering to particular paradigms and consequently sharing common interpretations. Kuhn noted several works, including Aristotle's Physica and Newton's Principia and Optiks, that defined the paradigms for particular scientific communities at particular times. Each work was "sufficiently unprecedented to attract an enduring group of adherents away from competing modes of scientific activity" and "sufficiently open-ended to leave all sorts of problems for the redefined group of practitioners to resolve."^*27 In each case, boundaries become self-defined by the limits of the questions asked, the models and exemplars used, the criteria determining who can 'belong' to the community, and the criteria determining what knowledge is accepted and validated.

This is normal science - textbook science - an autopoietic phenomena. It is primarily a puzzle-solving activity, "predicated on the assumption that the scientific community knows what the world is like."^*28 Research proceeds as scientists ask questions, make observations, interpret answers, and develop theories, all within the context of a particular self-contained paradigm and using particular examples and models. It is a recursive, poietic process that leads the community toward organizational closure. Belief in the paradigm is reinforced "by extending the knowledge of those facts that the paradigm displays as particularly revealing, by increasing the extent of the match between those facts and the paradigm's predictions, and by further articulation of the paradigm itself."^*29

Eventually, however, anomalies arise; "normal problems" resist solution by known procedures, or equipment designed for a particular purpose produces unexpected outcomes. "When the profession can no longer evade anomalies that subvert the existing tradition of scientific practice - then begin the extraordinary investigations that lead the profession at last to a new set of commitments, a new basis for the practice of science."^*30

Casting this description in terms of poietic systems concepts, anomalies occur when paradigms - which are conceptual structures - can no longer successfully couple to their conceptual environments. Note that in these cases the poietic characteristics - pattern of organization and structure - refer to intellectual or conceptual considerations. The generic notion of paradigm - the set of relations among beliefs, models, methods and other components defining a system of inquiry - is a conceptual pattern of organization. An actual paradigm - for example the beliefs, models, and methods espoused by the positivist epistemology - is the structure of a system of inquiry. Although structures and structural relationships exist in the conceptual domain, systems will still require specific structural inputs to 'survive.' For example, any particular scientific paradigm will require empirical observations of reality that can be interpreted through the conceptual models held by that paradigm. Data that cannot be so interpreted, will be 'wrong.' Structural coupling results from coherence between the observations and the interpretations. When such structural coupling becomes problematic, either the paradigm 'dies' or the system structure must be altered. Kuhn called these periods "scientific revolutions." "They are the tradition-shattering complement to the tradition-bound activity of normal science"^*31 and involve the generation of new paradigms through a return to more sympoietic-type approaches. During the revolutionary period boundaries are opened, leaving systems organizationally ajar. Scientific communities become able to alter the direction of questioning, to re-interpret the same information from a new perspective, and to allow new members into the community even if they hold different norms and exemplars. Systems with sympoietic characteristics are re-established, and then the process of movement toward organizational closure begins again.

The importance, values, and dangers of normal science - which is necessarily reductionist in at least some respects - correlate to its autopoietic characteristics. Reflection on the importance of such characteristics provides a caution for those who would criticize the reductionist approach outright. As noted above, organizational closure allows systems to build up high levels of complex organizational information - in this case paradigms - that would be difficult to achieve with a continually changing pattern of organization. While autopoietic systems have their limitations, this is one of their crucial advantages - one that has particular relevance to science. Kuhn's descriptions and examples help emphasize their value.

Such an autopoietic approach, however, also raises concerns.

This single paragraph provides a clue as to why Funtowicz and Ravetz^*35 chose the term post-normal science. They argue for development and application of a more sympoietic approach to science in specific contexts, notably in situations involving high risks, uncertainty, and divergent values. Of particular importance they advocate opening system boundaries by expanding the peer community; making the system of science organizationally ajar by encouraging inclusion of different perspectives, different types of knowledge, and different processes of validation.

There is a distinction, however, between the type of sympoietic science encouraged by Funtowicz and Ravetz and that described in most of Kuhn's examples - a distinction that makes post-normal science perhaps even more revolutionary - or not science at all. The distinction relates to the level at which change occurs and the factors that are involved. For explanation, I diverge to consider the notion of "paradigm."

 

Paradigms and the research holarchy

Kuhn is generally attributed with popularizing the term 'paradigm.' As early as 1970, however, he noted the confusion associated with its application, admitting his own ambiguity and stating that the term had "assumed a life of its own."^*36 Paradigm typically refers to a world view or basic set of beliefs, norms and values that guide thought and action.^*37 Many difficulties and ambiguities arise because - similar to the scalar concept of system in the biological domain - paradigm is a holarchical concept (Figure 2). It is often applied in reference to the basic set of beliefs guiding thought and action at different levels of research without acknowledging which level is being considered. Just as discussions about the biosphere differ from ones about an individual organism, discussions about an epistemological paradigm will differ from ones about a methods paradigm. Considering paradigms as holarchic emphasizes the nested nature of research and the constraints and possibilities each level of the process has on another. Ontological paradigms (reflecting beliefs about existence) influence epistemological paradigms (beliefs about knowledge), which influence the emergence of methodological paradigms (beliefs about the linkage between theory and method) and, in turn, methods paradigms (beliefs about technique). Each of these levels, in turn, affects their own creation through positive or negative feedback resulting from structural exchange with their environment.

Figure 3 lists several common paradigms at different levels of the research process relating them to three different approaches to inquiry. Positivist science refers to the conventional approach - the recent interpretation of 'normal' science. Postmodernism, I use in a broad, general, sense to refer to a variety of more specific arguments and positions. Post-normal science, I place in a middling position since, as I will argue below, I believe it arises from a new epistemological stance. The diagram illustrates differences among the three approaches by noting which paradigms are deemed most acceptable at the various levels according to each approach.

Aspects lower on the list are in some sense derived from those that are higher and must therefore be compatible with the 'higher' levels. Belief in a single, objective truth, for example, rests on the ontological position that reality is something concrete and immutable that exists 'out there' independent of the researcher. In contrast, the relativist position holds that socially constructed multiple realities exist, each resulting from the interaction between individuals and their worlds, and each with an equal claim to validity. The latter understanding of reality tends to be incompatible with belief in a single objective truth. As illustrated, however, compatibility at the lower levels is achieved across a wider range of possibilities. A realist, for example, recognizing that some information cannot adequately be captured in numbers may consequently use qualitative representations. Conversely, although a researcher who holds a pluralistic perspective may be less likely to place faith in the possibility for quantitative data and analysis to adequately represent phenomena, the epistemological position does not preclude its use.

One of Kuhn's valuable contributions was to force recognition of the underlying influences that are relevant - and important - factors at each level in the research holarchy.

Scientific revolutions happen when paradigms - at any level - are questioned due to increasingly unsuccessful structural coupling; when conceptual models relevant at the various levels of research and their subsequent interpretations are not coherent with the corresponding observations of reality.

By describing and naming paradigms, Kuhn has forced recognition of their underlying influence and has consequently enabled their questioning and consideration in a more pro-active manner. For example, Thompson^*39 suggests that the "best research" comes from scientists who struggle to be "post-paradigmatic" - guided by paradigms, but conscious of the restrictions imposed by such paradigms. It is also important, however, to recognize that the "best" approach to research will depend on specific situations and objectives. Funtowicz and Ravetz make a similar point regarding the broader notion of science by developing a three-zone typology expressing axiological and epistemological concerns and illustrating the appropriate application of post-normal science.^*40 The poietic system concepts described here have similar heuristic potential, but before considering such possibilities I consider the distinction between the sympoietic approaches described by Kuhn and promoted by Funtowicz and Ravetz that I noted above.

The difference between these approaches relates to the paradigmatic level being questioned. As described by Kuhn, revolutions in science have primarily occurred with respect to paradigms 'lower' in the research holarchy - with techniques, methods, and perhaps methodologies. I believe the significance of the sympoietic approach to science that Funtowicz and Ravetz describe is their focus on the epistemological level. In particular they note concerns for the type of knowledge that is validated and accepted, and how such validation and acceptance occurs.

As noted in Figure 3, the shift to a post-normal science involves a move away from the positivist paradigm, toward a pluralist epistemology. In addition, it is based on a critical realist position^*41 - a position that recognizes fundamental ignorance; an inability to 'know' truth, despite belief that reality may exist. Since some consider the realist-positivist paradigm to be a fundamental premise of science, this ontological-epistemological shift indicates one way in which the appellation post-normal 'science' is questioned.

This question also arises due to Funtowicz and Ravetz's recognition of, and emphasis on, the interaction between the generation and application of scientific knowledge. While some argue that separation of these two aspects is critical for maintaining the validity of science, Funtowicz and Ravetz claim the need for "a new political epistemology for science."^*42 They illustrate the axiological implications of the generation of scientific knowledge, debunking the claim of value-neutrality in many complex situations. Although I agree there may be some difficulty in calling their approach 'science,' if knowledge systems are recognized as boundless sympoietic systems, the distinction becomes moot - or at least less significant. Quibbling over terminology is less important than beginning a dialogue that recognizes the blurry distinction between the generation and application of science. In consequence, it becomes essential to focus on their inter-linkage and the subsequent implications. Despite my belief that this discussion is critical and represents a significant aspect of the epistemology articulated by Funtowicz and Ravetz, I will not pursue it here. Instead, I maintain a focus on the generation of knowledge, considering the implications of sympoietic systems for understanding different modes of inquiry and their appropriate application.

Current environmental and social situations on both global and local levels involve a wide range of complex issues, including climate change, biodiversity loss, and increasing disparity between rich and poor. Such situations, characterized by the divergent values, complexity, and uncertainty noted by Funtowicz and Ravetz and others, arise from complex interactions among natural and social systems. At its most basic, science is an attempt to understand both the interactions and the issues. Yet, as many have pointed out, our adherence to the conventional scientific approach has likely been a key factor contributing to the current unsustainable conditions.

Funtowicz and Ravetz promote the notion of post-normal science as a means for coping with some of the noted difficulties. Viewing these current situations with the aid of the poietic system heuristic suggests some concerns that are relevant to considering appropriate scientific approaches and for understanding post-normal science. In particular, I consider two lines of questioning that arise:

1. Regarding sympoietic systems as the subject of investigation:
   Are current approaches appropriate for dealing with the complex characteristics recognizable in many natural and social situations when they are viewed as sympoietic? If not, how can/should we cope with such complexity?
2. Regarding sympoietic systems as the process of investigation:
   Should we generate modes of inquiry that reflect sympoietic characteristics? Can we? If so, how? At which level of research should/can we encourage them? How should/can we decide when to apply a sympoietic approach?

Considering these issues shifts discussion from the evolution of science - what has and does happen - to the prescriptive possibilities - what should happen. Understanding the distinction between autopoietic and sympoietic systems can illustrate the seriousness of the difficulties, but can also suggest possible directions to pursue.

Sympoietic systems are prevalent in the complex, diverse, uncertain interactions that characterize current environmental and social debate. As illustrated above in Figure 3, our conventional approach to scientific inquiry tends to follow positivist, reductionist, experimental, and quantitative paradigms, especially with respect to understanding natural biophysical systems. The sympoietic system characteristics described above indicate that these conventional paradigms are unsuitable for understanding sympoietic systems. For example, given the unpredictable and changeable nature of sympoietic systems, the standard experimental method, which requires prediction, testing, and replication, is problematic. So, too, is the emphasis on boundary delineation as a critical aspect of system identification. The standard approach, aimed at controlling the context, precludes the possibility for understanding critical characteristics of sympoietic systems, in particular, their ability to adapt and evolve in response to environmental change. In addition, the approach provides a false sense of certainty with respect to autopoietic systems, since it prevents recognition of their inability to react to environmental change.

Many systems carrying sympoietic characteristics have been interpreted as autopoietic. Table 2 lists a few examples, noting concepts that illustrate autopoietic interpretations that seem inappropriate once sympoietic characteristics are recognized. The table also lists newer concepts that reflect a shift toward recognition of sympoietic characteristics. The examples listed are natural biophysical systems, but social examples, such as neo-classical economics and experimental behavioural research, are also prevalent.

Lack of recognition for the significantly different characteristics carried by sympoietic systems will lead to continued inappropriate approaches. This could include both inappropriate modes of inquiry (e.g. science) and inappropriate coping strategies (e.g. planning and management). The first challenge for any science, including post-normal science, is to develop a suite of approaches suitable for investigating the complex, unpredictable, evolutionary nature that characterize systems viewed as sympoietic.

I believe such developments and changes are occurring in some situations. Table 2 lists a few, but there are also others. Table 3 lists various approaches that have been articulated and practiced. I have categorized the two columns as positivist and post-normal/postmodern, although such labels must be considered loosely. The columns also reflect a difference between approaches applied in the 'harder,' natural sciences and the 'softer,' social sciences. For each level listed in the table, characteristics of the research situation are correlated to characteristics of the research approach. In addition, the two sets of characteristics are identified as poietic system characteristics, a connection that will be reconsidered below. The presence and practice of these approaches does not, however, indicate relief from the concerns noted. Autopoietic misinterpretations are still prevalent. Application of the autopoietic lens is so firmly entrenched - especially in the natural sciences - that we only 'see' autopoietic systems. Perhaps the most important aspect of the poietic systems heuristic is its comparative potential. Conceptualizing sympoietic systems forces recognition of their 'other' characteristics. Situations conventionally considered only in the left-hand column must be shifted to the right. The examples in Table 2 reflect such a shift. This shift also forces consideration of the need for different approaches.

The matching between poietic characteristics of situation and approach noted in Table 3 implies an answer to the question of how to cope with understanding sympoietic systems. As a general approximation, I suggest that, while autopoietic approaches are appropriate for understanding autopoietic systems, sympoietic approaches are essential for understanding sympoietic systems. This statement leads into the second line of questioning - considering investigative systems as poietic systems. What would autopoietic and sympoietic modes of inquiry look like, and what advantages and disadvantages would they have?

Following from the systems descriptions noted in the second section, consider two approaches to research - one which fits the autopoietic caricature, the other which fits the sympoietic caricature. Each approach would carry the various system characteristics. The autopoietic approach would be controlled from a central position, would contain restricted information, and would be focused on maintaining the status quo advocated by a particular paradigm. These characteristics would make the mode of inquiry efficient, developmentally oriented, able to build up complex information, and able to transfer that information as a package. It would also be less adaptable, would require specific structural inputs, would be dependent on a specific environment, and would have a finite trajectory. In contrast, the sympoietic approach would be open and lack centralized control, would have different information distributed among many different components, and would be focused on considering new possibilities and opportunities. It would be less efficient but more adaptable, and would transfer information in a disconnected, amorphous manner. In consequence, the potential for building up complex common understanding and knowledge would be restricted, but the possibilities for generating unexpected connections leading to new ideas, concepts, and understanding would be increased. Each approach, then, has both advantages and disadvantages, depending on the situation.

Perhaps the most critical advantages of autopoietic systems, are their efficiency and their ability to build up complex organizational information. The most critical disadvantage is the closure and isolation that is both generated and required by this efficiency and build-up. These descriptions illustrate the advantage and disadvantage of the positivist reductionist approach that has characterized science over the past couple of centuries. Consider a comment on physics, which is typically interpreted as the 'hardest,' most fundamental science:

In the centuries since the 'scientific revolution,' the positivistic epistemological paradigm, has primarily remained unquestioned - at least within the domain of 'science.' An autopoietic approach, it has tended to advocate autopoietic approaches at other levels: expert driven reductionist methodologies, experimental methods, and quantitative recording techniques. These autopoietic approaches have been instrumental in generating the wealth of knowledge and technical expertise we currently appreciate, yet they have necessitated the blinders that hide the impacts and implications resulting from application of this knowledge and expertise.

In contrast, the most critical advantage of sympoietic approaches is that they are organizationally ajar - open, yet not completely open, to new information. This means systems have an increased adaptive potential. They are able to cope with external surprises by creating new possibilities for structural exchange. They do this through alterations to system structure that may be generated when new organizational information is incorporated into the system.

The descriptions reflect two quite different approaches to inquiry - each of which would be appropriate in particular types of situations. The comparisons suggest that an autopoietic approach would be suited to situations where efficiency and the development of complex paradigms (patterns of organization) are more important than the ability to adjust to external changes. A sympoietic approach would be appropriate in situations involving the reverse conditions. The scope and depth of complexity and uncertainty involved in understanding current social and environmental situations indicates increasingly unsuccessful structural coupling between 'reality' and our 'scientific' paradigms - between our observations and our methods and models for interpretation. It is, therefore, essential at this stage to develop sympoietic approaches for science - to develop more organizationally open and structurally adaptive possibilities. This, then, is the significance of the epistemological revolution - the shift toward pluralism and toward democratization - that Funtowicz and Ravetz advocate by describing post-normal science. Generating sympoietic modes of inquiry - on all levels of the research process - is essential for dealing more appropriately with the prevalence of situations that exhibit sympoietic characteristics.

There is a considerable challenge here. Unlike physics, as noted in the above quote, the specified domain of post-normal science is the real world - including the complexity, uncertainty, and irregularity that is both inherent and ubiquitous.

The above discussion (see Table 3) has indicated that we can, and have, generated some suitable approaches.^*44 There is, however, a considerable lack of recognition for sympoietic systems and, consequently, for the need to consider sympoietic approaches. This is particularly the case at the interconnected methodological-epistemological level - a concern indicated by the call for post-normal science.

This leaves one aspect of the questions regarding sympoietic systems unanswered: How can/should we encourage sympoietic approaches? Or, to reframe the question: What can the poietic systems heuristic suggest for encouraging post-normal science? I consider some possibilities.

In the discussion above, I emphasize the importance of the epistemological position articulated by Funtowicz and Ravetz. Given the foregoing discussion, the most obvious suggestion is that to encourage post-normal science we must emulate sympoietic characteristics. In particular, we must encourage development of organizationally ajar science at an epistemological level. This means pluralism without chaos: the system of science must be open, yet not totally open, to new information and to organizational change.

Such a suggestion presents a significant challenge, primarily because autopoietic systems of inquiry - those with firmly entrenched paradigms - hold important advantages for developing any type of knowledge. As noted, paradigms both generate and allow the common understandings that are critical for the development of complex ideas, concepts, and theories. Is it possible to pursue research or develop a science without paradigms? Are they necessary on all levels of the research process? For example, is it possible to have paradigms to follow at the methods and methodological levels, but not at the epistemological level? Are paradigms inherently autopoietic, or is it possible to prevent the move toward organizational closure in the conceptual domain? For example, does insistence on the 'rightness' of pluralism reflect a move toward closure and development of an autopoietic - even if pluralistic - paradigm? Is it possible to be pluralistic and to build knowledge? Is building knowledge necessary for post-normal science?

To consider possible responses, I begin with explanations arising from application of an ecosystem metaphor. Since ecosystems can readily be interpreted as sympoietic, comparisons with current understanding about ecosystems may illustrate useful possibilities for encouraging post-normal science. Three considerations seem relevant. First, is to compare biophysical holarchies with research holarchies. In the former case, sympoietic ecosystems consist of numerous, diverse autopoietic organisms. Systems are inefficient yet adaptable at the broad ecosystem level. Complex information and the closure essential for its generation is contained at the organism level. As with such biophysical systems - a sympoietic epistemology may not have to encompass sympoietic approaches at 'lower' levels in the research holarchy. Research within a sympoietic epistemological paradigm may consist of numerous, diverse autopoietic and sympoietic methods, each one developing specific, complex information through application of a particular methods paradigm. Consider the potentials offered by advocating a wide variety of methods or - perhaps more importantly - by advocating the interpretation of results in a wide variety of different ways through application of different methodological paradigms. Consider the possibilities and advantages offered by a system that can incorporate a wide array of information, ranging, perhaps, from the results of experimental research to the artful, intuitive understanding generated by rich oral histories tied to a specific place. Such systems could have considerable adaptive advantages due to their wide range of accessible knowledge.

Second, internal structural coupling, as exemplified by successful or unsuccessful incorporation of an exotic species into an ecosystem, is the mechanism used by sympoietic systems to restrict information input. What are the implications and possibilities of structural coupling for an epistemological system? What structures in a mode of inquiry are essential for allowing it to remain organizationally ajar, instead of either closed or open? Consider the difficulties presented by attempting to incorporate information represented in a fundamentally different format than that which is standard for a particular paradigm. A classic and ubiquitous example is the difficulty of incorporating qualitative information into simulation models. Scientists accustomed to quantitative information find it difficult to accept, validate, and use qualitative information - and vice versa. How can such structural variations be overcome, or should they be? In addition, what are the implications for the integrity of the new information? For example, researchers attempting to translate traditional ecological knowledge into conventional scientific terms have concern for the integrity of the traditional information. Without the cultural context, significant connotations, implications, and understanding may be lost.

The third consideration, is Holling's infinity cycle and the notion of creative destruction.^*45 As applied to some ecosystem types, the model illustrates cycles involving periods of increasing organization and storage, followed by periods of catastrophic release, followed by renewal. This cycle is reminiscent of Kuhn's interpretation of the evolution of science: increasing closure resulting from relations developed among a scientific community and its paradigms, followed by revolution. If sympoietic systems in science must eventually become autopoietic, can we encourage their continual renewal without suffering major catastrophes?

These considerations provide more questions than answers, but they illustrate the value of conceptualizing reality as sympoietic through application of an ecosystem metaphor. Continuing in a similar vein, I look more specifically at sympoietic systems characteristics. As with the preceding considerations, rather than provide answers, I simply pose questions. Each set of questions arises from reviewing a particular sympoietic system characteristic. The rationale is to wonder if we can develop sympoietic approaches simply by supporting as many sympoietic characteristics as possible. Rather than attempt to develop a sympoietic system on a complete and comprehensive macro level, it may be possible to simply emulate the various characteristics of sympoietic systems - to focus on the micro level instead. Consider the following.

Information and control need to be distributed and shared among different components of a sympoietic system.

• Who, or what, can/should the components of the system be? What type and amount of information can/should these components have?
• How can/should the information be shared? How can we communicate among members of an extended peer community?
• Can we be comfortable without anyone knowing 'everything;' without anyone being 'in control'? Can we be comfortable among individuals with different types of expertise? Or among individuals who have gained knowledge through different means than our own? How do we develop trust among an extended peer community?

Maintaining dynamic tension is critical for system generation and continuation.

• How can we create such conditions? How can we invite discourse, encouraging the expression of different 'facts' and perspectives? What type of forum will enable such diversity?
• How can we maintain a balance between the need for a common language and the need for listening to different voices?

Structural coupling is also critical.

• What types of information are currently validated and accepted? Can 'non-scientific' types of information and knowledge be incorporated into any scientific system? If not, how can we develop a system structure to ensure that new organizational information can be incorporated?
• Can we alter the structure of science to ensure such opportunities? Can we promote such changes without altering the beneficial aspects of science?
• Can we be comfortable recognizing environmental uncertainty as an advantage?

Reproduction in sympoietic systems is amorphous and continuing.

• How can we ensure that new knowledge participants are not indoctrinated into specific, bound, paradigms, but receive a broad, eclectic range of ideas, perceptions and interpretations? Is it possible to do so without losing the advantages autopoietic paradigms provide? Is it possible to gain an 'adequate' education in this manner?
• How can we ensure that learning becomes a continuous process on both individual and social levels? How can we determine who should lean what?

Over the past couple of centuries, autopoietic type approaches have become the norm for science at all levels of the research process in many disciplines and institutions. Although valuable for the development of knowledge, there have been significant costs, evidenced by our inability to understand and cope with current environmental and social conditions. The complexities and uncertainties we now face require significantly different approaches. This recognition has prompted Funtowicz and Ravetz to advocate development of what they call post-normal science. Arguing for a methodological and epistemological shift, they suggest extending the peer community and accepting extended 'facts.' Moving toward such a post-normal science requires the development of sympoietic approaches as a complement to our autopoietic norms.

The system concepts described in this paper provide a starting point for addressing such issues. Contrasting the characteristics of autopoietic and sympoietic systems provides a useful heuristic that is relevant in two different ways. First, viewing many situations of concern as complex, self-organizing, boundless sympoietic systems forces recognition that anomalies ignored or considered irrelevant may be integral to the systems. Such conceptualization illustrates that conventional approaches are inappropriate because they cannot cope with sympoietic systems. Conventional approaches have been developed to investigate autopoietic systems and consequently lack the ability to recognize and understand sympoietic characteristics. One of the key benefits of the sympoietic lens is that it admits to the presence of the autopoietic lens - we can 'see' systems that exhibit both autopoietic and sympoietic characteristics. This allows for their distinction and comparison.

Second, the poietic system heuristic enables recognition that a choice of approaches is possible since it enables recognition of autopoietic, sympoietic, and combined modes of inquiry. In addition, the heuristic provides a tool for considering the advantages and disadvantages of the various possibilities and for considering when they might be applied most appropriately.

I interpret post-normal science as a sympoietic approach, at an epistemological level. It admits to the value of autopoietic approaches, but emphasizes the need for sympoietic approaches also. We must promote situational application of the different approaches. In particular, we must encourage development of sympoietic approaches in situations involving divergent values and high levels of complexity and uncertainty - situations readily interpreted as sympoietic. A conclusion that confirms the arguments of others. The poietic system concepts described here, however, provide a different logic to the argument. Characterization of sympoietic systems provides a conceptual model that more appropriately reflects the complexities and uncertainties that can be observed in many natural and social systems. Although many of the challenges I have noted are unanswerable, the heuristic generates many new questions and has the potential to introduce new perspectives, new understandings, and possibly new responses and directions.

In closing, I reconsider the positions I stated in the introduction.

The systems concepts introduced here have significant, wide-ranging implications. To describe the systems concepts, I have attempted to cover some breadth and use sufficient examples to generate clarity. However, many considerations, questions and possibilities remain unarticulated. Future work will consider further implications and details.

Kuhn's interpretation of western science - although not fully accepted - provides valuable understanding, especially when considered from a poietic systems perspective. I believe the concepts described here strengthen his interpretation. Recognizing these concepts as heuristics, however, emphasizes that application of a different heuristic may provide different, and valid, understanding. The question, then, becomes the usefulness of the heuristic and its subsequent interpretations. I believe the discussion here illustrates the usefulness of both Kuhn's interpretation and of the poietic systems heuristic.

Although I have some discomfort with the appellation, post-normal science, with respect to both components of the term, I agree in principle with the arguments of Funtowicz and Ravetz. In addition, I believe it is far more important to have labeled the concerns and initiated discourse than to worry about the details of terminology. The discussion presented here would be valid regardless of the name applied, so I have not entered the terminological debate.

My concern regarding the degree and level of compatibility between conventional and post-normal science was resolved in the above discussion by noting a distinction between epistemological and other levels of the research process.

I noted that this discussion is set within a western scientific cultural perspective. When I claim that 'we' must develop new approaches, I refer to people embedded within this culture. Understanding this scientific culture through the poietic systems heuristic has indicated its predominantly autopoietic nature. Applying the heuristic to other cultures and knowledge systems may provide interesting considerations, comparisons, and learning possibilities.

Finally, I reiterate my underlying concern for sustainability, which provides the normative basis for this discussion. For those yet unconvinced of the seriousness of current conditions, I have not provided arguments to engender such concern. They must turn elsewhere. For those who do recognize the fundamental nature of the problems we face, I hope to have provided a new way of looking at these situations that can suggest potential directions to pursue for coping with their complexity.

I gratefully acknowledge the assistance of many throughout the thinking and writing process that led to this paper. To Dr. James Kay for introducing me to post-normal science, for comments on earlier drafts, and, most specifically, for conversations that clarified the implications of the systems concepts for post-normal science. To Eric Tucs for comments, conversations, and questions that kept me reflecting on fundamental issues. To the post-normal science/complex systems discussion group convened by James Kay at the University of Waterloo (and beyond), which continues to provide a stimulating intellectual forum for contemplating such issues. To the many authors and scholars, especially those not directly cited, who have shaped my thinking over the years. Finally and importantly, to the Social Sciences and Humanities Research Council of Canada (Masters in Science Policy) for financial support during graduate studies prior to the writing of this paper, which reports on aspects of my Masters in Environmental Studies Thesis.

*¹ Christensen, Normal L., Succession and Natural Disturbance: Paradigms, Problems and Preservation of Natural Ecosystems in Ecosystem Management for Parks and Wilderness, J. K. Agee and D. R. Johnson eds., University of Washington Press, Seattle, Wash., 1988.