Luhmann's Sociological Theory:

Its Operationalization and Future Perspectives


Loet Leydesdorff

Department of Science and Technology Dynamics

Nieuwe Achtergracht 166

1018 WV  Amsterdam

The Netherlands



            Luhmann (1984) has proposed a second-order theory of social communications, but its formalization in terms of second-order systems theory has remained underdeveloped.  Second-order systems theory is a formal option, and furthermore Shannon's (1948) mathematical theory of communication is available.  The operationalization of Luhmann-type (reflexive) communications in terms of Shannon-type (first-order) communications has theoretical consequences: one is able to distinguish, more clearly than Luhmann did, between not (yet) meaningful information ("uncertainty") and its potential meaning after selection by an observing system.  Structural coupling between co-evolving systems can be distinguished from operational coupling between sub-systems.  This operationalization provides us with means to clarify, among other things, the theoretical debate between Münch and Luhmann about Parsons' concept of "interpenetration".  Technological developments can be analyzed in terms of operational and recursive coupling at the interfaces between sciences and markets.  In a triple helix model of university-industry-government relations codes of functionally differentiated communication can be translated into eachother.  Interorganizational configurations support the emerging communication systems.


Luhmann's Sociological Theory:

Its Operationalization and Future Perspectives


            Elaborating on the work of neurophysiologists like Maturana (e.g., 1978), Luhmann (1984) has proposed a second-order social systems theory.  This theory considers communications among human beings as units of analysis, and thereby explicitly contextualizes human actors to function as the carriers of a network of social relations.  On the one hand, this de-humanization seems a radical departure from sociological theorizing, which has aimed at "Verstehen" and sociological enlightenment about human intentionality.  On the other hand, Luhmann's (1984) theory differs from Parsons' (1937) systems theory of social action to such an extent that Luhmann's contribution has been characterized as a "second-order" systems theory.  But what does "second-order" mean in this context?

            In my opinion, second-order systems theory should be considered as a methodology: second-order systems can be defined only in terms of distributions.  While first-order systems can be identified theoretically, distributions are expected to contain an uncertainty.  Thus, second-order systems cannot be delineated clearly in terms of empirical observables, although one is able to specify an expectation.  However, a hypothesis requires a substantive specification.  Thus, a second-order theory of social systems requires the combination of the methodological perspective with a sociological theory.

            By considering the network of communication among people as the subject of sociology Luhmann has indeed provided us with a theory that takes distributions as its units of analysis.  However, his theory has hitherto remained speculative; if one has both theory and methods, one should in principle be able to specify testable hypotheses.  I shall argue below that the operationalization of concepts from Luhmann's sociological theory of communication in terms of a mathematical theory of communication enables us to clarify semantical confusions to the extent that theoretical questions can be made empirically researchable.


The distinction of "the social system of reference"


            At a time when an almost "incommensurable" divide between social systems theory and symbolic interactionism had been noted in sociological theory (e.g., Grathoff, 1978; Giddens, 1979 and 1984), Luhmann (1984) submitted a theory which combined elements from both these traditions.  On the one hand, Luhmann's theory shares a reference to Husserl's phenomenology with symbolic interactionism.  Luhmann, however, proposed that the constitution of "meaning" be operationalized in terms of codification in systems of communication.  On the other hand, Luhmann elaborated on Parsons' systems theory by stressing (functional) differentiation in systems of communication, but he distinguished more clearly than Parsons between society and the aggregate of human actors.  Parsons considered this distinction as a focus of organization, and he defined the role of the actor as the conceptual unit of the social system (e.g. Parsons & Shils 1951, at p. 190).  A social role, however, remains an attribute of an actor.

            Luhmann (1984) extended Parsons' methodological distinction into a difference at the epistemological level: he proposed to consider society and human beings as different systems of reference.  Although the social system and the personality are both contructed in terms of (inter‑)actions, the dynamics of the aggregates, i.e. their "life-cycles", are expected to be different.  Elsewhere, I have provided an operationalization of this model in terms of "parallel and distributed processes": each individual runs its own psychological cycle as a local processor, and the psychological systems disturb one another by communicating at the level of the network (Leydesdorff 1993a).  Let me here elaborate on the relevant statistics in order to further clarify these issues.

            In a standard (i.e. first-order) research design, actions can be attributed to actors as variables.  For example, one may score the behaviour of actors in a social setting.  The actors are then considered as the cases, and the attributes as variables.  Conventionally, one sorts the cases as the rows and the variables as the columns of a matrix.  However, the columns of the resulting matrix are a different frame of reference than the rows.  The columns teach us about the distribution of a particular variable (e.g., "right- and left-handedness in writing") over the relevant social system ("the domain"), while the rows inform us about the properties of individual actors.  In other words, changes in the attributes of observable cases are expected to have an effect on the distributions over the columns.  In the co-variation between rows and columns, (Shannon-type) information is communicated between the units (e.g., actors) and the system of their relevant distributions (cf. Theil 1972).

            One is able to define "structural coupling" and "operational closure" at this level of abstraction.  For example, each potential difference in a cell value on the occasion of a second measurement informs us in the vertical dimension of the matrix about a development at the level of the distributed system under study, and at the same time from the other perspective about a change at the observable actor level.  Each single cell value is an event that teaches us about a momentary and local coupling; another row or column can be relatively unaffected by this change!  Thus, the two systems of reference are coupled at specific sites of their structure.  Over time, the co-variation in the cell-values can be taken as an indicator of their mutual "interpenetration" (Parsons, e.g. 1968) or "structural coupling" (cf. Maturana 1978): the distribution at the level of the network does not contain information without the actors acting.  However, the relative weight of a co-variation in relation to the total variation over the columns and over the rows is expected to be different.[1]  The two systems, i.e. the first-order observable units of analysis and the distributions, are expected to develop on the basis of their total variation.  If co-variation is repeated over time, structurally coupled systems may exhibit a co-evolution.

            Figure 1 illustrates how co-varying systems by definition determine one another in the (symmetrical) co-variation, while they condition one another in the remaining variation (Leydesdorff 1995a).  In other words, to the extent that two systems do not disturb eachother in terms of a co-variation, they leave one another "free" to follow their own dynamics over time.  Giddens (1979 and 1984), for example, identified this relation between mutual determination and the conditioning of two interacting systems as the "enabling and constraining" functions of structure for action.


1Figure 1  Relations of variation (expected information content H), co-variation (mutual information T), and remaining variation (conditional entropies H(x|y) and H(y|x)) between two variables x and y (Attnaeve 1959).

            In summary, we have operationalized "action" as an attribute of the actor which at the same time can be considered as a communication with reference to the distributed system under study.  Communication systems are distributed systems; they contain an uncertainty, and therefore their boundaries remain an expectation.  The social system can be considered as such a distributed system.  But hitherto, our analysis has been so abstract that the reasoning can be applied to any research design, for example, to plants in a biological research design about the properties of plants.  We shall thus have to specify what turns a communication system into a human communication system.

            On the basis of this operationalization, we are able to understand what it means that distributed (i.e. second-order) systems are "operationally closed" because of their structural coupling in terms of variables.  Variables are specific: things (or beings) to which one cannot meaningfully attribute a notion like "left- or right-handedness" cannot be included into this distribution.  (Perhaps, they communicate in another dimension.)  The column of a research design contains a substantive selection.  Operational closure is thus a consequence of the research design.  By substantive specification, the researcher provides the uncertainty with a meaning: one specifies what will count as a signal and what as noise.  Although it remains uncertain what delineates the system, a selective criterion is provided by specifying the substance of the communication.

            Note that we have introduced the concept of substantive "meaning": variation can only be interpreted as meaningful information if one has a (sometimes implicit) expectation of what will count as a signal.  Events in this specific dimension can then be used for the update.  In a research design, one explicitly specifies this expectation, and thus the update can be made reflexively.  Meaning requires the introduction of an a priori system of reference (e.g., "an observer") in terms of which the information that has become available in the event ("the measurement") can be provided with a meaning.

            If one studies human relations, the meaning specified by an analytical observer at the level of the research design can be different from the one constructed by each of the observed participants.  Human communication involves both the variation ("uncertainty") and what this information means.  If an analyst maps into a matrix the various meanings as attributes of the observers and/or participants, one obtains another matrix containing the communication of meanings over the columns.  (The two matrices may co-vary.)  Thus, if the subjects under study are themselves reflexive, the research design "doubles" reflexively, and one obtains a design of second-order cybernetics.


Meaning and human communication


            Reflection is not yet a sufficient criterion for human communication.  An ordinary reflector or mirror turns an incoming light beam into an outgoing one.  A reflexive system, in general, generates an output as a function of an input.  In other words, a reflexive system can also be considered as providing an observed variation with a meaning.  As against ordinary reflexive systems, human observers can be considered as hyper-reflexive: in addition to providing all observed variation with meaning by attributing categories to it, human beings are able to communicate about these reflections, e.g. by using language as a communication medium.  Thus, human observers are able to distinguish the two matrices specified in the previous section in relation to one another, and if they wished, they could revise either of them by making other selections.  Meaningful communication contains two kinds of information: information in the sense of uncertainty and in the sense of meaning.  In my opinion, natural languages among other things codify the relation between information and meaning in human communication: a statement can have a meaning, and it may contain information. 

            In other words: a reflexive system uses its a priori state (i.e., its momentary structure) as a frame of reference to assess the relevance of the observable events.  By doing so, the system turns into an a posteriori state with reference to the observation.  This may, in turn, lead to further actions.  Thus, in the case of reflexive systems we have to extend the two dimensions of the above matrix of rows and columns by adding a time axis as a third dimension.  Matrices at each moment in time add up to a cube of information over time.  The succession of the reflections may exhibit an observable trajectory within this three-dimensional geometry (see Figure 2).




Figure 2

An observable trajectory of a (potentially complex) system in three dimensions




Figure 3

Selection among representations of the past using a fourth degree of freedom


            If the system itself develops in the way it reflects over time, it contains one more degree of freedom or, in other words, it has become hyper-reflexive.  A hyper-reflexive system operates in four dimensions; it can revise the meaning given in the past to a disturbance with hindsight, since it develops in the present (see Figure 3).  Both its elements and its operations are in flux (cf. Swenson 1989).  In my opinion, Luhmann's thesis can be understood as the hypothesis that the social system evolved to reach this fourth degree of freedom during the individual revolutions of the 16th and 17th centuries, and that the social communication system has been developing its hyper-reflexivity with increasing speed ever since.  This hyper-reflexivity enables the social system to functionalize the communications of which it consists with reference to its present state and its future options.  While a High Culture is presumed to have a static centre for the reflection, the modern social system is based on the assumption that reflections are distributed and can be communicated reflexively.

            In summary, uncertainty is generated when two systems (e.g. an actor and the network) relate; each co-variation is local and therefore occupies a structural position in either system at each moment in time.  The communication of uncertainty can be provided with a meaning over time to the extent that the receiving system is reflexive.  If the system has an additional degree of freedom in a fourth dimension (time and space), it can change its perspective on the position of a local event with hindsight.

The concept of "uncertainty" or variation in a cell value is equivalent to the definition of information in Shannon's mathematical theory of communication.  However, Luhmann (1984)-in accordance with other sociologists (cf. Bailey 1994)-focused on the concept of meaningful information.  Although human communication is specific in its capacity to communicate about meaning, this should not obscure the difference between the communication of meaningful information and information that is not yet meaningful.  Otherwise, one might lose sight of the basic operation of non-social, or not yet social communication.  The sender generates variation of which only the part that is communicatable is selected by the network.  The receiver selects by providing this uncertainty with meaning.  However, the messages can be meaningful at some places in the network, and meaningless at others.  Thus, the social system receives messages in a distributed mode.

            The distinction between information and meaning can be expressed in terms of "variation" and "selection".  The systems disturb one another in their co-variation, which is observable in terms of events (e.g., actions).  The disturbance, however, is local-it affects some rows and columns, and not others-and thus it occupies a position in each system of reference.  Because all distributed systems are constructed, and therefore operationally closed (see above), some disturbances will be selected and others discarded.  Selections can operate on one another: some momentary selections are sometimes selected for stabilization over time in the geometry of a three dimensional space.  Each stabilization is local and provisional.

            In a four-dimensional hyper-space, some stabilizations can be selected for globalization.  A global system is hyper-reflexive since it operates in a hyper-space: it has an additional degree of freedom with reference to the historical contingency of its genesis, and to all its instantiations (cf. Giddens 1984).  Consequently, a hyper-reflexive social system provides each event not with a single meaning, but with a distribution of meanings.  Instantiations can be considered as local stabilizations.  They are the momentary analogues of the longitudinal trajectories or, in other words, they constitute another geometrical representation of the complex and dynamic system.  The hyper-reflexive system, however, is by definition in flux.

            Two such systems are of particular interest to sociological theorizing: individual identities and social regimes.  In summary, we have explained how these co-evolving systems are reproduced as different systems by giving other meanings to the relations in terms of which they couple structurally.


The observables in sociological theorizing


            Unlike a trajectory or an instantiation, a regime is not observable.  A regime can be considered as a distribution of identities which remain in flux.  Sometimes, the one picture is more important than the other.  In general, one can observe only its instantiations or its historical genesis from a perspective, i.e. by using a geometrical metaphor (cf. Haraway 1988).  Discursive metaphors, however, may use different axes for reflection; to the extent that these perspectival axes are orthogonal, the windows of reflection will be incommensurable.  Thus, the semantic confusion in sociology is predictable following Herbert Simon's (1969) expectation that an emerging system will be evolutionarily more successful if it achieves nearly decomposability into orthogonal dimensions.

            The absence of a tangible substratum poses a methodological problem for second-order sociological theorizing.  While a psychologist may be able to identify a human identity as a unit of analysis, the sociologist has to be reflexive about the distributed character of observations.  Distributed observations contain an uncertainty, and therefore one can specify only an expectation.  In other words, an empirical account teaches us about the case which historically occurred, but it does not yet specify the range of cases which could have occurred.  The account enables us to specify a hypothesis (or perhaps a heuristics).  In this sense, the second-order programme is radically anti-positivist.

            But how can one test a thesis about a non-tangible substratum?  How can one distinguish between mere speculation and statements with testable consequences?  The problem of the inherent constructedness of the social system is amply reflected within the sociological tradition.  Each sociological construction can be deconstructed by changing the system of reference, for example in terms of the boundaries of the relevant domains in space and time.  In addition to the noted possibility of (nearly orthogonal) perspectives, the post-modern scepticism about "grand theorizing" has stimulated a proliferation of partial perspectives.  Partial perspectives, however, tend to proliferate the semantics without methodological control.  Consequently, a major problem in sociology today is one of parsimony: a methodology should help us to be selective vis-à-vis the wealth of narratives.

            Second-order systems theory provides us with such a criterion.  The geometrical metaphors which specify the observations positively can be formulated in terms of selective-i.e. negative-operations on otherwise noisy data.  The selections are expected to change the shape of the distributions.  Thus, one is able to predict the probability distribution of possible events on the basis of a theoretical statement, although one is not able to predict any single event.  Additionally, selections can be formulated as conditions in computer code, and then the systems under study can be simulated, in principle.  The algorithmic approach of second-order systems theory provides us with options to combine the various geometrical perspectives in terms of their relative weights.  This simulation allows us to specify an expectation with respect to the possible further developments of the system.  Of course, such a prediction will be probabilistic, and thus as fallible as a weather forecast.  However, the simulation results and the observed variations may allow us to improve the theoretical specifications which go into the simulations in a subsequent round.

            As noted, an evolutionary expectation can be specified with reference to the further development of the various discursive reflections.  Scientific discourses reflect on the social systems under study.  Inasmuch as these systems of discursive reasoning ("the paradigms") are evolutionarily developing communication systems, they can be expected to develop along the trajectories set by their historical genesis, and to differentiate along the different axes of the system under study.  There is no expectation of a "meta-paradigm," since one is not able to stabilize a geometrical metaphor in four dimensions.  But the lowest, i.e. most parsimonious number of nearly incommunsurable paradigms which can be stabilized in relation to one another may sometimes prove to be specifiable (Leydesdorff 1994a and 1996).


From observable identities to computable distributions


            "Complexity", "chaos", "noise", etc., continuously irritate all systems at the places of their local carriers.  Variation is continuously generated in the network by its lower-order carriers which are expected to operate according to their own dynamics.  But only the co-variation with the network is selected.  If this uncertainty can be provided with a meaning, the uncertainty may be reduced by a further selection.  Reduction of the uncertainty by a second-order selection contributes to the system's (provisional) stabilization.  Codifications through recursive selections provide higher-order stabilizations.  Each higher-order stabilization remains provisional, since the configuration develops over time (as in a life-cycle).

            In principle, the constructed distributions are in flux; they do not exist objectively, but as an expectation of their further operation.  In other words, the distribution of a system itself contains an expected information, i.e. a prediction about the system's future operation.  (This prediction can be improved if more, e.g. previous, states of the system are known.)  The second-order systems under study do not have to be identified; they can be specified as a difference or a distribution of a variable to be attributed to cases which can be observed.

            For example, one is able to observe the first-order systems which span the second-order systems as distributions, and then to study the behaviour of these distributions over time.  By studying the observable rows of the matrix as sub-dynamics ("actions") of the communication system, one obtains an expectation at the level of the emerging system.  In other words, the sub-dynamics correspond to the "genotypes" of the system, while the second-order results are "phenotypical."  Only sub-dynamics can be discursively reflected into a geometrical narrative without too much confusion.

            Of course, the "phenotypical results" can again be interpreted, and a second-order matrix can be constructed reflexively.  Thus, the systems are constructed in layers of reflexivity.  When first-order systems operate, they produce disturbances which can be provided with meaning by each of these systems (if they are evolutionarily able to do so).  In doing so, they generate a second-order layer of communication.  If the systems are hyper-reflexive, they can shift the focus of their reflection back and forth from the first-order to the second-order system under study, and thus reconstruct their reconstructions.  Selection upon a pre-selected distribution potentially closes the system by stabilizing the representation locally.  A further selection reopens it globally, for example as an expectation of its potential developments in the future.

            The crucial yardstick for scientific communications-as opposed to integrated belief systems-lies in the capacity of scientific discourses to specify such expectations reflexively.  This possibility exists in principle since a distributed system consists only of expectations.  At this stage of cultural evolution, specification of the expectations contained in the discursive system seems to provide the highest level of reflexivity the social communication system is able to achieve (Leydesdorff 1994b).


Consequences: Luhmann and beyond


            The introduction of a formal perspective allows us to clarify, for example, theoretical debate between Münch and Luhmann about "interpenetration" in terms of operational concepts.  However, I wish first to address a flaw in Luhmann's work about uncertainty and the reflexive communication of information.  This operationalization enables us to disentangle the various concepts of "interpenetration" in terms of structural and operational couplings.  In a final section, I address "technology" as an inter-systems relation.


a. the communication of information and meaning


            The wealth of heuristic semantics which Niklas Luhmann has developed over the last decade cannot hide the fact that the formalization of his systems theoretical perspective has remained underdeveloped.  The various concepts are not operationalized, and the codification is at some places confused.  Luhmann appeals to systems theory to introduce counter-intuitive arguments, but he warns against the unreflexive application of biological metaphors from systems or evolution theory to social systems (cf. Leydesdorff 1993b).  In short, Luhmann has not provided us with means to test whether his statements are more true than others (e.g., Maturana's), and therefore the reader often has no option other than to trust the author, or not.  At certain places, Luhmann himself signals this problem.  For example, Luhmann (1990) mentions that a more elaborated systems theoretical notion is needed to clarify inter-system dependencies.  I shall return to this issue in a later section.

            As noted, the introduction of second-order cybernetics as a methodological apparatus is not theoretically neutral: it changes the epistemological status of theorizing.  Theories provide us with "genotypical" specification of sub-dynamics, i.e., with partial perspectives.  In order to proceed from discursive speculation to analytical models and empirical research, the substantive theories have to be formulated as conditional statements.  In principle, these conditions can be combined as "do while" and "if then" statements in algorithmic code, making the observable events testable against a distribution of possible events.  Thus, the analysis of the history of the current system is only a first step: it provides us with a description of the co-variations that have been the case.  The next step is to specify theoretical hypotheses about the co-varying systems that potentially explain why the observed events occurred, and then to develop means to test such hypotheses against simulations of other possible operations of the assumed system.  Only thereafter, one is able to check systematically (as contrasted to incidentally) for flaws and mistakes in discursive metaphors.

            Luhmann, for example, has not wished to distinguish between communication, as the transport of information, and human communication which implies information, message, and understanding.  On the contrary, he has defined these three operations as the very unit of communication.  But as argued above, meaning is reflexive upon the information contained in the variation.  It requires a second operation by an observing system, while the observed variation itself is the consequence of a first-order operation.  In human communication, these two operations are coupled, yet this coupling is not complete but reflexive, i.e. selective in the time dimension.  Some information is provided with meaning, while other information is discarded as noise.  Sometimes a message can be rich in information, but one is nevertheless not able to understand it.  At other times, we give meanings to things that are not (yet) based on human communication.

            In order to "understand" a communication, the receiver needs to decompose the incoming variation into a signal-i.e. the meaningful information-and noise.  Additionally, human receivers are complex enough to distinguish between these two dimensions and the contextual position of the information at each moment.  Thus, each reconstruction is one among a range of possible reconstructions, since the decomposition can be revised; the understanding is one among a range of possible understandings.  But the distributedness of the understanding should be attributed to the distributed system.

            Luhmann has focused on the (local) understanding of the communication.  Thus, he has chosen the reconstruction of the signal at the receiving actor as his frame of reference.  However, actor's understanding is local and contingent.  What does it mean for the distributed system ("the social network") that the communication is understood locally?  The social system "understands" the information in the messages in a distributed way, i.e. not as a (provisionally) stable meaning, but as a global hyper-meaning.  Thus conceptualized, the social system is one layer more complex than it would be without this conceptualization.  It operates with four degrees of freedom, not with three.  While Luhmann has distinguished three operations in addition to self-organization (variation, selection, and stabilization), self-organization can be understood in our model as the next-order recursion of the selection.  Some local stabilizations are selected for globalization into a prevailing regime.

            In other words, where Luhmann (1984, at p. 103) has defined information as "a difference that makes a difference" (Bateson 1972, p. 489), we maintain that this is a definition of meaningful information.  The more parsimonious concept of information as "a difference" (or a distribution) allows us to understand distributed systems as themselves containing an expected information.  The recursion of the selection should then be understood in an orthogonal model: the different layers build upon one another as (nearly) decomposable dimensions without necessary hierarchies.  In general, the next level of communication should not be understood in terms of aggregates of lower-level units, but in terms of their interactions.  The hierarchical model of relations describes only the special case that order has (provisionally) been stabilized.

            The relevant differences define the systems under study as distributions.  A distribution, however, can be considered as a probability distribution; variation in probabilities presupposes a selection from randomness.  The crucial point is the recursivity of the selective operation: a probability distribution can have a probability, a selection can be selected, etc.[2]  What is being communicated is system-specific ("operationally closed"); how it is communicated is the subject of a second-order cybernetics. The systems drive one another by (sometimes even stochastic) variation generated in their relations.  This noise can be "locked" (Arthur 1988) into a next-order systems level as a signal if the communicating systems are able to process the uncertainty with reference to a previous state.  Reflexive systems may then be able to communicate by bouncing the information back and forth, and thus a next-order level of systems can be generated.  Maturana (1978) has called this evolutionary event the generation of "a consensual domain."  Hyper-reflexive systems, however, gain one more degree of freedom which allows them to adjust internally to the further development of the emerging (next-order) systems level, i.e. to develop further according to their own rules.  Such systems are able to self-organize their development given disturbances at lower levels and selective hyper-cycles.

            Why is the distinction between information (uncertainty) and meaning so important?  A systems theoretical conceptualization which is not sufficiently sophisticated may easily lead to semantic confusion and scholastic debates.  The partial perspectives may contradict one another.  The specification of rational expectations requires a rich and free discourse, in which there is no place for "right" and "wrong" on a priori grounds.  One should not fight about speculations, but instead elaborate them into testable hypotheses.  (Of course, testing in this context implies the use of simulations.)

            Luhmann tends not to specify expectations, and to use systems theory without sufficient precision.  For example, Luhmann (1993, at p. 446) recently specified the relations between sub-systems of society as "structurally coupled," and therefore he expects them to be operationally closed.  However, sub-systems belong structurally to a system, and thus they can be operationally coupled within this system.  Additionally, they are structurally coupled to the carriers of the distribution, i.e. in this case the actors involved.  While the latter (structural) coupling is operationally closed, operational coupling between sub-systems requires two kinds of operation, namely one with respect to the super-system (the network) and another with respect to each of the other sub-systems (via the network).  This coupling between sub-systems remains a historical variable dependent upon developments at the system's level.  Consequently, the further development of an emerging operation at an interface, e.g. in the case of technological developments, can never be excluded.


b. the problem of "interpenetration"


            Luhmann (1984) has indeed specified the relations between the social communication system and what he calls "individual consciousness systems" (i.e. actors) as "structurally coupled:" the social communication system cannot operate without individuals who communicate, but only the message-the action-is communicated and not the actor.  Actors and social (communication) systems exchange information through interpenetration.  The interpenetration is an event which can be attributed to the actor as action and to the social system as communication because of the structural coupling between these systems.  As noted, the social system then has its own dynamics.

            In reaction to this redefinition of "interpenetration," Münch (1982/1988) has emphasized with reference to Weber's sociology of religion that "interpenetration" refers primarily to the interpenetration of subsystems among one another (e.g., to the interpenetration of cultural meaning and power in society), and with the social system at large, since particularly this type of interpenetration should be considered as constitutive of Western modernity (cf. Münch 1982, at pp. 480f.; Münch 1988, at p. 204; cf. Leydesdorff 1993b).  In the previous section, I have specified the interaction between sub-systems as "operational coupling."

            Following Weber, Parsons assumed that "interpenetration" (the relations between subsystems of society, and the internalization of cultural and social objects into the personality) can be understood in terms of the same cybernetic relations among all stable systems of social interaction.  For example, as Parsons (1968, at p. 437) put it:


            "The phenomenon that cultural norms are internalized to personalities and institutionalized in collectivities is a case of the interpenetration of subsystems of action, in this case social system, cultural system and personality (...).  Here the critical proposition is that institutionalized normative culture is an essential part of all stable systems of social interaction.  Therefore, the social system and the culture must be integrated in specific ways of their interpenetration."


            The discussion between Münch and Luhmann has revealed that the relations between the social system and the cultural system differ in important respects from those between the social system and the personality.  Although Parsons noted the specificity of interpenetration among systems, he did not distinguish sufficiently among the various kinds of interpenetration.  The relations between subsystems (e.g., the social system and culture) are based not on structural coupling, yet they are structural: the subsystems are expected to be contained within a system.  Coupling is therefore of another nature, namely operational: subsystems are expected to update in relation to one another if this is functional for the system which contains them.

            Münch (1982/1988) defined the social system as the Parsonian action system.  From this perspective the social system operates by actors taking action.  If one additionally accepts Luhmann's distinction between actors and the social communication system, action is in itself already a form of interpenetration.  If in action the subsystems of the social system have additionally to be coupled, the two subsystems have to be made relevant for one another in the same event.  Thus, the communication itself is differentiated, in addition to being a communication in the two structurally coupled systems.  The event integrates the specified internal dimensions of the relevant systems operationally, and the two systems which are coupled structurally; the explanation requires a cross-tabulation.[3]

            One expects all subsystems of society potentially to resonate in all human communication.  For example, the truth of a message, or what it may mean emotionally, is often relevant in the background of a communication.  Thus, a whole distribution of dimensions of the communication should be declared in each action/communication, although a number of these may have a value of approximately zero in specific communications.  If we add subsequently the time dimension to this complex, different frequencies may be involved for the self-referential update within each (sub‑)system.  Furthermore, not all actors are expected to be involved in each update.  In other words, the distributed systems may update with a spectrum of different frequencies (Leydesdorff 1995c).  For example, relatively small economic transactions can have a cumulative impact on change in political power-relations, but the latter may go (temporarily) unnoticed for some of the actors involved.  In such a case, the actors couple structurally with the communication system by acting in one dimension, but their communications fail to couple operationally to a second dimension of the social system.


c. Technological developments as inter-system dependencies


            Interpenetrations among sub-systems of the social system may begin to co-evolve if the signal can be bounced back and forth at the system's level.  Thus, a specific sub-system ("consensual domain") may evolve.  However, the constitution of new sub-systems at the interfaces may change the whole configuration, while sub-systems are supposed to be functional to the further development of their super-system.  Luhmann (1990, at p. 340) has indicated that this may have happened more recently in the case of "technological developments":


            "The differentiation of society changes also the social system in which it occurs, and this can again be made the subject of scientific theorizing.  However, this is only possible if an accordingly complex systems theoretical arrangement can be specified."

            If the categories of the differentiation themselves change historically, one has to attach a suffix with a time indicator to the categories.  Soon a calculus becomes increasingly necessary, one that makes it possible to change both the values of the variables and the categories themselves, for example by declaring the variables (x) as fluxes (dx/dt).  Particularly when studying new technologies, one needs such more abstract models (cf. Blauwhof 1995).  The technological trajectories and regimes (Dosi 1982) can then be considered as consequences of non-linear interactions at the interfaces between the sciences ("supply") and markets ("demand").  Thus, the question of the relations between algorithmic modelling and discursive understanding is most prominent in the discussions about the relevance of evolutionary economics for technology studies (Leydesdorff & Van den Besselaar 1994).

            What does the stabilization of an interaction between functionally differentiated sub-systems mean?  At the level of a single system, stability requires a form of integration.  Indeed, an important condition for the development of modern high-tech sciences seems to be the increasing integration of political, economic, and scientific orientations in research practices (Gibbons et al. 1994).  Integration in the sense of de-differentiation, however, would be evolutionarily unlikely, since the social system would lose its capacity to handle complexity.

            Technological developments can also be considered as the result of inter-systemic resonances which have been stabilized as new functions of the social system during the last century.  The stabilization of interfaces and the construction of integration can then be considered in terms of the sub-dynamics of an emerging higher-order communication system.  This higher-order communication, however, is expected to contain a new epistèmè[4] as it regime: in addition to the communication of substantive novelty and methodologically warranted codification ("truth"), high-tech sciences are able to translate representations from other sub-systems of society into scientific knowledge, and vice versa, to legitimate research results in "trans-epistemic" cycles of communication (Knorr 1981).  In other words, one is institutionally warranted to change the code of the communication (for example, because of a flexible division of labour within the research group).

            What has changed during the last century?  The epistèmè of the modern sciences has been distinguished from those of pre-modern times in terms of its functional differentiation and the universalistic orientation of communications.  The modern sciences have been differentiated as quasi-autonomous sub-systems of society since the 17th century (e.g., Merton 1938 and 1942; Luhmann 1990).  The new systems of reasoning have distinguished themselves from integrated belief system as discursive systems of rationalized expectations.  Differentiation, however, requires at least two levels of communication: substantive novelty ("context of discovery") and metholodogical warrant ("context of justification").  Accordingly, the substance of the communication has both a value in itself and a function for the emerging higher-order system of universalistic communication.  (In the modern sciences, this super-system is also considered a universal system.)

            From this perspective, the 18th century can be considered as the age of the establishment of modern culture and its universalistic discourse (cf. Foucault 1972; Luhmann 1982).  The Napoleonic wars tried to impose the universalistic discourse of the French Revolution onto the European social system.  The early 19th century, however, witnessed the institutional differentiation between civic society ("the political economy") and the various modern states (e.g., Marx 1857; Gouldner 1976).  After the revolutions of 1848 and 1870, the national systems became fully institutionalized and stabilized.

            The "scientific-technical revolution" of the period 1880-1900 (e.g., Habermas 1968; Braverman 1974) can then be considered as an inter-systemic resonance between these functionally and institutionally differentiated sub-systems of the sciences and modern industry.  "Lock-ins" between interacting differentiations are expected to emerge for merely stochastic reasons (Arthur 1988).  These "lock ins", however, occur locally, and therefore their pattern is socially distributed.  Thus, the "wedding of the sciences and the useful arts" (Noble 1977) was not based on a deliberate design, but was rather the unintended outcome of a distributed process (Giddens 1984).

            If these processes can additionally be stabilized into co-evolutions, the resulting patterns can be expected to drive the system into yet higher-order complexities of communication (cf. Maturana 1978; Nelson 1994).  The emerging epistèmè is based on this interaction of the older differentiation with the institutionally organized differentiation between the political economy and the post-Napoleonic state.  After the "scientific-technical revolutions" (1880-1900), however, the patterns of interaction had still to be established.  From this perspective, the history of science and technology in the twentieth century can be considered in terms of the exploration of the potentials for recombining the various sub-dynamics of the emerging system.

            While functional differentiation requires (as noted) the codification of the communication at two levels, the "high-tech" sciences require hyper-cycles of communication in at least three dimensions.  "Triadic" communication systems (ijk) are able to encompass the functionally differentiated ones, and to translate them into one another (ij 6 ik).  Thus, they are able to change the code internally, and by evolutionarily alternating between codes they may develop stable resonances that reconcile what was considered irreconcilable from the perspective of the previous epistèmè.  If the modern epistèmè was based on the geometrical metaphor of a univeral panopticum, the new epistèmè is based on a multitude of partial perspectives, and therefore it requires algorithmic modelling and dynamic representations (e.g., video-clips).

            The emerging patterns of the high-tech sciences are not expected to replace the older models, but to encompass them and to guide their future development.  The next-order regime entrains the trajectories on which it builds (Kampmann et al. 1994).  In other words, the "big science" communities are a result of the "epistemic drift" of translations between economic innovations and research questions; and vice versa, of the possibility to merge fundamental and applied research questions in terms of selections of relevant representations (Elzinga 1992).  These communication systems contain more than a single codification, and additionally they are able to translate between these codifications internally by using a spiral model of communication.  Consequently, developments should be analyzed in terms of processes of representation and communication within relevant scientific-political-economic communities: high-tech sciences develop by communicating in terms of recursive selections on interactively constructed representations (Ahrweiler 1995).

            For evolutionary reasons, one would expect that the regime of the emerging higher-order network of communications will prove to be more viable than that of the less complex ones if specific resonances can be sustained institutionally.  As noted, the emergence of "big science" during the 20th century can be considered as the institutional acculturation of the new epistèmè of science-technology-economy.  The reflexive organization of these institutional patterns in new forms of S&T policies was apparently delayed until the second oil crisis of 1979, when the post-war system entered into a serious crisis at the level of the global economy.  The gradual emergence of stable patterns of scientific reproduction in fields like "artificial intelligence", "biotechnology", and "advanced materials" in the 1980s indicates the viability of the triadic model of communication (cf. Van den Besselaar & Leydesdorff 1993; Ahrweiler 1995; Leydesdorff & Gauthier, forthcoming).  In future work, I hope to elaborate on the co-evolutionary model of this Triple Helix (Etzkowitz & Leydesdorff 1995).





Ahrweiler, Petra (1995). Künstliche Intelligenz-Forschung in Deutschland. Die Etablierung eines Hochtechnologie-Fachs [Artificial Intelligence Research in Germany: The establishment of a high-tech field].  Münster/New York: Waxmann.

Arthur, W. Brian (1988). `Competing technologies,' in: Dosi, G., C. Freeman, R. Nelson, G. Silverberg, and L. Soete (eds.), Technical Change and Economic Theory, pp. 590‑607. London: Pinter.

Attneave, F. (1959). Applications of Information Theory to Psychology.  New York: Holt, Rinehart and Winston.

Bailey, Kenneth D. (1994).  Sociology and the New Systems Theory.  New York: State University of New York Press.

Bateson, Gregory (1972). Steps to an Ecology of Mind.  New York: Ballantine.

Blauwhof, Gertrud (1995).  The non-linear dynamics of technological developments: an exploration of telecommunications technology. Unpublished Ph.D. Thesis, University of Amsterdam.

Braverman, Harry (1974). Labor and Monopoly Capital. New York/London: Monthly Review Press.

Dosi, Giovanni (1982).  "Technological paradigms and technological trajectories," Research Policy 11: 147‑62.

Elzinga, Aant (1992). "The Theory of Epistemic Drift.  A way of relating the social and the epistemic," Revue Roumaine de Philosophie 36: 45-61.

Etzkowitz, Henry & Loet Leydesdorff (1995). "The Triple Helix---University-Industry-Government Relations: A Laboratory for Knowledge Based Economic Development," EASST Review 14, nr. 1: 14-19.

Foucault, Michel (1972 [1969]). The Archaeology of Knowledge. New York: Pantheon.

Gibbons, Michael, Camille Limoges, Helga Nowotny, Simon Schwartzman, Peter Scott, and Martin Trow (1994). The new production of knowledge: the dynamics of science and research in contemporary societies. London, etc.: Sage.

Giddens, Anthony (1979) Central Problems in Social Theory. London: Macmillan.

Giddens, Anthony (1984) The Constitution of Society. Cambridge: Polity Press.

Gouldner, Alvin W. (1976). The dialectic of ideology and technology. London: Macmillan.

Grathoff, Richard, ed. (1978).  The Theory of Social Action.  The Correspondence of Alfred Schutz and Talcott Parsons.  Bloomington and London: Indiana University Press.

Habermas, Jürgen (1968). Technik und Wissenschaft als "Ideologie". Pp. 48-103 in: Technik und Wissenschaft als "Ideologie".  Frankfurt a.M.: Suhrkamp.

Haraway, Donna (1988). "Situated Knowledges: The Science Question in Feminism and the Privilege of Partial Perspective", Feminist Studies 14: 575-99.

Kampmann, Christian, Christian Haxholdt, Erik Mosekilde, and John D. Sterman (1994). `Entrainment in a Disaggregated Long-Wave Model.' In: Leydesdorff and Van den Besselaar (1994), pp. 109-24.

Knorr, Karin D. (1981). The Manufacture of Knowledge: an essay on the constructivist and contextualist nature of science.  Oxford, etc.: Pergamon Press.

Leydesdorff, Loet (1993a). "`Structure'/`Action' Contingencies and the Model of Parallel Processing", Journal for the Theory of Social Behaviour 23, 47-77.

Leydesdorff, Loet (1993b). "Is Society a Self-Organizing System?"  Journal of Social and Evolutionary Systems 16, 331-49.

Leydesdorff, Loet (1994a).  "New Models of Technological Change: New Theories for Technology Studies", in: Leydesdorff & Van den Besselaar (1994), pp. 180-192.

Leydesdorff, Loet (1994b).  "The Evolution of Communication Systems", Int. J. Systems Research and Information Science, 6, 219-30.

Leydesdorff, Loet (1995a).  The Challenge of Scientometrics: The development, measurement, and self-organization of scientific communications.  Leiden: DSWO Press, Leiden University.

Leydesdorff, Loet (1995b).  "The Possibility of a Mathematical Sociology of Scientific Communications", Journal for General Philosophy of Science (forthcoming).

Leydesdorff, Loet (1995c).  "The Operation of the Social System in a Model based on Cellular Automata," Social Science Information 34, Nr. 3 (forthcoming).

Leydesdorff, Loet (1996; in preparation). "The Non-linear Dynamics of Sociological Reflections".

Leydesdorff, Loet, and Peter Van den Besselaar, eds. (1994).  Evolutionary Economics and Chaos Theory: New Directions in Technology Studies. London: Pinter.

Leydesdorff, Loet, and Élaine Gauthier, "The Evaluation of National Performance in Selected Priority Areas using Scientometric Methods", Research Policy (forthcoming).

Luhmann, Niklas (1977). "Interpenetration- Zum Verhältnis personaler und sozialer Systeme". Zeitschrift für Soziologie 6: 62-76.

Luhmann, Niklas (1982). Liebe als Passion (Frankfurt a.M.: Suhrkamp).

Luhmann, Niklas (1984). Soziale Systeme. Grundriss einer allgemeinen Theorie (Frankfurt a.M.: Suhrkamp).

Luhmann, Niklas (1990). Die Wissenschaft der Gesellschaft (Frankfurt a.M.: Suhrkamp).

Luhmann, Niklas (1993). Das Recht der Gesellschaft (Frankfurt a.M.: Suhrkamp).

Maturana, Humberto R. (1978). "Biology of language: the epistemology of reality", in: G. A. Miller and E. Lenneberg (Eds.), Psychology and Biology of Language and Thought. Essays in Honor of Eric Lenneberg, pp. 27-63. New York: Academic Press.

Marx, Karl (1857). Grundrisse der Kritik der politischen Oekonomie.  Moscow: Marx-Engels-Lenin Institute, 1939.

Merton, Robert K. (1938). Science, Technology and Society in Seventeenth-Century England.  Bruges, Belgium: Saint Catherine Press.

Merton, Robert K. (1942). "Science and Technology in a Democratic Order", Journal of Legal and Political Sociology, 1: 115-26.

Münch, Richard (1982). Theorie des Handelns.  Frankfurt a.M.: Suhrkamp; Engl. translation: Understanding Modernity.  London/New York: Routledge, 1988.

Nelson, Richard R. (1994).  "Economic Growth via the Coevolution of Technology and Institutions", in: Leydesdorff and Van den Besselaar (1994), pp. 21-32.

Noble, David (1977). America by Design.  New York: Knopf.

Parsons, Talcott (1937). The Structure of Social Action. Glencoe, IL.: The Free Press.

Parsons, Talcott (1968). "Interaction: I. Social Interaction," in: The International Encyclopedia of the Social Sciences, Vol. 7, pp. 429-41.  New York: McGraw-Hill.

Parsons, Talcott, and E. A. Shils, eds. (1951). Toward a General Theory of Action. New York: Harper and Row.

Shannon, Claude E. (1948). "A Mathematical Theory of Communication," Bell System Technical Journal 27: 379‑423 and 623‑56.

Simon, Herbert A. (1969). The Sciences of the Artificial. Cambridge, MA/ London: MIT Press.

Swenson, Rod (1989).  "Emergent Attractors and the Law of Maximum Entropy Production: Foundations to a Theory of General Evolution," Systems Research 6: 187-97.

Theil, Henri (1972). Statistical Decomposition Analysis. Amsterdam/ London: North‑Holland.

Van den Besselaar, Peter & Loet Leydesdorff (1993). "Research Performance in Artificial Intelligence and Robotics. An international comparison," AI Communications, 6: 83-91.




     [1] This is true unless the matrix is symmetrical, which is an exceptional case that would need to be explained.

     [2] The general concept is "probabilistic entropy", but it goes beyond the framework of this article to elaborate on this concept.  Probabilistic entropies are nested in different layers of communication (cf. Leydesdorff 1995b).

     [3] The two systems (i.e., the social and the psychological one) may internally process the two dimensions in this communication differently.  A single two-dimensional information content of a message can be decomposed in various ways (cf. Theil 1972; Leydesdorff 1995a).

     [4] "By episteme, we mean, in fact, the total set of relations that unite, at a given period, the discursive practices that give rise to epistemological figures, sciences, and possibly formalized systems; the way in which, in each of these discursive formations, the transitions to epistemologization, scientificity, and formalization are situated and operate; the distribution of these thresholds, which may coincide, be subordinated to one another, or be separated by shifts in time; the lateral relations that may exist between epistemological figures or sciences in so far as they belong to neighbouring, but distinct, discursive practices.  The episteme is not a form of knowledge (connaissance) or type of rationality which, crossing the boundaries of the most varied sciences, manifests the sovereign unity of a subject, a spirit, or a period; it is the totality of relations that can be discovered, for a given period, between the sciences when one analyses them at the level of discursive regularities." (Foucault 1972, at p. 191).