The Construction and Globalization of the Knowledge Base in Inter-human Communication Systems

The Construction and Globalization of the Knowledge Base in Inter- human Communication Systems Canadian Journal of Communication 28(3), forthcoming Loet Leydesdorff Science & Technology Dynamics, University of Amsterdam Amsterdam School of Comm uni cations Research (ASCoR) Kloveniersburgwal 48, 1012 CX Amsterdam, The Netherlands loet@leydesdorff.net; http://www.leydesdorff.net/ Abstract The relations hip betwee n the “knowle dge base” and t he “globalizatio n” of comm unication system s is discussed from the perspective of c ommunication theory. I argue that inter-human c ommunication takes place at two levels. At the first level information is exchan ged and provided with meaning an d at the second level meaning can reflexively be communi cated. Human l anguage can be considere d as the evolut ionary achievem ent which enables us to use these t wo channels of communication simultaneously. Providing meaning from the perspective of hindsi ght is a rec ursive operation: a meaning t hat makes a di fference can be c onsidered as kn owledge. If the production of knowled ge is socially or ganized, the perspective of hindsight can furt her be codifi ed. This adds globalization t o the hist orically stabi lized patterns of communicat ions. Glo balization can be expected t o transform the communicat ions in an evol utionary m ode. However, the self-orga nization of a knowledge-based society remains an expectation with the status of a hypothesis. Keywords: s ocial systems, informat i on theory, gl obalization, m ediated, soci ology Introduction A body of literature is currently devoted to the globalization of the economy and the so-called knowledge-based society (e.g., David & Foray, 2002; Leydesdorff, 2001; Mokyr, 2002). How can the concepts of “knowledge based” and “globa lization” be theoretica lly related? In this paper, I argue that communicat ion theory is needed for unde rstanding this relationship. A knowledge base is generated within (for exam ple, scientific) communication when another, that is, global, dimension can be added to the communication. Globalization of the communication operates selectiv ely upon stabilizations that are shaped historically. Language can be considered as an evolutionary achievement that enables us to construct meaning at the supra-individual level. First, th e expected information contained in a message can be provided with meaning by selecting a “signal” from the “noise.” Once meaning has been attributed to the signal loc ally, this can again be communicated. Recursively some meanings can be selected as more meaningful than others by using c odifications of meaning in the communication. A meaning that makes a di fference with reference to a code can be considered as knowledge. In other words, the cons truction of discursive knowledge is a result of using language in increas ingly codified communication. Meaning is provided reflexively, that is, with hindsight. The em ergent order of expectations is constructed and continuously reconstructed on the basis of the communications that have been organized previously. Although a know ledge-based order remains em ergent and historically embedded, the codification in knowledge-based systems of communication provides a basis for the reconstruction of the social system different from its historical construction. The subdynamics of historical c onstruction and reflexive reconstruction operate with different perspectives on time (Urry, 2000). As knowledge production became increasingly organized in scientific discourses du ring the past two centuries (e.g., Gilb ert & Mulkay, 1984; Leydesdorff, 1995; Stichweh, 1984), the knowledge-based perspective has becom e an institutionalized subdynam ic of the social system (Whitley, 1984). From this perspective, the historical configurations provide the communication systems with stabilities needed fo r the m eta-stable or globalized com petition in a next-order layer. This layer, however , can only be accessed through a knowledge-based representation. In order to make my argument that the evolutionary reo r ganization of communication can increasingly operate as a feedback on the histor ical development of the social system, I shall proceed step-by-step. Following Luhmann (1984), I first distinguish betw een action systems (Parsons, 1937) and communication systems as different types of systems. Communication systems process information differently from ac tion systems and therefore can be expected to provide information with another kind of mean ing. In a second step, I distinguish between observed and expected informa tion. By elaborating Shannon’s (1948) mathematical theory of communication into a non-linear dynamics of communication, a more abstract concept of meaning can be defined. The conditions for developing a knowledge-based system of communication can then be specified. First- and second-order observations Whereas Parsons’s (1937) theory of social sy stems focused on actions at the nodes of the network, Luhmann’s (1984) sociological theory of communication, in contrast, focuses on the links that potentially coordinate the networ k into an evolving system. The links can be considered as the operators of a networ k that emerges as a system among the nodes (Barabási, 2002). The links provi de momentary structure to th e network. In social network analysis, for example, this structure is exam ined in terms of dim ensions using factor analysis. Observable actions can be attributed to agencies . The latter have been considered as units of analysis in the sociological re search design. In communication theory, however, one focuses on the communications as operations. Units of operation cannot be observed without a further level of reflection. The communicative operations, for exam ple, may also change the units of analysis. The observables, however, can be analyzed as th e traces of previous communications when the communicative operations are prop erly specified. For example, an institution may have been shaped as the result of a social conflic t. Luhmann (1984, p. 226 [1995, p. 164]) emphasized that “communication cannot be observed directly, only inferred” (italics in the origin al). 1 Some authors have called this reflexive inference “second-order” observations (Bäck er, 1999). Because one needs theoretical guidance for making an inference, a “second-or der observation” can be considered as the specification of an expectation. However expecta tions are, in important respects, different 1 Luhmann (1984: 226): “Die wichtigste Konseque nz dieser Analyse is t: daß Komm unikation nicht direct beobachtet, sonder nur erschlossen werden kann.” 2 from observations. For example, expectations operate in terms of uncertainties, while observations serve the observer main ly for reducing the uncertainty. Consequently, a network system of communicat ions cannot be expected to observe like a human being. The use of anthropomorphic meta phors (e.g., “observer”) m ay easily lead to misunderstandings: not the human observers, but the observations and the observational reports play a key role in the communication. Th e system of reference is different. When an observation is reported, the observation can contribute to updati ng the expectations entertained at the network level. As noted, th is second-order exchange cannot be analyzed without a theoretical perspective. Observations can then be inte rpreted discursively in relation to one another. Expectations--“second-order” or theoretically informed observations--can thus be further specified. If we assume that expectations operate as the substance of co mmuni cation at the level of social systems, how can such a system be re produced? In order to answer the question in terms of a model it is first necessary to ab stract from specific exp ectations. In my opinion, Shannon’s (1948) mathematical theory of comm unication can be useful for answering this question because the concept of “ex pectation” is fundamental to th e abstraction in this theory. In Shannon’s theory, a distribution (e.g., of links and nodes in a network system) contains an uncertainty that can be communicated as the expected inf ormation of the message that this configuration has occurred. The definitions at this level are fo rmal; that is, they do not refer to specific system s of reference. Specification of the systems of re ference requires substantive theorizing, for example, about the role of expectations in inter-human communication. In other words, we have first to specify how meaning is ge nerated in communication and then again communicated and reproduced. Information, uncertainty, and meaning Although sociology and the mathema tical theory of communicati on both focus on distributed systems (e.g., networks), words like “information,” “uncertainty,” or “communication” m ean different things in other inte llectual traditions. Furthermore, the distributions in social systems develop at various levels at the same time. The socio logical re search design therefore needs methodological guidance. However, the m athematical theory of communication has been discredited in the soci al sciences with the argumen t that Shannon (1948) defined “information” abstractly as uncertainty (e.g., Bailey, 1994). This definition appears counter- intuitive, since one tends to associate “infor mation” with a message that informs a system receiving the signal. Fro m the perspective of a receiving system , information can be defined only as meaningful information or “a difference that m akes a difference” (Bateson, 1972, p. 453; Luhmann, 1984, p. 103). The use of two concepts for information has resu lted in a lot of semantic confusion. Let us first specify the diff erence between inform ati on as uncertainty and information that is meaningful to a system receiv ing this information. Shannon detached himself from the implications of his definition of information as uncertainty by stating that the “sem antic aspects of communication are irrelevant to the engineering problem” (Shannon & W eaver, 1949, p. 3). The Shannon-type information precedes the information form ed within a system. 3 The system of reference is not yet specified (Theil, 1972). The message only inform s us that an event (or a series of ev ents) has happened. However, the unit of measurement of the uncertainty thus generated (i .e., bits of information) remains formal and therefore dimensionless. 2 When a system receives this (Shannon-type) in form ation, it may be disturbed and therefore initially become more uncertain (about its environment for example). By processing the uncertainty internally, the receiving sy stem can sometim es--that is, if it contains subs tantive dimensions--discard part of the information as noise. The remainder is then selectively designated as meaningful information. After th e de-selection of the noise, the meaningful information potentially reduces the uncertainty. Brillou in (1962) proposed to call this decrease in the uncertainty “n egentropy.” The meaningful info rmation contains a selection on the uncertainty that prevails. Despite the confusion of the two concepts of information in the litera ture (Hayles, 1990), the biological theory of autopoiet ic systems (e.g., Maturana & Varela, 1984) and the information- theoretical approach (e.g., Abramson, 1963; Thei l, 1972) have been consistent in excluding each other’s definitions of “information” for analytical reason s (Boshouwers, 1997). Biological systems can be considered as “natural ,” and therefore the biolo gist is inclined to begin with the specification of an observation rather than the uncertainty of an expectation. As Maturana & Varela (1980, p. 90) formulated it: Notions such as coding and transmission of inform ation do not enter in the realization of a concrete autopoietic system because they do not refer to actual processes in it. While these authors insisted on the biological realization of “actual processes,” Shannon’s co-author Weaver (1949, pp. 116f .) noted the problem of defining “meaning” from a mathematical perspective when he formulated the following: The concept of information developed in this theory at first seems disappointing and bizarre-- disappointing because it has nothing to do with the meaning, and bizarre because it deals not with a single message but rather with the statistical character of a whole ensemble of messages, bizarre also because in these statistical terms the two words information and uncertainty find themselves to be partners. I think, however, that these should be only temporary reactions; and that one should say, at the end, that this analysis has so penetr atingly cleared the air that one is now, perhaps for the first time, ready for a real theory of meaning. My purpose is to abstract from observers as biological system s and to discuss systems of communication that are able to communicate about meaning and expectations in addition to observations. I therefore follow th e intuition of Weav er and try to define “meaning” first without reference to an y specific “realization.” 2 Shannon (1948) defined the uncertainty or info rmation as equal to the probabilistic entropy of the communication system. Thermodynami c entropy, however, is different from probabilistic entropy because the former is measur ed in terms of Joule/Kelvin, while bits of information are dimensionless. 4 The meaning of information can be defined only with reference to a system that is able to organize the information. The generation of meaning therefore assum es a system operating over time. The specification of a system of reference provides the (Shannon-type) inform ation with system-specific meaning. That is, a system contains a substance in which information is communicated. This substance can also be cons idered as a medium of the communication. A probabilistic entropy is generated whenever the communication system operates and the medium is consequently redistributed. This Shannon-type information can be m easured, but the measurement results could additionally be pr ovided with a substantiv e interpretation if the system of reference were to be specified. For example, information theory can be elabor ated into a statistics for the study of econom ic transaction processes (Theil, 1972 ) or for the study of biological evolution processes (Brooks and Wiley, 1986). Substantive theorizing is need ed for the specification of the relevant system(s) of reference (MacKay, 1956) so that another theory of comm unication --such as one concerning economic exchange relations-- can be generated (Ste inmueller, 2002). The mathematical theory of communication can be used as a formal m ethodology for studying the non-linear dynamics of substantive systems thus specified. Luhmann’s sociological th eory of communication Luhmann (1984 and 1986) distinguished between individual (psychological) systems and social systems. Both social and psychological systems were specified as substantively different from biological systems to the extent that they are able to process expectations by providing observations with meaning. However, when they receive si gnals, these two types of systems can be expected to use different dynamics for their respective update. This is because the concept of “observation” has a differe nt status in these two contexts. The social system can only process information and m eaning in a distributed mode; that is, by communicating it as the expected information content of the communication. This exchange of the expectations provided us above with the second-order domain. This domain is empirical, but reflexive. That is to say the observ ation at this level remains an expectation that can only be made by providi ng a theoretical interpretation to first-or der observations. Thus, the second-order domain emer ges within the first-order and it rem ains referential to it. One is thereby able to obs erve not only what the observers under study are observing, but also how these observers are pr oviding meaning to their observations and communicating about this. Furthermore, when obser vations are attributed as one type or the other, this attribution can be expected to remain uncertain. The attributions can be de- constructed and reconstructed through furthe r communication. The opportunity to distinguish between the exch ange of expectations and the reports of observations enables us to redefine the relatio nship between autopoietic systems theory and information theory at the level of social syst em s. The exchange relations take place at two levels or, in other words, in different dimensions of the communication. These dimensions may additionally interact. Giddens (1979) has called this duality a double hermeneutics, but he juxtaposed the two hermeneutics in terms of participant-observers and external observers. When one appreciates the two levels as dimensions of th e communication, the in teraction terms can also be specified. 5 In a process that communicates both meaning and information, uncertainty is generated in more than a single dimension. Using informa tion theory, the uncertainties in different dimensions can be measured as m ulti-variate probability distributions. As noted, understanding the meaning of the results requires that the system(s) of reference be specified properly. This theoretical specification ha s to be com plex enough to account for the information contained in the messages in the va rious dim ensions. If there are two dimensions, one can analyze both of them and their co-variation. It is important to note that th e specification of uncertainty in terms of a (second-order) system of expectations may become abstract, because a discourse about expectations no longer refers exclusively to concrete (first-orde r) observable s, but also to meanings attributed to these observables. Furthermore, a social distribution of expectations can be provided with m eaning recursively. Recognizing meaning as one among any number of possible meanings opens a horizon of expectations (Husse rl, 1939; Luhmann, 1984, pp. 114ff.). I shall argue below that this next-order relationship of meaning to other possible meani ngs can be distinguished from the historical generation of meaning. Stabilization and globalization The provision of meaning requires an operation over tim e by a receiving system, but this system does not yet have to be stable for any period longer than this single operation. A communication system may also be volatile. In o rder to be identified as a system, it is assumed that the system under study updates in term s of repeatedly distinguishing between signals and noise. In this model, a time axis re presenting selections diac hronically is declared that stands perpendicular to the selections made by the system at each m oment in time (Figure 1). Meaning reduces the uncertainty by selecti ng from incoming information over tim e. The variation over time (or “change”) can be considered as a diachronic outco me of the interactions between variation and selection at different mome nts in time. Selection is a recursive operation. Meaning processing systems can be expected to select som e of the previous selections for stabilization. Since se lections reduce the uncer tainty; stabilization reduces the uncertainty with a next order of magnitude. At a next r ound, stabilizations can further be selected for globalization. Potentially different stabiliza ti ons can again be considered as observable variations, but this would imply the perspective of a next-order selecting system . That is, a distinction between variation and stabilization is ma de. The next-order, or globalized system , is able to select among the results of the first-or der stabilizations. Gl obalization can thus be considered as a second-order selection process. The next-ord er system builds upon the lower-level ones by selecting among them and by potentially rewritin g the previously attr ibuted meanings in terms of their relative weights in the distribution of possible meanings. 6 variation and selection at each moment in time; change and stabilization over time time selections at t = t emerging systemness and stabilization with potential feedback from the organization in the present: goblalization selections at t = t − 1 selections at t = t − 2 uncertainty; disturbances; noise Shannon-type information Figure 1 Selection of variation at each moment in tim e; stabilization of change over time; globalization as a feedback mechanism in the present The recursivity of the operati on of providing m eaning by a next-order system adds another degree of freedom to the system. However, this globalization of the system is not completed, but remains under construction on top of the pr evious stabilizations. Stabilization and globalization can be considered as differe nt subdynamics operating on each other. The constructing system comes under th e selection pressure of its global dim ension as specific constructions are historically realized among the other possible constructions. As a system is shaped historically along the time dimensi on in a forward m ode, global meaning can be attributed only with hindsight. This is achieved by comparing the historical system with other envisioned possible systems. This operation is kn owledge-based because the alternatives were not yet all realized. However, the next-o rder selection operates upon their temporarily stabilized representations in the present. A reflection operating with reference to an instance invoked in the present (x t ) can be distinguished from the recursive update of m eaning operating with reference to a h istorically previous state of the system (x (t-1) ). Dubois (1998) proposed to ca ll this recursion on a present state of the system an “incursion.” An incursion evaluates a representation from a next-order perspective. However, the operation remains histor ical in the evolution of the system, and an incursive system can therefor e be expected to develop re cursively along the time axis. “Incursion” occurs within systems under historical conditions; th at is, as an empirical relation to recursively constr ucted and reconstructed trajectories. An incursive system is able to select among its current representations of the past in terms of its future “survival value” in a next-order se lection environment (Rosen, 1985). Incursion and recursion can be expected to intertwine as m odes of communication in a social system that operates in terms of expectations . Analytically, however, incursi on p rovides the social system 7 with a sub-dynamic different from historical recursion since the incurs ive system operates on a representation of the system in the present. Because of the ability to select among a variety of possible representations of the system , an incursive system can learn to “anticipate” possible further developments and, therefor e, become increasingly “self-organizing.” Additionally, a meaning-processing system can construct (knowledge-based) representations that compete with thos e previously generated. Unlike artificial systems, social systems remain histor ical and therefore ref erential to first- order observations. The virtual ope ration of the global dimension can be expected to interact with other dimensions of the sy stem . Because of this interacti on with its history, the meaning- processing system can be expected to fail empiri cally to se lf-organize. Here, self-orga nization remains only a subdynamic of the historical syst em . The social system can be expected to exhibit a mixture of contingencies of its constr uction in history and its next-order incursions, upon these manifestations. Analytically, howev er, a double-layered structure sustaining the exchange of both information and meaning is requ ired before the “globalization” of a system of expectations can become historically important. At the first level, variation is selected and then selections can be chosen for stabilization so that the structural advancements of the netw orks can also be reta ined historica lly. The stabilizations function evolutionarily as rete ntion mechanism s that have been shown to provide “survival value” for the configurations that were rea lized. At a second level, these historical configurations can fu rther be selected as second-or der variations by another (next- order) dimension of the system. This globalizi ng operation of the system remains “virtual” from a first-order perspective (Giddens, 1984) because globali zation incurs on the observable instantiations of the system using another dime nsion of that system . The additional dimension is spanned in terms of the dist ribution of observable instances. Codification and the change of meaning Whereas positive theo ries are able to specify observable variation (at each m oment in time) and change (over time), the specification of a “s election” requ ires an analytic perspective. Selection is a negative operat ion: it reduces the variation. In that a negative operation cannot be directly observed, a theoretical inference is needed. Stabilizations can again be observed. The negative sign of the selection can be expect ed to lead alternately to “observable” and “expected” events with each consequent turn. Stabilizations can be disti nguished analytically from lowe r-level variations when the selections that preceded the former are analy tically specified. Globaliz ation does not require a single stabilization since a great variety can be selected from . With hindsight one can ask whether an identified unity has remained th e same despite ongoing processes of change. However an identifiable system can be expect ed to develop along an observable trajectory, while under selection pressure from a next-order “regime” (Dosi, 1982). The next-order regime remains pending as a late nt form of selection pressure on the systems and subsystems upon which it rests. By attributing an analytical identity to this next-order system the analyst would reduce its comple xity by choosing a perspective. The self- organization of the social system cannot be identified em piri cally because it can be expected 8 to remain uncertain and therefore other appreciations remain possible. 3 Thus, the identification of a global system is analytical: th e self-organization of a social system can be specified only as a hypothesis. The questi on then becomes whether entertaining this hypothesis adds to our theo retical understanding. In other words, the self-organization of th e social system beyond its stabilization, is an intrinsically knowledge-based assumption. What is observable provides us with fragments of the global system, which can only be appreciated as instantiations on the basis of entertaining the hypothesis of other options. In this cont ext, Luhmann (1984) proposed to distinguish among three representations of the social system relevant for pur suing analysis: society as the global system of communication, the historical organization of communication, and concrete interactions. By using social network analysis, interactions can be analyzed as be ing organized by latent dimensions (Burt, 1982; Lazarsfeld & Henry, 1968) . This organization of the interactions can be appreciated by a reflexive actor. However, th e further self-organizatio n of the interactions by a next-order level of global organization assu m es another dynamic at the structural level. The flux in the latent dimensions may be expe ri enced as counter-intuitive by inf ormed and reflexive actors. The interpretati on can then only be knowledge-based. Knowledge based, as for example in the case of a child who knows that its parents entertain a level of communication to which it has no acce ss. Here the child m ay begin to understand that this next-order level co mmunication can be expected to redefine its own situation beyond its control or understanding. Wh ether the fam ily is stabili zed or destabilized by the communication between the parents remains stru cturally beyond the children’s control. However, the situation can be reflected in a th erapeutic setting. A globalized system of communications self-org anizes the interactio ns and organizations subsumed under it by selecting (in the present) using the hypothesized degree of freedom for a next-order reflection. By specifying and ente rtaining this hypothesi s, one is able to distinguish analytically between a system deve loping its complexity hi storically, that is, along a trajectory, and a self-o rganizing system that has one more degree of freedom for adjusting to its environments from the perspe ctive of hindsight. W hen this additional degree of freedom is used reflexively within the sy stem as a dimension of uncertain communication, a meaning-processing system is generated on top of the inform ation-processing one, on which it reflects. In a meaning-processing system, m eaning can be changed by being communicated. An interaction term is generated that may provide feedback on the definitions. The first-order exchange of information is provided with mean ing that can further be exchanged at a next- order layer, and this potentia lly changes the system. Whereas the Shannon-type information provided us only with distributio ns and differences, “the differe nce that makes a difference” can then be communicated as “meaningful info rmation,” i.e., an interaction term between meaning processing and information processing. 3 An identity can perhaps be defined as a codi fied and, therefore, symbolically stabilized system that is able to entert ain its relation to its own next-o rder system reflexively without loosing stability. 9 Language and culture I submit that language can be considered the e volutionary achievem ent that enables us to attach meaning to the exchange of inform ati on and to exchange meaning at the same tim e (Luhmann, 1984, p. 208 [1995, p. 152]). The analytical distinction is cont inuously blurred in observable practices because language enables us also to operate informally and in a network mode. A message can be provided with meaning, and the meaning can be distinguished from the expected information contained in the m essage. Thus, through the linguistic ability communication processes are doubled into two interacting layers within the exchange. This occurs however, in an uncertain mode. The differentiation between the two channels of communication--uncer tain inform ation and meaning--is not “hard-wired.” Inter-human communication can therefore b e expected to remain prone to failure. Where codification help s us to keep the various m eanings separate, the uncertainty introduced by this “double herm eneutics,” allows both for formalization and for informal communication. The two layers can be expected to interact continuously. In other words, one can always consider mean ingful what has been previously communicated as uncertainty, and one can discuss and dec onstruct discursively what previously was codified. The exchanges of meaning and un certainty co-vary in the communication of “meaningful information.” The various exchange processes can be expected to generate uncertainty (i.e., introduce ne w variation) by interacting. Participants and observers are able to take part both in the hi storical construction and in the recursive feedback of providing meaning to the reconstructions. This occurs at potentially different times, in different places, and to diffe rent degrees. When linguistic capacities enable agents to counteract natural constraints at the supra-individual le vel by constructing and maintaining a specific codificat ion as a cultural feedback loop, a double herm eneutics is implicated reflexively (Giddens, 1976). The possi bility of alternating perspectives between the role of participant-ob server and external observer, provides the basis for improving one’s analytical understanding of the reflexive operation of inte raction (Parsons, 1968). Two (or more) perspectives for the selection can be entertained and traded-off. A biological system has to adapt--at the risk of extinction--because the s election pressures are “natural.” Selection environments, however , can already vary am ong ecological niches. Biological exchange relations can therefor e experience additional selection pressure. Moreover, the two types of selection pressures cannot be distinguished by the agents living in these niches unless these carrying agents have th e reflexive capacity to entertain concepts for mapping their situation. When these concepts can be communicated, a semantic domain is generated operationally on top of the consen sual domain (in which the agents rela te structurally) (Maturana, 1978). The system of communications can then become self- organizing and autopoietic in the sense of in creasingly controlling its own reproduction. This happens in biological cultur es like insect populations. Human language extends the biological concept of a semantic domain because order is not constructed in a biological envi ronment and then stabilized, bu t remains f lexible and under construction as an expectation of order. This co nstructed order can also be changed by a next- order system or at a next moment in time, i. e., by adding reflexively a new dim ension to the 10 system. Note that “reproduction” in this cas e no longer means reproduc tion in the biological sense, but rather the further developmen t of the communication through the system’s previously organized retention mechanisms. The lower-level system s upon which the next- level systems build, can be innovatively recons tructed given the additional degree of freedom provided by linguistic exchanges. The represen ted and the representing systems may thus begin to co-evolve. The number of reflexive couplings maintain ed between the layers, may perhaps be considered as the evolutionary variable because it lim its the complexity that can be pr ocessed by the communication. A meaning-processing syst em, for example, can be expected to evaluate its options also in an anticipatory mode because it can entertain and exchange different representations of itself. A knowledge-processing system like a scientific communication system can sometimes reconstruct the system under study--to the extent that one is also able to intervene techno logically in what is repres ented. Translation and knowledge generation The globalized super-system can be expected to remain latent, but as a next-order selection mechanism it opera tes with the resilience of a regime. The complex system can be represented only in terms of its “instantiations” or equivalen tly--along the time dimension--its trajectories. The codes of the communication at the level of these subsystems tend to close the operation by attributing specific meaning to the (Shannon-type) information, but this closure can be altered when exchanges among diffe rent codes become historically possible at a next-order level. The changes in meaning due to the exchanges at the level of codes, can be considered “translations” of the information. Tran slations enable us to change the meaning of an uncertainty from one system of reference to another. Through translation, discursive knowledge is generated endogenously because one h as to distinguish among possible meanings in different contexts. For example, when one discusses energy shortages in political deba tes, a physicist would first have to translate this concept of “energy” as a shortage of energy carriers (e.g., oil or gas) because “energy” is defined in physics as a conserved entity. Just as in the co ntext of physics there can never be a sho rtage of energy, words often have different meani ngs in different cont exts; yet if one is knowledgeable about the differences between the codifications, transla tion becomes possible. The differentiation of the codes possesses an evol utionary function that potentially furthers the development of communication systems. Transl ations of m eaning within the system from one code to another, provide an internal m echanism for the regenera tion and transformation of the organization of meaningf ul expectations. The trans lation mechanism can theref ore be made “functional” to the “reproduction” of th e knowledge-base of the so cial system. Novelty is generated when new representa tions emerge from the innovative recom bination of codes. For example, technological innovation can be co nsidered as the successful generation of new combinations of “demand” and “supply.” Wh ile “demand” is mainly codified in term s of markets and users, the “supply” side is organized in terms of technologies--and the latter m ay in turn be based on specific combinations of insights from different specialties. The stabilization of innovation into a new system potentially reorganizes the system s from which 11 it was generated. The emergence of the Internet can be studied in these terms, as can the railway systems in the 19 th century (Urry, 2000). Whereas mutation prevailed as a stochastic m echanism for generating variation in biological systems, it is selective and re flexive interactions am ong (sub)sy stems of the social system that generate second-order variation. Each hist orical stabilization with in an evolving system will be continuously disturbed by interactions with other parts of the system . When disturbances can be provided with meaning at various systems leve ls, the historica lly embedded codes are expected to ch ange gradually. The recursivity in this process of reflexive refinement (locally) improves the system a nd generates knowledge endogenously. Meanings which are functional are distingu ished from those which are not (or no longer) functional for the reproduction of the communication. Solutio ns to puzzles can be communicated as potential innovations and then selectively codified in a next round (Kuhn, 1962). To paraphrase Bateson (1972), this endogenous generation of discursive knowledge can perhaps be defined as the exchange of “mean ing which makes a difference.” W hen repeated over time, the different selections generate coupl ings both horizontally and vertically within the knowledge base of the system (Simon, 1973). Some couplings (and not others) are stabilized and provision ally locked-in (Arthur, 1994) form ing communication channels that can be considered as codifications of pr evious communications. Communication along a provisionally constructed and stabilized channel can be recei ved as signals by specific system(s) whereas higher-order languages, such as scientific discours es enable us to refine the distinctions of signals from noise (Luhmann, 1990). When a coupling is provisionally stabilize d, a next-round of discur sive reflection and deconstruction may enable us to reconstruc t the system under study and perhaps renew it by searching for a solution to the puzzle different from the ones generated “naturally” or at previous moments in time. In Holland, for ex am ple, a polder vegetation can be considered “natural,” while the polder as a technical system of water managem ent remains artificial. The social system is increasingly ab le to replace its historically given base with an evolving knowledge base that operates analytically independent on the histor ical organization of meaning. The discursive reconstr uction of codification in rele vant exchange processes (e.g., markets, sciences, etc.) enables us to decons truc t an interface in term s of its composing (sub)dynamics and then sometimes reconstruct it m ore efficiently by shaping technological artefacts. This technological r econstruction of the natural base can be reinforced at the global level and therefore develop be yond the control of individual or group intentions (Marx, 1857). (Sub)systems shape one another historically along the time axis by providing meaning to their mutual disturbances. The reconstructions increasing ly invert the time axis in an evolutionary manner, from within. This proc ess uses another (“incu rsive”) dynam ic of the communication. As the interactions gain syst emic momentum, the em erging patterns guide the selections with hindsight because the constructs becom e structured as another network layer. The global perspective, although evolut ionarily emerging, m ay eventually begin to drive the historical development of the syst em into a self-organizing mode. The “phase space” of alternatives provided by globalization offers possibilities for restructuring the system. This development is knowledge-based a nd selective because some of the envisioned possibilities are no longer “naturally given.” 12 The globalized system can be hypothesized as se lecting upon historical manifestations that are recognized in the present. The representations refer to expectations th at can be entertained about other (“adjacent”; cf. Kauffman, 2000) possi bilities. Globalization can be considered as an evolutionary process that inverts the time ax is by opening and then al so organizing a space of envisioned possibilities from which one is able to select incursively (Van Lente and Rip, 1998). The theoretical representations remain c onstrained by the manifestations because one cannot select upon what has not yet been envisioned. Regimes and identities Globalization provides an additi onal dimension to the social system, which can only be accessed by the individual through p reviously stabilized systems of communication. The individual in an increasingl y knowledge-based society has the opportunity to trade-off between stabilization and globali zation. At moments when the two operations can be mapped onto each other, an identity can be temporarily maintained. At other moments, one interacts with the social system, and a next-order sel ection may rem ain external to the individual. Thus, next-order selection by th e social environment and intern al reflection at the individual level compete in providing meaning. Both autono my and inclusion are part of one’s personal development in these complex dynamics (Weinstein and P latt, 1969). The two systems (the individual and the social) can be considered as structurally coupled as they provide complexity to each other (Maturan a & Varela, 1980). However, the axes of the systems stand perpendicular to each other, and this difference can be m ade reflexive within both systems. Whereas the indiv idual is consciously intereste d in maintaining identity , uncertainty can be expected to prevail in the network mode of communication. The network improves by developing discursive knowledge, whereas the individua l can feel alienated from ongoing processes of codification in subsystems of social comm unication in which one is not or only partially included. The ne tworks generate meanings and discursive knowledge at the supra-individual level, whereas the individual seeks to reflec t on these meanings from an identifiable perspective. The social system reflects the complex dynamics of each individual, but in a distributed mode. The interactions among mutual expectat ions potentially add another dynam ic to the aggregate of individual expect ations. And as the human carrier s continue to communicate in terms of information and m eaning, this commun ication system becomes increasingly shaped so that it enables them to process two or even more dim ensions of the communication at the same time without continuously becoming confused. Thus, the development of the communicative capacities of the carriers promotes the furthe r differentiation within, and codification of, the social (s ub)systems, and vice versa. As long as the actors operated on the basis of a single (e.g., cosmologically given) meaning at the level of the social system , this system could still be consid ered as an identity. As the actors increasingly deconstruct and recons truct meaning within communication, the difference between the communication of uncer tainty and meaning can be reinforced by codification. When this diffe rentiation can further be stab ilized--by communicating through symbolically generalized media for exam ple--one can expect the social system to becom e globalized. 13 As a historical phenomenon, the globalization of th e social system remains part of the social system as one of its subdynamics. The super-sy stem fails to exist physically or biologically, since it is functionally defined as a knowledge -based operation, one of changing what can be observed. It can only be considered as “real” in the sense of critical realis m while it exhibits itself in terms of its continge nt instantiations (Bhaskar, 1997; Sayer, 2000). The globalized super-system, however, can be considered as another subsystem--one that modulates local communications--in terms of, for example, these communications’ sym bolic value (Leydesdorff, 1993). For analytical reasons, a reflecting communi cation must be distinguished from the communication on which it reflects. Thus, the soci al system is in need of a medium that enables us to communicate in terms of both information exchanges and in terms of the meaning of this exchange. These two dimensions have not been considered here as different systems, but as differentiations of the linguistic operator: the tw o layers are not “hard-wired.” The distinctions remain reflex ively constructed, although such di stinctions can be codified historically and stab ilized for considerable periods of time. For example, social meaning can be codified at the supra-ind ividual level such as in the case of the development of a scie ntific paradigm . The differe ntiation between common-sense knowledge and scientific (i.e ., potentially counter-intuitiv e) knowledge can be made functional to the further devel opment of the individual and/or the society, but again along different axes. In a scientific discourse, the two dimensions (the embeddedness in a social group as an individual and th e cognitive partaking in a scientific development by providing contributions) can no longer be expected to co incide as system s of reference (Gilbert & Mulkay, 1984; Leydesdorff, 1995). Thencefor th, the problems of differentiation and integration increasingly become puzzles for th e social coordination and for the individuals living in a knowledge-based society (Bernstein, 1995). Meaning at the level of the social system is t hus different from m eani ng at the level of the individual. The social level can be considered an outcome of the interactions among individuals. This emergence of social meaning has also been called “situational m eaning” in symbolic interactionism (Blum er, 1969; Kno rr-Cetina & Cicourel, 1981). Hum an carriers are increasingly able to distinguish reflexively between the social mean ing of a comm unication and their personal meaning. This differentiation can be reinforced as the difference between personal (or tacit) knowledge and discursive (i .e., potentially counter-in tuitive) knowledge (Leydesdorff, 2000). Social order can be entertained as an exp ectation that is informed by the m anifest institutionalizations of the system (Giddens , 1984). Social order, however, consists of expectations being exchanged among individuals realizing their life-cyc les in interaction with the contingencies of their biol ogical bodies. Within the life-cy cle of an individual certain problems have to be solved: real-life condi tions place constraints on the differentiation between possible meanings and the distribution of events (Haberm as, 1981, 1987). At the level of a social system, this na tural constrai nt can be relaxed: diffe rent solutions can be achieved by various subsystems. The function of differentiation can be expected to rem ain uncertain, or in flux, more than in the case of a biological system. As the social system becomes further differentiated into various subsystem s, discourses and codes (such as between an economy, scientific discourses, health care sys tems, etc.)--- the 14 carriers are burdened with making translati ons among the specific languages. Furthermore exchange media must be developed and potenti ally globalized in each of these subsystem s. Reflexive discourses enable the carriers to make these translations of meanings in various ways. As they are able to communicate about their respective solutions, for exam ple, using public or symbolic media, they may subjectively envisage their historic al solution as a global solution. However, with hindsight, each solution at the level of the social system can be recognized as the globalization of a specific medium (e.g., a television channel like CNN). Globalization of the social system can be expected to be differentiated into a range of local solutions that have been provisionally stabiliz ed. The social system continuously loops back into itself (e.g., between local and global) sin ce no informal transgress ions are intrinsically forbidden. A globalization represented at a specif ic moment in time will be perce ived in the future as an instantiation of this subdynam ic because th e globalization itself remains dynam ic. The dynamic perspective contains more com plexity th an one is able to observe of any discrete instance. This more comple x system of comm unications builds on its subdynamics by selecting from them, but not in any a priori prescribed order. In a ddition to institutions, the observable manifestations include rules and re gulations as the dynamic counterparts of the institutions (Giddens, 1979 and 1984). These units of analysis can be considered as the footprints of the communications that have served us hitherto. The organization of communications develops along trajectories that have been in stitutionalized and codified for historical reasons. Institutionalization and stabili zation are historically observabl e, whereas codification can be considered as providing meaning to some m eani ngs but not to others. Therefore, the latter operation is part of the incursive process of cu ltural evolution. As noted, this process inverts the time axis and highlights the construction of meaning within the system as a future- oriented operation that takes plac e in the present. By providing new meaning to its previous instantiations, the sys tem is increasingly able to rewrite its history; for exam ple, by making new selections from the pers pective of hindsight. As the new meanings are increasingly science-based, the rewritings can become in creasingly technology-ba sed. The system atic organization of knowledge production (e.g., in R& D laboratories) drives the social system into its globalization. Here I have followed Luhmann in defining the soci al system as structur ally coupled to hum an agency. And whereas the social system is ge nerated from and based in the network layer among the carriers at the nodes, the links themselves are not stable. That is; they can be considered as operators that induce change . Virtual communications among machines can then also be constructed using specific media of communication as next-order codifications-- for example, into scientific and technical di scourses. However, these virtual comm unications have to be provided with meaning locally, sin ce otherwise they would lose their historical significance. Thus, the evolutionary metaphor and the historical metaphor are two sides of the sam e coin in the case of the social system’s operation. Th e co-evolution of incu rsive and recursive subroutines drives the developm ent of the knowledge base as an emerging dimension of the social system. Geometrical metaphors reduce the algorithmic complexity in the systems under study by providing us with perspectives that can provisionally be stabilized by codification. The codes of the communication span a universe that processes complexity, 15 while the interfaces between th e social and the in dividual ground the communication. This structural coupling sets the discursive stage for what can historically be considered as “natural,” “social,” or “virtu al.” The cap acities to channel inform ation selectively continuously challenge the (sub)systems and their interactions to explore innovative recombinations of meaning. Communication in the knowledge base Methodologically, one is able to measure co mmunications in term s of changes in the distributions of what is exchanged. By defin ition, a redistribution generates a probabilistic entropy. Using the abstract concept of “probabi listic entropy,” “varia tion” and “selection” can be considered as two geometrical meta phors for the study of the sam e (algorithmic) process: the observable variation provides us with a historical description of the evolutionarily deselected cases. In ot her words: the concepts of probability and probabilistic entropy provide us with a common denominator for “selection” and “variation” when using the mathematical theory of co mmunication. This reflects a pers pective very different from that of (Darwinian) evoluti on theory in biology, where thes e two operations have been considered analytically independent. An empiri cal event can be expected to occur (with a probability); yet the occurrenc e of events remain s uncertain b ecause of selection pressure. At each instance, randomness, or variation, inter acts with determinati on (or selection). In mathematics, one can deduce a consequence by making an abstract in ference, but one can no longer expect to be completely certain (nor completely uncertain) when studying empirical operations. The theoretical specific ation of the substances of the comm unication enables us to select the relevant dimensions from an ab stract phase space of possible dimensions. A specific theory of communication can consequently be generated in each dimension that can be specified as relevant. The mathematical concepts are thus provi ded with substantive meaning. When it is assumed for example, that atom s are exchanged among molecules, chem istry can be developed as a discourse that refl ects upon these exchanges (Mason, 1992). When molecules are exchanged, biology can be considered as a discourse that is able to appreciate these processes in theoretical terms (Maturana, 1978). Wh en m eaning is communicated, psychology and sociology provide us with releva nt reflections (Luhmann, 1986). These latter two discourses refer to different systems of reference, nam ely individuals and the coordination mechanisms am ong individuals, re spectively. Because of th e differentiation within the social system, refl exive study of this system--sociology--can be expected to proliferate its discourses e ndogenously (Leydesdorff, 1997). The substantive interpretations of the various subdynamics provide us with metaphors. These metaphors can be thought of as geometrical stabili zations that are useful in the codification of a specific type of specialized communication (Hesse, 1980). Th ey provide analysts with a structure that can sometimes be stabilized (i nto a scientific discour se) and perhaps even globalized (into a paradigm). As substantively specific, the us e of m etaphors enables us to handle the algorithmic complexity in the data without beco m ing continuously confused by the ongoing processes of cha nge at various levels. 16 Whenever a variation occurs that is not comp letely random, one selection or another can be hypothesized, since the variation co uld have been different. Whethe r the result is perceived as an observable variation or as the result of a (hypothesized) selection depends on the perspective of the analyst. Thus, the concepts of “varia tion” and “selection” can be considered as two geometrical perspectiv es on the algorithmic operation of generating probabilistic entropy. The system of reference for the va riation is different from that for the selection. In classical ev olution theory, for example, variat ion occurs at the species level, while selection is attributed to nature as a super-system . The notion of geometrical metaphors used for communication brings language into play (Hesse, 1988). The perspectives can be considered as codified at the supra-individual level, that is, by using one selection or another for the stabilization of the complex dynamics into a discursive (that is, geometrical) represen tation. This representation can contain an instantiation at one m oment in time or entail the developm ent of a trajectory over tim e. However, an information calculus is required for the specifica tion of the probabilistic operation itself (Bar-Hillel, 1955). The synchronic representation of an instanti ation highlights the complexity, whereas the diachronic representation of a trajectory focuses on the dynamics. The complex dynam ics, however, develops beyond these representations in terms of fluxes. This global level of communication can be accessed in terms of expectations on the basis of the codified knowledge contained in the representations, for example, with the help of an algorithmic simulation model. In the algorith mic m odel, the representations are recursively repre sented as subroutines (Leydesdorff & Dubois, 2003). Conclusion Human language first enables us to codify the relationship between uncertainty and the meaning of a message. As m eaning is genera ted interactively by us ing language, it can be considered as a reflexive function of language at the level of the social system (Pask, 1975). I have here deviated from Luhmann’s social systems theory by considering language as the “first-order” operating system of society. Languages spontaneously emerge among hum an beings as “natural languages” (Pinker, 1994); ho wever languages can also be cultivated using higher-order codifications, for exam ple, in scientific discourses. Luhmann in 1984 proposed that “meaning” was itself the operator of social systems. “Meaning” was therefore considered as a kind of transcendental precondition of both reflection and language, but not as a consequence of the lingu istic operation itself (Luhmann, 1971). Luhmann had not yet really reflected upon the idea of meaning from a more abstract perspective such as the mathematical th eory of communication, or non-linear dynamics. Luhmann’s focus on using historical examples implied that the globalization of meaning was considered a virtual operation that could only be discussed with reference to Husserl’s transcendental phenomenology. The social construction of meaning was anal yzed as thoroughly contingent, by scholars working in the pragmatist tradition and in sy mbolic interactionism (Blumer, 1969; Mead, 1934). The exchange of meaning provides communi cation with a second layer that reflects and interacts with the first layer of uncertain exchanges in the social system. I propose that 17 language can be considered as the evolutiona ry achievement that facilitates communication by using these two channels simultaneously. This duality enables us to capture the complexity of the algorithmic opera tion, a problem having two sides: an a priori expe ctation and an a posteriori update value. Meaning draws in the present upon the information exchanged when it “incurs” on the informati on. It is also conti nuously generated as a recursive (i.e., historical) update of previous meaning(s). The difference between these two operations can be observed in soci al systems since meaning can be codified at different levels of the same system. The organization of scientific knowledge production for example, has led to the differentiation between the context o f justific ation and the context of discovery (Popper, 1935). Knowledge produced in th e context of discovery can be recognized as a knowledge claim when submitted to a journal for furthe r communicatio n. Accordingly, the two contexts provide different meanings to the same knowle dge and distinctions in the epistem ological status of this knowledge (Gilbert & Mulkay, 1984). The meaning of a communication can be changed with hindsight, by being exchanged. It can temporarily be codified and historically stabilized, as for ex ample, in social institutio ns. The incursive subroutine, however, doubles the historical authentic ity of the social system by providing it with an evolutionary layer that is future-or iented in the present. Thus the complex social system as the subject of cultura l evolution is able to reconstruct its history continuously, and from different perspectives, wh ile operating reflexively and in a distributed mode. Because of this distributed mode, the reconstruction cannot be completed without a further selection. Through translation by a rece iving discourse, a prev ious codification can sometimes be stabilized since the m eaning may have been rewritten as hi storically specific. The appreciation of incursion within the recurs ive routines enab les us to reconstru ct the historical dimension of these processes. In other words, the system re-doubles in the representation because the linguistic operation itself is layered. A knowledge-based system can entertain more than two representations beca use what was first considered as levels can also be considered as dimensions of codificat ion along different axes. Over tim e--yet another dimension--the operation is neither completely h istorical nor completely reconstru ctive: it contains uncertainty also in the time dime nsion. Cultural evoluti on, once unleashed as an interaction among these degrees of freedom, dr ives the communicative competencies of the carriers at an increasing speed since the social system provides us with ranges of options for realization at each reflexive turn. While reflection is often associated with a turn of 180 degrees, the addition of another dimension to the social system assum es a tu rn of only 90 degrees. Interaction terms among these perpendicular dynami cs of reflection can lead to non-li near recombinations of meaning. However, the number of possible combinati ons--and therefore the selection pressure-- increases exponentially with the number of relevant dimensions . As the awareness of these dimensions is absorbed, the carriers can experience this pressure as globalization of the social system. Insofar as the dimensions can be specified as hypotheses, they can be m ade the subject of discursive reasoning. The analytical specification of these systems of reference can be expected to become crucial for th e further developm ent of knowledge-based communication systems. 18 • The author is grateful to Gretchen M. Dee and Iina Hellsten for detailed comments on a previous draft. References Abramson, Norman. (1963). Information Theory and Coding . New York, etc.: McGraw-Hill. Arthur, W. B. (1994). Increasing R eturns and Path Dependence in the Economy . Ann Arbor: University of Michigan Press. Bäcker, Dirk, Ed. (1999). Problems of Form . Stanford: Stanford University Press. Bailey, Kenneth D. (1994). Sociology and the New Systems Theory . New York: State University of New York Press. Bar-Hillel, Y. (1955). An Examin ation of Information Theory, Phil. Sci. 22, 86-105. Barabási, A.-L. (2002). Linked: The New Science of Networks. Cambridge, MA: Perseus Publishing. Bateson, Gregory (1972). Steps to an Ecology of Mind . New York: Ballantine. Bernstein, Richard B. (1995). The New Constellation: The et hical-political horizons of modernity/postmodernity . Cam bridge, MA: MIT Press. Bhaskar, Roy (1997). On the Ontological Status of Ideas, Journal for the Theory of Social Behaviour 27(3), 139-147. Blumer, Herbert. (1969). Symbolic Interactionism . Perspective and Method. Englewood Cliffs: Prentice Hall. Boshouwers, Stan. (1997). In-Formare: De Wereld van het Kensysteem, Unpublished MA Thesis, Cognitive Artificial Intelligence, Utrech t University. Brillouin, Louis. (1962). Science and Information Theory . New York: Academ ic Press. Brooks, Daniel R., and Wiley, E. O. (1986). Evolution as Entropy . Chicago/London: University of Chicago Press. Burt, R. S. (1982). Toward a Structural Theory of Action . New York, etc.: Academic Press. David, Paul A., & Dominique Foray. (2002). An Introduction to the Economy of the Knowledge Society. International Social Science Journal , 54(171), 9-23. Dosi, Giovanni. (1982). Technol ogical Paradigms and Techno logical Trajectories: A Suggested Interpretation of the Determinan ts and Directions of Technical Change, Research Policy, 11, 147-162. Dubois, Daniel. (1998). Co mputing Anticipatory Systems with Incursion and Hyperincursion. In Daniel Dubois (Ed.), Computing Anticipatory Systems, CASYS-- First International Conference . AIP Conference Proceedings 437 (pp. 3-29). Woodbury, NY: American In stitute of Physics. Giddens, Anthony. (1976). New Rules of Sociological Method . London: Hutchinson. Giddens, Anthony. (1979). Central Problems in Social Theory . London: Macmillan. Giddens, Anthony. (1984). The Constitution of Society . Cambridge: Polity Press. Gilbert, G. Nigel, & Michael J. Mulkay. (1984). Opening Pandora’s Box. A Sociological Analysis of Scientists’ Discourse . Cambridge: Cambridge University Press. Habermas, Jürgen. (1981). Theorie des kommunikativen Handelns . Frankfurt a.M.: Suhrkamp. Habermas, Jürgen. (1987). Excursus on Luhm ann’s Appropriation of the Philosophy of the Subject through Systems Theory. In: The Philosophical Discourse of Modernity: Twelve Lectures , pp. 368-85. Cambridge, MA: MIT Press. [ Der philosophische Diskurs der Moderne: zwölf Vorlesungen . Frankfurt a.M.: Suhrkamp, 1985.] 19 Habermas, Jürgen, and Niklas Luhmann. (1971). Theorie der Gesellschaft oder Sozialtechnologie . Frankfurt a.M.: Suhrkamp. Hayles, N. K. (1990). Chaos Bound; Orderly Disorder in Contemporary L iterature and Science. Ithaca, etc.: Cornell University. Hesse, Mary. (1980). Revolutions and Reconstructions in the Philosophy of Science . London: Harvester Press. Hesse, Mary. (1988). The Cognitive Claims of Metaphor, The Journal of Speculative Philosophy , 2(1), 1-16. Husserl, Edmund. (1939). Erfahrung und Urteil: Untersuchu ngen zur Genealogie der Logik . Prag/Hamburg: Claassens & Goverts, 1948. Kauffman, Stuart. (2000). Investigations . Oxford, etc.: Oxford University Press. Knorr-Cetina, K. D. and A. V. Cicourel. (Eds.). (1981). Advances in Social Theory and Methodology. Toward an Integration of Micro- and Macro-Sociologies London: Routledge & Kegan Paul. Kuhn, Thomas S. (1962). The Structure of Sc ientific Revolutions . Chicago: University of Chicago Press. Lazarsfeld, P. F., & N. W. Henry (1968). Latent Structure Analysis . New York: Houghton Mifflin. Leydesdorff, Loet. (1993). ‘Structure’/’Action’ Contingencies and the Model of Parallel Distributed Processing, Journal for the Theory of Social Behaviour, 23, 47-77. Leydesdorff, Loet. (1995). The Challenge of Scientometrics: the development, measurement, and self-organization of sc ientific communications. Leiden: DSWO Press, Leiden University, < http://www.upublish.com /books/leydesdorff-sci.htm > Leydesdorff, Loet (2000). Luhmann, Habermas, and th e Theory of Communication, Systems Research and Behavioral Science 17, 273-288. Leydesdorff, Loet. (1997). The Non-linear Dynamics of Sociological Reflections. International Sociology, 12 (1), 25-45. Leydesdorff, Loet. (2001). A Sociological Theory of Communication: The Self-Organization of the Knowledge-Based Society . Parkland, FL: Universal Publishers, < http://www.upublish.com /books/leydesdorff.htm > Leydesdorff, Loet, and Daniel Dubois (2003). Anticipation in Social Systems. Paper to be presented at the 6 th International Conference on Computing Anticipatory Systems CASYS'03 , Liège, Belgium, August 2003. Luhmann, Niklas. (1971). Sinn als Grundbegriff der Soziologie. In Habermas & Luhm ann (1971), pp. 25-100. Luhmann, Niklas. (1984). Soziale Systeme. Grundriß einer allgemeinen Theorie. Frankfurt a. M.: Suhrkamp. [ Social Systems . Stanford: Stanford University Press, 1995.] Luhmann, Niklas. (1986). The autopoiesis of soci al system s. In F. Geyer and J. Van Der Zouwen (Eds.), Sociocybernetic Paradoxes . London: Sage, pp. 172-192. Luhmann, Niklas. (1990). Die Wissenschaft der Gesellschaft . Frankfurt a.M.: Suhrkamp. MacKay, Donald M. (1969). Information, Mechanism, and Meaning . Cam bridge, MA, and London: MIT Press. Marx, K. (1857). Grundriße der Kritik der Politischen Oekonomie . Moscow: Marx-Engels- Lenin Institute (1939). Mason, Stephen F. (1992). Chemical Evolution: Origins of the Elements, Molecules and Living Systems . Oxford: Oxford University Press. Maturana, Humberto R. (1978). Biology of langua ge: 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. 20 21 Maturana, Humberto R., and Va rela, Francisco J. (1980). Autopoiesis and Cognition: The Realization of the Living . Dordrecht, etc.: Reidel. Maturana, Humberto R., and Va rela, Francisco J. (1984). The Tree of Knowledge . Boston: New Science Library. Mead, George Herbert. (1934). The Point of View of Social Behaviourism. In Mind. Self, & Society from the Standpoint of a Social Behaviourist. Works of G.H. Mead . Vol 1 (pp. 1-41). Chicago and London: University of Chicago Press. Mokyr, Joel. (2002). The Gifts of Athena: Historical Origins of the Knowledge Economy. Princeton and Oxford: Princeton University Press. Parsons, Talcott S. (1937). The Structure of Social Action. Glencoe, IL.: The Free Press. Parsons, Talcott S. (1968). Social Inte raction. In David L. Sills (editor), International Encyclopedia of the Social Sciences , Vol. 7 (pp. 429-441). New York: MacGraw-Hill. Pask, Gordon. (1975). Conversation, Cognition and Learning. Elsevier: Amsterdam. Pinker, Steven. (1994). The Language Instinct: The New Science of Language and Mind . London, Etc: Penguin. Popper, Karl R. ([1935] 1959). The Logic of Scientific Discovery . London: Hutchinson. Rosen, Robert. (1985). Anticipatory Systems: Philo sophical, Mathematical and Methodological Foundations . New York: Pergamon Press. Sayer, Andrew. (2000). Realism and Social Science . London, etc.: Sage. Searle, John R. (1998). Mind, Language and Society: Philosophy in the Real World. New York: Basic Books. Shannon, Claude E. (1948). A Mathemati cal Theory of Communication I and II. Bell System Technical Journal , 27, 379-423 and 623-656. Shannon, Claude E., and Warren Weaver. (1949). The Mathematical Theory of Communication . Urbana: University of Illinois Press. Simon, Herbert A. (1973). The Organization of Com plex Systems. In Howard H. Pattee (Ed.), Hierarchy Theory: The Challenge of Complex Systems (pp. 1-27). New York: George Braziller Inc. Steinmueller, W. Edward. (2002). Knowle dge-Based Economies and Information and Communication Technologies. International Soci al Science Journal , 54(171), 141- 153. Stichweh, R. (1984). Zur Entstehung des Modernen System s Wissenschaftlicher Disziplinen. Physik in Deutschland, 1740-1890 . Frankfurt A.M: Suhrkamp. Theil, Henry. (1972). Statistical Decom position A nalysis. Amsterdam: North-Holland. Urry, J. (2000). Sociology Beyond Societies: Mobili ties for the twenty-first century . London and New York: Routledge. Van Lente, Harro, and Arie Ri p. (1998). Expectations in Te chnological Developments: An example of prospective structures to be filled in by agency. I n Cornelis Disco and Barend van der Meulen (Eds.), Getting new technologies toge ther: Studies in making sociotechnological order (pp. 203-229). Berlin: Walter de Gruyter. Weinstein, Fred, and Gerald M. Platt. (1969). The Wish to Be Free: Society, Psyche, and Value Change . Berkeley, etc.: University of California Press. Whitley, Richard D. (1984). The Intellectual and Social Organization of the Sciences . Oxford: Oxford University Press. Loet Leydesdorff is Senior Lecturer for “Communication and Innovation in Science and Technology” at the Amsterdam School of Co mmunications Research (ASCoR) of the University of Amsterdam. See at http://www.leydesdorff.net/list.htm for his other publications.