7 The third position: beyond artificial and autopoietic systems
Reacting to the reduction of cognition to the computational Gestalt of Artificial Intelligence (the first position (A)) and to the cellular Gestalt of autopoiesis in cybernetics (the second position (B)), I declare a third position (C) in the conversational Gestalt of dialogical understanding, concerned also with its collapse by submittance to a monolithic perspective (model monopoly).
Two rival positions
Reductive claims have been voiced in both the first and the second position. In the computational Gestalt, physical symbol systems are regarded by Newell and Simon (1976) as necessary and sufficient means to generate intelligence. In the biology of cognition, proposed by Maturana and Varela (1980:82) in the cellular Gestalt of self-reproduction, autopoiesis is specified as necessary and sufficient to characterize the organization of living systems.
While Leibniz' monadology has been the negated point of departure for my own path, even when working in the perspectives of A and of B, different portions of the heritage left by Leibniz have been adopted by those two approaches emerging from the cybernetic foundations laid in the 1940's. Let me review briefly:
The Physical Symbol Systems Hypothesis (A)
Leibniz' idea about computation and artificial language came to be pursued within tradition A, defining itself as concerned with artificial intelligence (AI). In terms of this position, Plato's steersman, as a metaphor for cybernetics, may be seen as a regulator, operating through the use of symbolic representations, in spite of their limiting rationality (Simon 1969). A good regulator of a system requires a model of that system, although one must take care to distinguish between relevant levels (Conant and Ashby 1970; Newell 1982).
Simon (1969) is concerned with objects of design and description in terms of a hierarchical order of objects in the computational gestalt. The organization of components, rather than their physical properties, is emphasized. This comes to expression in the physical symbol systems hypothesis, proposed by Newell and Simon. A physical symbol system is a machine that produces through time an evolving collection of symbol structures, and which "exists in a world of objects wider than just these symbolic expressions themselves". They claim that
(a) "[a] physical symbol system has the necessary and sufficient means for general intelligent action" (Newell and Simon 1976:116).
The way in which two systems, A and B, form an communicative dyad may be grossly visualized in this way in terms of this position:
A and B each forms partly open physical symbol systems, linked together via their respective input-output facilities, and with access to a common protocol device that provides the rules and conventions for the production of outputs and processing of inputs. If organized by layers in the gestalt of the computer network, there would be different protocols for each layer, ensuring virtual communication between peer processes of the higher layers, as distinct from the physical communication between the lowest layers.
The principle of Autopoiesis (Operational Closure) (B)
Leibniz' theory of self-contained monads, each living in a closed universe, with no windows to the outside, may be recognized in the principle of cellular self-creative production or autopoiesis (Maturana and Varela 1980) as advanced by the other tradition, (B). The cybernetic notion of circular loops is retained, but stripped of external reference and mapping relations. The term 'in-formation' is used only reluctantly in the sense of forming from within. Regulation, if having sense at all here, would have to involve the relations of production of some order, maintaining the presence of participants in a continuous self-creative circle (Varela 1979).
While the above position (A) regards physical symbol systems as mediating, referential systems that have inputs and outputs and exist "in a world of objects wider than just these symbolic expressions themselves" (Cf. Newell and Simon 1976:116), a different kind of claim is advanced in the second position (B). Concerned also with the organizational basis, the latter is a biological approach to cognition as a self-organizational phenomenon which is "concealed through such magic words as ...symbolization" (Maturana 1980:56). An autopoietic organization is defined in the gestalt of cellular self-production to mean "processes interlaced in the specific form of a network of productions of components which realizing the network that produced them constitute it as a unity" (Maturana and Varela 1980:80). If living systems are machines, then they are physical autopoietic machines, transforming "matter into themselves such that the product of their operation is their own organization" (p.82), and if a physical system is autopoietic, then it is living. Hence,
(b) "autopoiesis is necessary and sufficient to characterize the organization of living systems" (Maturana and Varela 1980:82).
In contrast to the physical symbol system, an autopoietic system does not have inputs or outputs; its organization involves processes which are self-productive and non-symbolic, interlaced in the specific network of production of components which realizing the network that produced them constitute it as a unity (Maturana and Varela 1980: 80-81). That defines the identity of the individual, as distinct from the phenomenology that this entity, once constituted, can generate in its interactions, and which depends on the interaction of such unities in the media in which they exist, i.e. through structural couplings. Individuality and the phenomenology of unities in their interaction are distinct non-intersecting domains such that one cannot be reduced to or deduced from the other, albeit both domains may be considered from the same meta-position.
The scientific networks pursuing the two rival perspectives
Consistent with these different perspectives, cognitive interaction came to be viewed differently:
(A) In the first tradition, declared by Simon (1969) to belong to the sciences of the artificial, symbolic representations are considered basic, and human and computer intelligence are viewed as employing roughly 'the same symbol-manipulating processes'. Computer simulation of cognitive processes is seen as one of the few investigative means to study such complex processes.
(B) In the other tradition, Maturana (Maturana and Varela 1980) argues that such 'magic words as...symbolization' serve only to conceal the autopoietic phenomena of cognition and linguistic interaction. Systems operating in and by themselves (auto) need to be distinguished from those made for and by others (allo), and require additional operational explanations.
There has thus been a reductionistic tendency in both positions: for A in the artificial Gestalt of computer representation, for B in the biological Gestalt of cellular reproduction.
However, qua networks of cognitive interaction among scholars, each of the two rival traditions, A and B, has exhibited behavioural modes that conform to principles of its rival. In the first tradition, A, the network soon disengaged itself from other cybernetic traditions, defining its domain as that of AI in a self-enclosing manner. This fits the principle of organizational closure, postulated in the rival tradition. In the latter tradition, B, symbolic representations in an allopoietic system were resorted to in order to demonstrate autopoiesis (Varela 1979). This use of computer simulation is quite consistent within the rival tradition.
The third position (C) -- beyond artificial and autopoietic reduction
Thus, at least in the context of scientific communities themselves, behavioural modes conforming to both the artificial Gestalt of A and to the biological Gestalt of B can be shown to exist in each of the systems concerned. This, in turn, is one of the concerns of the present position, C, in the conversational Gestalt of dialogue and complementarities. The boundary of the first position has been crossed in the direction of the second by Winograd and Flores (1986). The second appears to have been transcended by Varela (1979) in his focus upon complementarity and conversation. In the congenial third position (proposed by Bråten 1984, 1998ab) intersubjectivity and dialogicity are considered as prior to the monological reason assumed by the first position, and as allowing intersubjective commun(icat)ion to be explained without reducing such a phenomenon to only linguistically mediated events.
Consciousness as dialogical
The above case may serve to demonstrate the need for complementary descriptions of sociocultural systems in terms of rival modes. Qua participants in meaning-processing systems, the scientists carry out operations involving symbols and representations within a self-enclosed domain which they maintain qua living systems. Their understanding is achieved through their involvement in a more or less shared culture of symbols and artefacts. This shared understanding is neither subjective nor objective: it is intersubjective, generating the subject-object complementarity.
Pattee (1978), concerned with biological systems, makes a strong claim concerning irreducible subject-object complementarity:
"There is, to be sure, wide recognition that systems theories need a richer basis for modelling living systems. Bråten...has summarized these approaches which apply adjectives such as self-referential, dialectical.... These are in the right direction, but fail to convey the essential epistemological shift from what I call the classical paradigm of the objectivity of systems to the subject-object dualism characteristic of life." (Pattee in Klir (ed.) 1978: 518)
Modifying his former B position to allow for complementary explanations, Varela has taken account of this point, and I stand partly corrected, albeit not quite in the above sense. Having worked from both the A and B positions, I have previously failed to demonstrate the reversible Gestalt-switches involved, and to disclose conditions for shifts between epistemological modes that are operational within sociocultural systems and conform to both positions. This search is still in progress in terms of the basic distinction between the mode of felt immediacy and the mode of simulational mediacy. Earlier, I have rather been concerned with conditions that, respectively, promote and prevent the intersubjective crossing of perspectives that is viewed as a prerequisite for consciousness and a creative meaning horizon. When Varela (1979), of the B tradition, and Winograd (1980), of the A tradition, each recently took account of the rival tradition, their writings serve to demonstrate the kind of creative consciousness that may emerge from the crossing of perspectives. But I find in their work a lack of awareness of certain irreducible properties of sociocultural systems:
o the unique character of intersubjectivity (as distinct from the subject-object dualism which it generates but does not equal) which may be realized by subjects engaging in felt immediacy, as well as in the more commonly entertained notion of representational mediacy;
o the symbolic operations in a meaning-tight manner internal to the domain (involving artefacts and symbols that cannot be reduced to linguistic sequences, but are none the less symbols); and
o the potentials for consciousness as dialogue between conceptual perspectives.
The latter point is stressed by Pask (1978) in his conversation theory. While not sharing his definition of social systems as autopoietic (see Zeleny (ed.) 1981) -- they involve both nature and culture -- I consider much of his work as oriented towards a third position. Pask struck the dyadic keynote in a seminal paper on the meaning of cybernetics (in Rose (ed.) 1970). About the same time I wrote a working paper on the 'Human Dyad', suggesting a principle of requisite inconsistency (Bråten 1971). Independently, and from different points of departure, we have reached a common concern with consciousness through a dialogue between perspectives - either inside one's own mind, or in a human community. From this, it is only a short step to study what happens when a conceptual agreement or contract (Rommetveit 1974), a consensual world of meanings, breaks down.
In some laboratory studies of small groups (using Blakar's map design and generating referent systems for computer simulations), I have noticed that under break-down conditions participants may shift from an insider's mode of participation to an almost outsider's mode of reflection. In two different series of experiments, their taken-for-granted shared world of meanings was distorted: in one study I switched labels in a story that had to be subjected to ethical judgements by a pair of persons; in another study the participants had a routing task with reference to maps that were not identical, but were believed to be so. It was rather a surprise to observe how long participants carried on without suspecting anything. Self-reflective comments upon video playback indicate that, when they became aware something might be wrong, they resorted -- among other strategies -- to shifting among perspectives and alternating between different kinds of symbolic representations. Why would such a break-down evoke such a resort to a boundary position of reflection? One reason may be that, since they are usually being involved in an intersubjective world of meanings, the insiders cannot refer to it qua insiders.
The above incidents occurred in the artificial setting of a laboratory. However, events in more naturally evolving groups, discussing 'Our Computer Milieu of Yesterday, Today, and Tomorrow', also indicate that shifts from an insider's position to a boundary position, through the group's own usage of self-reflective video playback, have seemed to restore a meaning-tight platform for agreement and disagreement. The conversation sometimes breaks down and participants may show anger when others falsely attribute viewpoints to them that are based on misunderstanding. Relief is expressed when the platform is re-established even when there is disagreement, as long as it rests on intersubjective understanding.
Thus, as long as the platform for this understanding is not broken, one may carry on without being conscious of what one is doing. Indeed, in daily life and practical affairs, consciousness is sometimes highly unoperational. It would be like asking the bicycle rider to be aware of the working of his feet, or the scientist at the keyboard to note the touch of her fingers. Operations would break down. Even in the context of scientific pursuit, in spite of the falsification ideal, biased and persistent behaviour from a single perspective may sometimes seem to be preferred to conscious and discursive modes, producing results arising from the fact that one is blind to occurrences that do not fit the shared understanding.
The third path as mediating between the first and the second position
In his ontological scheme, Maturana (1988:29) compares these two explanatory paths: first, an explanatory path that assumes objectivity and "is constitutively blind (or deaf) to the part played by the observer in the constitution of what is accepted as explanation, and second, the path (his own position) that brackets objectivity in the awareness of the observer's part and praxis of living in language.
Now, the concern of the third position is to assume that there are blinkers involved in any path or position (including of course the third), for example, the lacking awareness of the intersubjective prerequisite for being able to bracket objectivity (and which Husserl discovered, arriving at his radical notion of lifeworld (Lebenswelt)).
Such blindness, however, may be replaced by creative consciousness when different paths are allowed to mutually specify and complete each other through dialoguing. Then the observer-independent world (universum) of the first position and the multiple subjective observer-specified worlds (multiversa) of the second position may be seen to be generated by the complementary perspectives of participants living in dialogue and co-creating an intersubjective world (interversum2).
Thus, the third position, or more adequately labelled in this context as the third path"3, is concerned with the kind of creative consciousness and praxis that permit the unfolding of perspectives that specify each other and the changing wholeness which they constitute through their mutually completing each other in an on-going conversation. This defines the conversational Gestalt as a continuous conversation between complementary perspectives, including those of (A) and (B) in the format of C[(A),(B)].
Submittance to a monolithic perspective: model monopoly
Current reflections of Leibniz' ideas
I have indicated how different portions of Leibniz' heritage may be seen to be reflected by the (A) and (B) traditions. In AI his conception of Characteristica Universalis came to be pursued:
"If controversies were to arise, there would be no need of disputation between two philosophers than between two accountants, for it would suffice....to sit down to their slates, and to say to each other (with a friend as witness, if they liked): Let us calculate." (transl. in Russell 1961:572-73).
This is Leibniz' dream of how the processing of contrary viewpoints could be carried out in terms of a generalized logic. It may appear to be inconsistent with his ontology of a multiplicity of subjective viewpoints, each a self-enclosed monad with an autonomy that makes each the source of its own activity (Discourse on Metaphysics, XI).
His monadology may be recognized in Maturana's position, assuming multiple subjective observer-specified worlds (multiversa), and in the principle of autopoiesis. Leibniz points out that as a result of "the infinite multitude of simple substances, there are as it were so many different universes", such that the universe is multiplied perspectively "according to the different points of view of each monad", each with a certain self-sufficient autonomy "which makes them the source of their internal activities, and so to speak, incorporeal automata" (Monadologie, Reclam ed. 1954:9,18,57).
Leibniz' submittance to Aristotle's model monopoly of logic
Since such a world view acknowledges the existence of a multiplicity of subjective viewpoints as constituting reality, it might be expected to provide sufficient inoculation against submittance to a monolithic perspective, to a model monopoly. But apparently, this is not sufficient. Even the holder of a pluralistic world view like Leibniz may come to submit to a monolithic perspective:
"He (Leibniz) did work on mathematical logic which would have been enormously important if he had published it; he would, in that case, have been the founder of mathematical logic, which would have become known a century and half sooner than it did in fact. He abstained from publishing, because he kept on finding evidence that Aristotle's doctrine of the syllogism was wrong on some points; respect for Aristotle made it impossible for him to believe this, so he mistakenly supposed that the errors must be his own." (Russell 1961:173; my emphasis).
Provided that Russell's recount is adequate, Leibniz himself appears to have submitted to a mono-perspective. In spite of his entertaining an ontology that assumes a plurality of subjective viewpoints, he appears to have surrendered to the symbolic power of Aristotle's doctrine.
This, then, is one of the concerns of the third position: Conditions for a continuous conversation between complementary perspectives that are not arrested and lost in some monolithic perspective (model monopoly) in the face of a reductive claim.
Sometimes, such single-mindedness or blindness is carried over to conversational contexts - in a boardroom or classroom, in a computerized network, or - as I shall examine below in A terms - in a socioeconomic planning establishment, identified by Rød-Larsen (1976) to be monolithic.
An example of model monopoly: socioeconomic planning in post-war Norway
The socioeconomic establishment in Norway after the second world war exemplifies a case of social systems planning and attempts to steer in terms of symbolic representations. It illustrates an operational mono-perspective, to the exclusion of any rival perspective, or, in A terms, the establishment of a model monopoly, which was considered to be the only valid steering and complexity-reducing 'map' to which one tried to adjust the national economic 'territory'.
In this establishment, Professor Ragnar Frisch played a crucial role. During the economic crisis in the 1930's, a new paradigm for a more active role of the state in the national economy was being developed -- in Britain by Keynes, in Holland by Tinbergen, and in Norway by Frisch, who in 1969 shared with Tinbergen the first Nobel Memorial Prize in Economics. When called to his Chair in Oslo in 1932, Frisch had his version of the paradigm ready. From this chair he educated those who came to occupy leading positions in dominant financial institutions in Norway. Among these were the State Department of Economics, which was responsible for the implementation of praxis in accordance with his paradigm, and the Central Bureau of Statistics, which was responsible for generating and processing data in accordance with his socioeconomic model. Even politicians who were to become members of the Norwegian parliament and government came under his influence and served to strengthen the model monopoly.
Directly and indirectly, the second world war paved the way for the establishment of macroeconomic steering in terms of this model monopoly. This happened in two ways. First, the Norwegian economy had been virtually destroyed during the occupation, and one was faced with the formidable task of rebuilding it. This called for a monolithic effort. Second, during the war Frisch was a prisoner of war in the Grini Camp, together with many of the later leaders of the Norwegian government. He organized seminars on economy, teaching the leaders-to-be how to steer in terms of his paradigm (cf. Röd Larsen 1976, and personal communication with T. Gulbrandsen, H. Hansen-Bauer, L. Johansen and P. M. Munthe).
Thereby a model monopoly was established, not merely because the Frisch model was adopted (and developed) as the only valid model for the reconstruction of the Norwegian economy, but also through a subsequent process of self-validation. At the university department steering devices were worked out in terms of the model; at the State Department of Economics they were put into effect. The government attempted to adjust the economic 'territory' to the 'map'. The Central Bureau of Statistics gathered and processed data - again in terms of the model. This mode of social systems planning and steering partly worked - at least until the international economic territory started to exhibit its decline in the middle of the 1970's.
Mechanisms promoting submittance to a model monopoly
The model power thesis (Bråten 1973c) may explain part of this genesis of a socioeconomic model monopoly. A given universe of discourse, in this case the national economy, E, may be predefined in such a manner that only one of the participant actors, A, is rich in relevant concepts and symbolic representations which reflect his own interests and perspectives. This actor is termed the model-strong actor -- in this case the socioeconomic establishment described above, or in the more limited case, Ragnar Frisch and his close collaborators.
Definition: If all the elements and relations in E, which are describable in terms of B's perspective, also are describable in terms of A's perspective, and there are elements in E that are describable only in A's perspective, but not in B's, then A is the model-strong actor and B the model-weak one, with respect to E.
The perspective of B is 'swallowed by' or properly included in A's perspective, preventing any dialogue between intersecting and non-empty complementing perspectives.
Certain ideal type pre-conditions furthering such a submittance to a model monopoly may be specified:
(c1) A stable domain is well-defined by the "model-strong" source of knowledge,
(c2) who is present or (re-presented by symbolic artifacts) in a situation of closed interaction which excludes other perspectives and alternative sources of knowledge
(c3) than the one who has defined the domain and on whose premises the knowledge has been developed and, hence, is
(c4) acknowledged as the source of valid replies about the domain in question (Bråten 1983).
Given such pre-conditions, a state of submitting to the model power of the "model-strong" source will be seen to be implied by these two postulates in conjunction (Bråten 1973c, 1981b):
(Pl) A representation of X is required for the efficient control of X.
(P2) Participants in symbolic interaction seek to employ representations (models) of their domain of discourse, including each other.
The first postulate is a version of the Conant-Ashby theorem (Conant and Ashby 1970). The second is consistent with symbolic interactionism (Mead 1934). Consistent also with a tenet of the first position (A), models or representations are here seen as internal complexity-reducing and coding simulation devices, used by participants to generate tentative answers to self-imposed questions about the productions and interpretations of the other participants (Bråten 1974).
Now, assume a closed domain of discourse between model-strong A and model-weak B (according to the above definition and ideal-type specifications). This may occur in a classroom, in a boardroom, in the Socratic dialogue Gorgias (Plato 1952), in the context of what Habermas (in Habermas and Luhmann 1971) terms 'the ideal speech situation', or in the above context of national economy planning and steering. The second postulate (P2) implies that in such a situation
(Implic. 1) the model-weak B will seek to adopt the models offered by the model-strong A.
As long as these models have been developed in terms of A's premises, and are accepted by B as the only valid ones, it follows from the conjunction of this implication (Implic. 1) and the first postulate (Pl) that:
(Implic. 2) The more 'successful' the model-weak B is in his adoption of A's models, the more B comes under A's control.
The ultimate control is reached when B's adoption gives A the power not merely to simulate B's behaviour, but even to simulate B's simulations, which are now carried out in terms of the models or simulation devices developed on A's premises (Bråten 1973c).
The socioeconomic planning case in the above light
Members of the socioeconomic planning elite in Norway, with the Frisch paradigm as the A-perspective, themselves acknowledged the need for a strong position and stated: 'Manipulation of aggregated macro-entities was something which only an authoritarian political-administrative system could carry out' (Röd Larsen 1976). If model power mechanisms operate, together with attempted adjustments of the 'territory' to the 'map', control may then be in part established, especially since the map is already part of the territory -- at least as long as the territory does not exhibit catastrophic shifts.
A post-war incident during proceedings in the parliamentary budgetary committee may illustrate the model power of professor Frisch. The members were uncertain about his definition of a key term and had to obtain his view. Frisch, however, was abroad at the time, in London. The committee sent him a cable. Only when the chairman could read out his cabled reply, did the committee resume its work.
A shift eventually did occur, owing to the international economic depression that occurred in the middle of the 1970's. But even though this control was finally broken, this case not only exemplifies the establishment of a model monopoly, but also shows that the cybernetic paradigm of social systems regulation through symbolic representations is operational under certain limiting conditions. The historic task of reconstructing a country called for joint and cooperative efforts, a common model platform and firm leadership. The Norwegian economy was steered in a certain manner, and this mode of steering worked as long as the international environment exhibited a certain degree of stable growth. But even when the environmental shift occurred, the socioeconomic elite continued for a while as if their model were still valid. It took outsiders to recognize that the territory had changed, and that alternative maps were required. There was a gradual recognition of self-regulatory market mechanisms, and of system-internal participatory transformations, operating under their own law, conforming to the conditions consistent with the closure position (B).
Resolving a model monopoly ~ reactivating dialogue between perspectives
Modes of dissolving conditions for a model monopoly (in A terms) or a mono-perspective (in B terms) are sometimes required because such a systems state not only entails blindness, but is sometimes -- though not in the above case -- downright oppressive to the participants. For example, a consensual life-world of meanings in a tight local community or network may seem to ensure the autonomy of the collectivity, but may deny it to its individual members. Extreme cases like that of Jonestown and, perhaps, present-day Iran need not concern us here. Daily life cases such as in the jury room, or in the local network of the firm, provide sufficient grounds to the need to indicate modes that may dissolve a mono-perspective and create climate for dialogue, for example through awareness that "coercion by way of the best argument" presupposes a rationally pre-defined universe of discourse in terms of one perspective.
Of particular importance in the present context of cognition and praxis are cases of model monopoly preventing discoveries and delaying what Kuhn (1970) terms 'change of paradigm'. The case of Leibniz submitting to the syllogistic perspective of Aristotle is only one of a number of instances in the history of Western science and philosophy. Another example is the case of how Euclid's geometry for more than twenty centuries was accepted as final and unquestionable. When the fifth postulate is finally questioned by Gauss, he does not publish his doubts. They are only voiced when, in the 1830's, an outsider, Bolyai's father, tells Gauss that his son has come up with an alternative geometry. Then, and only then, the model monopoly of the Elements is cancelled by the alternative perspective of Gauss, Bolyai and Lobachevskij.
The previous list of conditions (c1 - c4) promoting a model monopoly gives us the clue to resolution modes. The situation may allow for re-definition and re-framing in a manner that may cancel one or several of these conditions, re-activating a transition to the state of complementary non-empty perspectives in dialogue:
(-c1) Shifting the boundary of the universe of discourse or re-defining the domain in a manner which reveals the limit of the monolithic perspective. For example, if Leibniz had permitted himself to re-define the domain of syllogistic principles into a domain concerned with formal logic, not psycho-logic, that is, distinct from, or only intersecting with Aristotle's domain of necessary inferences of thought, he need not have submitted in the way in which Russell reports him to have done.
(-c2) Opening up for rival knowledge sources and admitting propositions in alternative languages or in rival frames, for example, in the way in which a leading figure of each of the two traditions, respectively, Terry Winograd (1986) of (A) and Francisco Varela (1979) of (B) each has taken account of the rival perspective of the other tradition, the latter in terms of complementarity in line with the third path.
(-c3) Developing knowledge on own premises, taking a boundary position which allows for the crossing of boundaries and a reflective view on the relations between the premises of the mono-perspective and one's own perspective that has been passivated in the process. There are many ways of cancelling or escaping submittance to a mono-perspective, such as resorting to a meta-level (G. Bateson 1973), entering a dialectic modus (Herbst 1976), stepping back for reflection (cf. Schön 1983) or, just break off interaction while developing and consolidating your own perspective (cf. Næss 1982:126-27).
(-c4) More generally, being aware of the kind of conditions that may promote a model monopoly and evoke psycho-logical tendencies towards cognitive consistency, monadic closing and conformity which prevent a creative horizon in the individual and in the community, for example awareness of the way in which a quest for explanation on a rational ground, as in parts of (A) compels us to stay within domains defined in terms of rationality, and of the way in which our biological roots of closure, as emphasised in (B), may contribute blinkers in our understanding of the other, and compel us to take an either-or stand instead of unfolding a mutually complementary ground, in the individual and in the community. Here, Habermas' recipe for discursive modes contributes to the resolution of any monolithic perspective or model monopoly, since they invite an awareness of world views underlying statements made in the speech situation. But this should naturally also be applied to underlying rationalistic world views, including his own.
A cybernetic pathway towards a dialogics of understanding?
The above modes provide a platform for posing this question: Could there be a dia-logic of intersubjective understanding which would facilitate conversation between contrary or complementary viewpoints, for example, between those of the first (A -- the artificial symbol processing) position and the second (B -- the biological autopoiesis) position, without super-imposing a controlling perspective upon either of them? If this might be envisaged, might there be a cybernetic pathway towards such dialogics, that is, to a cybernetics of dialogical understanding?
The I-You concern of second-order cybernetics of Heinz von Foerster (1984) and in Pask's (1981) conversation theory may open for this, provided it be applied to cybernetics itself without insistence upon the right or privileged cybernetic language. Even though the two traditions, concerned respectively with (A) Artificial Intelligence and (B) with the biological roots of cognition, differ sharply in their views of objectivity and representations, they have a common concern with cybernetic machines, circularity, referentiality and construction: the privileged language of (A) allows us to talk about circular feedback-loops and object-reference in the system constructed by the designer; the language of (B) is suitable for talk about organizational closure and self-reference in the domain specified by the observer as a closed system that specifies the world brought forth. Thus, their common cybernetic ground may offer some promise of conversation between the differing viewpoints, as exemplified by the way in which the boundary between them has been crossed by Varela (1979) and Winograd & Flores (1986).
Perhaps we could envisage, after all, a cybernetic language that may carry across traditional boundaries and permit on-going conversations between viewpoints, without their being encapsulated in locked rooms without any doorway between them, as it were, without the dialogue being arrested in terms of a monolithic perspective accepted as final (endgültig) by the participants.
The implicate order of cybernetics appears, however, to emphasize observers, not participants (cf. Skjervheim 1974); to entail subject/object dualities, not intersubjectivity; to emphasize construction, not mutual constitution, and to focus on linguistic mediation, not felt immediacy (cf. Bråten 1988). It remains to be seen whether it be open for a third path that may apply to dialoguing participants in a lifeworld that in part has to remain unspecified as the horizon within which meaning-processing participants maintain and transform the identity of themselves and of their networking in virtue of a more or less shared understanding of themselves and of their world, co-created through the (proto)conversations in which they exist.
Some recent advances in theory and praxis
Some recent advances, however, within and outside the above traditions may come to provide tentative replies to the above question about a possible pathway.
In cognitive science the ground of AI has been shaken by the reactivated concern with self-organizational neural networks, for example by Fukushima's (1988) Neocognitron model, and by the revolt against AI's rationalistic tenets by several, including Terry Winograd, who refers inter alia to the works of Heidegger and Maturana. The latter, in turn, has brought forth a scheme that includes the ontic commitments of the first position of AI in order that they be bracketed by the second.
Social Systems theory
In social systems theory Niklas Luhmann (1986) has formulated his theory in terms of a meaning-processing variant of the notion of autopoiesis, and which breaks with the biological ground of the second position. He regards consciousness (albeit not in a dialogical sense) as the distinctive autopoietic modus of the psychic system, and communication as the corresponding self-reproductive modus of social systems. This may be seen to concern self-reproduction of historically realized monological communities which have submittance as a form of external relation (with one community submitting to or forcing the other to submit in interaction -- since communication across its systems boundary is not possible in his theory). Boundaries cannot be crossed by members (they belong to the social system's environment) and their I-awareness functions only as long as there is consensus. In the third position I acknowledge the prevalence of such monological modi, but not their invariance: the collapsed dialogical circle may activate itself permitting boundaries to be crossed. Pointing to the unsolved aspect of the (self)observer's taking an outside and an inside position, Luhmann is open to this possibility:
"It is possible to differentiate subsystems within the social system, giving them the special function of 'observing society'. This still implies self-observation, because the subsystem can only operate within the society. It can look at its social environment....but this observation itself (1) is part of the autopoiesis of the social system, and (2) becomes self-observation as soon as it tries to observe and control its epistemology. Somehow, a 'Third Position' (Bråten, 1984 and Chapter 12 below) would be required, one which contains the possibility of shifting between external observation and self-observation and of defining rules which tackle the paradox of being at the same time inside and outside the system" (Luhmann 1986:187-88).
We have to consider dialoguing in praxis, though, in order to arrive at an understanding of the ways in which boundaries may be crossed.
In systemic psychotherapy the invention by Tom Andersen (1987) of the reflecting team of therapists which change place with the patient-therapist participants, approaches the actualization of the third position in praxis (cf. also Hoffman 1988).
The basic dialoguing attitude called for is strikingly actualized also by Harlene Anderson and Harold Goolishian in their family therapy work with patients who have a long clinical career before they come to see them. Being concerned with the prerequisites for dialoguing on the premises of the patient's perspective in the intersubjective setting of the conversation, Anderson and Goolishian bracket any deficiency preconceptions. They regard the ideal therapist not as an expert on pathology, but as a participant manager of conversation in which the therapeutic conversation does not differ from any other regular conversation4:
"The therapist does not control the interview by influencing the conversation towards a particular direction in the sense of content or outcome, nor is the therapist responsible for the direction of change. The therapist is only responsible for creating a space in which the dialogical conversation can occur and for being participant in maintaining the conversation. Bråten [....] describes this as a conversation that is intersubjective and one in which participants can make room for the creativity and consciousness of each other.
The goal of creating a space for, and participating in dialogical conversation is central to the therapist's position." (Anderson & Goolishian 1987).
Considering the implications for clinical theory they thus identify their praxis with that of being actual companions to the patient in the mode of dialoguing in which no position of rational expertise is taken, and control is avoided even in the sense of imposing upon the other any label or invite submittance to any monological or monolithic perspective. Anderson and Goolishian's point out, though, that one "must be careful, however, not to confuse a mono-perspective with problems; that is, to think that a mono-perspective is always problematic. It may or it may not be." I accept this correction of my original declaration of the third position (Bråten 1984); I have not differentiated clearly enough between the time for dialoguing and the time for monological implication (one cannot walk both to the right and to left, even though both directions would have to be considered).
Dialogicity in psycholinguistics
In psycholinguistics Ragnar Rommetveit (1988) demonstrates dialogicity and the socio-cognitive embeddedness of languaging. Communication comes about through the complementarity between the speech acts of the speaker and the listener in the intersubjective lifeworld in which they exist. Rommetveit points to the convergence of axiomatic features of a number of dialogical approaches to language and mind, and which may be captured in my third position terms about the dialogical constitution of the mind.
Dialogical conception of the mind
In the psychosocial domain, implied by a postulate about the innate dialogical constitution of the mind (formulated in third-position terms), I have put forth a theoretical account of infant-adult protoconversation in terms of dialogic closure (Bråten 1988) -- an adapted version of the principle of operational closure (assumed in the second position). I have suggested that the two preconceptions about cognition and communication may be fruitfully reversed in this way: First, the individual mind may be seen as a system of participants in conversation, e.g. in terms of my thinking, creating, and feeling with You. Second, certain events of communication -- while distinguished by the observer to be realized by two bounded organisms -- may be considered to be generated by one dialogically closed organization in the modus of felt immediacy.
I submit that a circular dyadic organization is the primary characteristic of the individual mind, involving the initial participation of the Virtual Other, in the place of which the Actual Other may step into the dialogic circle, while the identifying form of the dialogical maintains itself. This entails that the even the newborn baby should be able to perceive and relate, imitate and complement, the adult Other in an immediate sense. Recent infancy research, including the discovery of protoconversational patterns in the mother-infant dyad, reveals such capacities in the newborn (cf. the references to M.C. Bateson, C. Trevarthen, and others in Bråten (1988b).
Summary and conclusion
The third position has been recapitulated in the special sense of being concerned with a continuously unfolding pathway towards the understanding of the dialogical roots of human understanding, and in the general sense of considering relations between complementarities in terms of an on-going dialogue between complementary perspectives without their becoming encapsulated or lost in a monolithic perspective that arrests the dialogue.
Thus, the concern with the complementarity of modes conforming to the respective tenets of the two rival positions, respectively, Artificial Intelligence (A) and Self-organizing systems (B), surveyed and compared elsewhere (Bråten 1982b), may be seen as a special case of a more general concern with the unfolding of a complementary ground for creative consciousness and conversation, and for the way in which such dialoguing sometimes is arrested when faced with the demand in science for a unified reduction, or in praxis, for the final (re)solution.
Recent advances in theory and in praxis have been referred to as indicative of the third position as a dialoging path towards our unfolding a complementary, intersubjective ground for our understanding. This includes a postulate about the dialogic constitution of the mind in terms of Virtual Alter. It permits dialoging -- and modeling -- to be seen as a creative alter-nating circle with the Other, actual or virtual.
1 Talk at Colloque de Cerisy, Perspectives Systemiques 2, Cerisy la Salle, 25 Sept. 1988 (published in Praxis et Cognition, ed. by Bernard-Weil & J-C. Tabary, Paris 1992:205-14), fused here with notes for a talk at the 6th International Congress of Cybernetics and Systems, Paris, 10-14 Sept. 1984 (based on my article in Kybernetes (1984) vol. 13: 157-63).
2 G. de Zeeuw proposed this third-position term upon my talk on this theme at the Gordon Research Conference on Cybernetics and Cognition (Wolfeboro, June 1986).
3 H. Maturana (personal communication, Bielefeld, Sept. 1988) suggested that this would have been a more adequate label for the position I have declared. I agree.
4 Added 1999 to this Sigma edition: cf.. also H. Anderson's book on Conversation, Language and Possibilities (Basic Books, 1997), in particular the chapter on Therapy as Dialogical Conversation.