Almost 20 years ago I started the Radical Constructivism web page with the intention to collect sources available on the internet pertaining to constructivist ideas and concepts in order to make it easier for us (who are working with constructivist approaches) to quickly access the variety of different aspects and approaches that come with the constructivist perspective. Later this effort has been accompanied by founding the journal Constructivist Foundations in order to provide a serious scholarly publication platform for constructivists. Have these two little initiatives been enough or do we need further means and tools to facilitate current research in constructivism and to attract new researchers? In my talk I shall review the current format of the journal and report on two new ideas that are currently being implemented: (a) the Constructivist E-Paper Archive (CEPA.INFO), a research tool providing a single-point access to the scholarly literature on constructivism which sometimes is hard to get by. In addition to full texts it enables annotations, collecting quotes, and the creation of reading lists. (b) the Constructivist Encyclopedia (CE), an open-access publishing initiative where qualified authors with relevant expertise provide an accurate and reliable source of information relevant to constructivist approaches. It targets both the educated layperson as well as the advanced specialist.

Radical Constructivism has been subject to extensive criticism and denigration such as that it is a naturalized biologism which supports an "anything goes" philosophy of arbitrarily constructed realities. In an extreme case RC is equated with intellectual silliness. These accusations are to be refuted. Based on the concept that cognition can work only with experiences, I investigate the question of where their apparent order comes from. Arguments are presented that favor the amorphousness of the "external" world. To support the idea of "internal" order I review results in formal network research. The properties of such networks suggest that order arises without influence from the outside. The conclusion are threefold. RC based on network models (a) does not need any empirical support and is therefore no biologism nor naturalism, (b) forgoes arbitrariness, and (c) goes beyond narrative (armchair) philosophy.

“L’intelligence organise le monde en s’organisant elle-même”, wrote psychologist Jean Piaget and referred to the claim that the human mind actively constructs the world it lives in. Indeed our common sense self-understanding describes us as beings of free will. And it is this freedom to take the decisions we fancy that is thought to separate us from any machine, endowed with artificial intelligence or not. Already more than 50 years ago Alan Turing set out to correct the view that a machine is not human-like until “it can write a sonnet… [however] not only write it but know that it had written it.” His solution, however, to settle at the “polite convention that everyone thinks” in order to prevent us from falling prey to solipsism, did not catch on among proponents of A.I. If we declare fellow humans equipped with free will there will be no reason to assume that we have to exclude sophisticated artifacts from becoming conscious as well.
Let us reconsider the “polite convention” by picking up the idea that we indeed cognitively construct our world. This perspective rests on the insight that cognition is an organizationally closed system, a network of dynamically interacting elements whose properties are solely determined by the properties of other elements. However, what appears to be a solipsistic position—the one Turing warned us to take—is merely agnostic in a Wittgensteinian sense. We are asked to refrain from clinging to a mind-independent reality as the anchor of cognition. At the same time we open the door for a genuine A.I. that is capable of gaining cognitive independence of its creators. Although a subjective worldview is the result of a construction process, the construction process itself is not solipsistically arbitrary. The interlocking of continuously added elements of experience forms a hierarchy of schemata, which ultimately results in interdependencies among these elements and thus in canalizing forces that make arbirariness impossible. Such a structure does not only describe human cognition but applies also to the design of artifacts. The challenging question is whether these canalizations drive both humans and intelligent machines into the position of Huxleyan Helpless Spectators, or whether they are the conscious constructors of their respective world, as Piaget wants it.

Constructivism comes in many variants, which gives rise to the assumption that it is not (yet) fully developed. I will review the most important versions related to cognition. This allows to underline the common ground, to spot the differences, and to formulate general principles of the (cognitive) constructivist endeavor. By making these principles explicit I will also hint at their implications for scientific research and for the future of constructivism.

Gemäß des Konstruktivismus sind wir die Konstrukteure unserer eigenen Welt und nicht durch eine externe Realität determiniert. Seine am meisten konsistente Formulierung, der Radikale Konstruktivismus (RK), behauptet, daß wir unsere Erfahrung nicht transzendieren können. Deshalb macht es wenig Sinn zu sagen, daß sich unsere Konstruktionen den Strukturen einer externen Realität annähern. Der Vortrag soll die Grundzüge des RK darstellen, sofern dies bei seiner Heterogenität möglich ist. Fragen werden behandelt wie etwa: Sind unsere Konstruktionen arbiträr? Wer macht wirklich die Konstruktionen? Zeichnet dafür eine Seele verantwortlich? Zum Abschluß wird ein Vergleich gezogen zwischen der hier vorgebrachten Aufassung und traditionellen Seele-Begriffen.

Volgens het constructivisme construeren wij onze eigen wereld eerder dan dat deze door een externe werkelijkheid zou gedetermineerd zijn. In de meest consistente formulering, het radicale constructivisme (RC), wordt beweerd dat wij onze ervaringen niet kunnen overstijgen en we daardoor geen toegang hebben tot een absoluut perspectief. Het heeft dus geen zin te zeggen dat onze constructies stap voor stap de structuur van een externe werkelijkheid benaderen.
De voordracht zal de fundamentele principes van het enigszins heterogene RC presenteren. De volgende vragen worden o.a. besproken: Wat zijn de gevolgen van zulk een wereldbeeld? Kunnen we het gebruiken om kunstmatige cognitieve wezens te bouwen —een methode om al te vrijblijvende filosofische speculaties over waarheid en werkelijkheid te vermijden? En zijn onze constructies willekeurig?

Was ist (wissenschaftliche) Erkenntnisgewinnung? Diese Fragestellung soll aus der Perspektive des Radikalen Konstruktivismus beantwortet werden. Zunächst führe ich drei prinzipielle Untersuchungsebenen ein. Danach zeige ich, wie diese Ebenen mit dem Radikalen Konstruktivismus zusammenhängen. Schließlich führe ich aus, welche Möglichkeiten sich aus einer konstruktivistischen Perspektive für die Wissenschaften ergeben.
Ich unterscheide drei Schichten des Herangehens an (wissenschaftliche) Probleme. Auf der Ebene der Natur- und Deduktiv-Wissenschaften finden wir eine strikte Systematik und quantitative Exaktheit. Die zweite Ebene wird durch die Philosophie abgedeckt. Sie bietet gemäß John Searle [1] den begrifflichen Rahmen für Fragen, zu deren Beantwortung noch keine systematische Zuordnung gefunden wurde. Wenn sich eine solche Systematik etabliert, wird aus dem philosophischen Problem ein (natur-) wissenschaftliches. Trotz strenger Argumentationslogik besitzt damit die Philosophie keine strikte Systematik: Es werden Argumente hervorgebracht und von anderen Philosophen übernommen, die plausibel erscheinen, und daraus Schlußfolgerungen gezogen, die einleuchtend sind. Schließlich treffen wir noch die Ebene des gesunden Menschenverstandes (common-sense) an, die sich durch rasche aber vage qualitative Abschätzungen auszeichnet.
Um die Beziehungen zwischen diesen Ebenen verstehen zu können, müssen wir genauer betrachten, was "Kenntnisgewinnung" bedeuten soll, also worin Zweck und Ziel der Wissenschaft liegen. Das sind Finden von Erklärungen und Treffen von Voraussagen. Beide Komponenten sind unverzichtbar. Erklären, also das Einordnen von neuen Erfahrungsinhalten in ein bestehendes Netzwerk von vorher gemachten Erfahrungen und Erklärungen, findet man auf allen drei Ebenen. Es kann als das Konstruieren von Sinn im Strom der Erfahrungen aufgefaßt werden. Das Erzeugen von erfolgreichen Voraussagen setzt dementgegen quantitative Exaktheit voraus. Darin unterscheiden sich die Ebenen. So kann sich beispielweise auch die Philosophie trotz redlicher Argumenationslogik nicht der Unschärfe qualitativer Schlußfolgerungketten erwehren.
Diese Ausführungen entsprechen der Position des Radikalen Konstruktivismus (RK). Er läßt sich in vier Punkten festlegen [2]. Das Radikal-Konstruktivistische Postulat sagt, daß das kognitive System (Verstand, mind) operational geschlossen ist. Die Epistemologische Folgerung daraus ist, daß die Existenz einer "äußeren Realität" weder verneint noch bestätigt werden kann. Die Methodologische Folgerung besagt, daß Erklärungen in dem Sinne zirkulär sind, daß sie sich auf andere Erfahrungsinhalte beziehen und dadurch Sinn erfahren. Schließlich verneint das Postulat der Konstruktionsbeschränkung die Möglichkeit beliebiger (Sinn-)Konstruktionen.
Zum Schluß möchte ich die Konsequenzen einer radikal-konstruktivistischen Auffassung von Wissenschaft betonen. So revidiert der RK die Aufassung von Sprache als instruktive Informationsvermittlung, was naturgemäß Auswirkungen auf die Vermittlung wissenschaflicher Sachverhalte, wie z.B. in der Ausbildung, hat. Ferner ist der RK bemüht, die wissenschaftsgeschichtlich auftretenden Paradigmenwechsel zu erklären, da nicht der ontologische Stellenwert des Wissens im Vordergrund steht, sondern seine Generierung. Letztlich bietet der RK Argumente sowohl gegen ein "Anything goes", welches jegliche Form der Erkenntnisgewinnung der wissenschaftlichen gleichzustellen trachtet, aber auch gegen authoritäre Wissensansprüche. Dies beinhaltet auch eine Selbstanwendung des RK auf sich selbst, wodurch der RK notwendigerweise "radikal" sein muß.
[1] Searle, J. R. (1999) The Future of Philosophy. Philosophical Transactions: Biological Sciences (The Royal Society), vol. 354, no. 1392, pp. 2069-2080.
[2] Riegler, A. (2001) Towards a Radical Constructivist Understanding of Science. Foundations of Science, special issue on "The Impact of Radical Constructivism on Science", vol. 6, no. 1. In press.

We can say that from a RC perspective the purpose of science is not to seek for truth or to map out 'reality'. As Glasersfeld put it, science is the attempt to make sense in the manifold of experiences, to look out for regularities. As scientists we do this with greater rigor than in everyday situations, but this is a rather gradual than fundamental difference. At both ends of the continuum of common-sense and scientific cognition we employ a set of mental scaffolds. As findings from psychology (and ethology) show, scaffolds guide us (and animals) in what we perceive and how we reason. We are not passively flooded by information from the outside. Rather, as RC suggests, we actively construct our world. This constructivist view on science goes beyond the conception of instrumentalism which acknowledges only the utility of theories as science as sole raison d'etre. The mental scaffolds in a scientist's mind form an operationally closed system. No semantics from an outside reality is intruded upon them. They create meaning by relating to each other rather than referring to the metaphysical (and possibly science-impeding) concept of a mind-independent reality.

There is clear evidence that the ability to anticipate future events does not rely on an internal model which evolve faster than the system it models. Rather, anticipations are the result of internal canalizations which inevitably 'force' a particular path, whether in the physical or the abstract realm. Neisser's characterization of perception as a schemata controlled 'information pickup' corresponds to this perspective. An organism's schemata determine the way it is looking at the environment. The schemata construct anticipations of what to expect and thus enable the organism to actually perceive the expected information. Based on the constructivist concept of 'operationally closed systems', I arrived at a similar conclusion which I called the 'constructivist-anticipatory principle'. While the information processing metaphor creates a 'computational bottleneck' by claiming that the flow of information is unidirectional from perception to action, this principle reverses the proceedings. It is not the entirely available information from outside which is filtered for relevant issues in order to control the behavior of an organism. Instead, the constructivist-anticipatory algorithm guides perception and behavior. A set of internal hypotheses contain expectations the organism has with respect to the environment. Once such a hypothesis is invoked, it asks for sensory data only at certain moments in order to verify the validity of the hypothesis. In other words, instead of trying to reduce the complexity of all sensory data, the schemata merely need to check whether a sensory detail is present. Conceptual models of this sort can account for the fact that we find anticipations not only in humans but also in animals of various levels of complexity. Furthermore, due to the algorithmic nature of the constructivist-anticipatory principle, it provides the possibility to implement anticipatory behavior in artifacts which might help us to further investigate the working of anticipation.

Traditionally, philosophy of science has been concerned with questions about the logic of scientific investigation. It provided explanations of historical changes in research methodologies--most prominently based on sociological models--and it tried to be a normative instrument which tells scientists how to carry out science.
Recently philosophy of science has turned towards the psychological aspects of scientific praxis, i.e., it has started to focus on the people who are carrying out science. The basic assumption is that the some general cognitive processes serve as a vehicle for both scientific and non-scientific thinking. Not only can such a psychology of science account for the creative aspects of science--as opposed to a merely rationalist-logical system in which according to its definition as deductive system no creativity is possible--it also enables us to think about another perspective: Can machines perform creative science as well? Can a psychology of science provide insights and mechanisms which--in the long run--can be automatized and therefore passed over to artificial artifacts which then will carry out scientific reasoning? Computers are already used as a supportive tool in virtually any discipline, be it “number crunchers” in mathematics and physics, or as databases to store large amounts of data, in both cases they are, again, used at the deductive end of scientific activity. An original example is the proof of the famed four-color conjecture which demonstrated that using the power of hundreds of hours of computation on supercomputers are necessary to calculate individual cases of the problem is an alternative to proving the problem in a traditional mathematical way. But this program did not come up with the four-color conjecture in the fist place. It merely tracked it unremittingly down.
So what about the generation of new hypotheses and, in a larger framework, of entirely new theories? Proposals have been around for many decades already, such as the program “Bacon”. I will argue that these systems, while being a first important step, are flawed in the sense that they are artificial intelligence programs whose input is fed by humans and whose computational output is interpreted by humans. In order to act creatively it is not sufficient to simply combine existing pieces in a unique way. Furthermore, artifacts must be able to communicate their insights with humans; they must be able to deliver scientific explanations. What constitutes for a “sufficient” explanations? Are there merely logical or also psychological criteria that define a scientific explanation?
My conclusions will be that the purpose of science is not to seek for the Truth, as the matter of science is not mapping out a “reality”--regardless of how much this idea serves as a psychological motivation for people to become scientists. Rather, it consists of fairly sophisticated scaffolds which both permit predictions and create explanations.

In diesem Vortrag wird die Relevanz asymmetrischer Prozesse in lebenden Systemen sowohl in prozeduraler als auch in struktureller Hinsicht diskutiert. Verhalten wird hier nicht als ahistorischer logischer Problemlösungsprozess verstanden; es wird vielmehr gezeigt, dass Verhalten – und damit Kognition – durch Antizipation gesteuert werden. Auf struktureller Ebene ist evident, dass ein Zuwachs an Verhaltenskompetenz sich ähnlich wie bei biologischen Strukturen immer auf bereits vorhandenen Komponenten gründen muss. Aber auch Reversibilität ist möglich. Dadurch ergeben sich Auswirkungen auf die Frage, ob Evolution gleichbedeutend mit Fortschritt ist, und auf das Design von künstlichen Systemen.

(1) I will outline implementation principles for serendipitous web agents that cope with information overload in an expectation-driven way. In contrast to information retrieval agents, the agents' goal is to find out general systematic properties of web links which are hidden from a human web designers and users. (2) The recent End of Science affair triggered by John Horgan reminds us that we have to seriously think about the possibility that the progress in human science will decay and finally arrive at a cognitive barrier. In contrast to Horgan's romantic view of science, according to which we have to seek for The Truth, the matter of science is not the reality. Rather, it consists of fairly sophisticated scaffolds which both permit predictions and create meanings.

In this talk I will point out the importance of the three concepts, how they relate to each other and how they contribute to cognitive processes ranging from artificial life to distributed information systems.
In general, canalization is the restriction of developmental diversification (or complexification) due to system-internal functional dependencies among components. As such, it accounts for to evolutionary phenomena as well as ontogenetic (learning) aspects.
In particular, I will outline implementation principles for serendipitous web agents that cope with information overload in an expectation-driven way. In contrast to information retrieval agents, the agents' goal is to find out general systematic properties of web links which are hidden from a human web designers and users.