The article has been divided into three sections and proceeds from general insights and observations in innovation research across firms and scientific institutes to a narrower focus on radical innovations in science and their organizational environments and concludes with a specific example, namely Heinz von Foerster’s Biological Computer Laboratory at the University of Illinois in Urbana which in the short period of its existence between 1958 and 1975 exhibited a large number of essential characteristics for radically innovative research. In fact, it is astonishing in retrospect that Heinz von Foerster shaped the organizational design of the BCL in a way which nowadays seems to be the most promising and most fruitful configuration for a permanent proliferation of highly innovative and, at times, also highly implausible and counter-intuitive ideas, theories, mechanisms or instruments.

An extension of the calculus of indications (of G. Spencer Brown) is presented to encompass all occurrences of self-referential situations. This is done through the introduction of a third state in the form of indication, a state seen to arise autonomously by self-indication. The new extended calculus is fully developed, and some of its consequences for systems, logic and epistemology are discussed.

Key words: self-reference, self-referential systems, calculus of indications, paradoxes, autonomy

I met with Heinz von Foerster, Chairperson of the ASC Board of Trustees, last month at his home nestled in the Santa Cruz Mountains. Seated with the 88-year-old physicist – his frail body somehow persistent, eyes flashing with intellectual vigor – what emerged was a clear commitment to a set of guiding principles. Famed as a robust raconteur, von Foerster explicated his dedication to the path that has led him, with characteristic dignity, to these penultimate days he enjoys at Rattlesnake Hill in coastal California.

Excerpt from the introduction: “Can machines be intelligent?” “Can machines think and understand?” These are questions of epistemology. Since the concepts of intelligence, thinking, and understanding have been thought of until recently only in the context of mental activity in homo sapiens (or other species), these questions should only be asked when we know what we mean by intelligence and thinking, or when we have an “understanding ot “understanding.” ” The formulation in quotes suggests that the fundamental issue associated with these concepts is the epistemology of recursion, that is of concepts being applied to themselves. The issue here is not an isolated case, as indicated by the numerous attempts to grasp the logic of self-referring concepts (1), for instance self-reproduction (the reproduction of reproduction) (2), self-explanation (the explanation of explanation) (3), autonomy, i.e. self-regulation (the regulation of regulation) (4, 5), and many more.

Purpose: To point out the relevance of Heinz von Foerster’s work to modern embodied cognitive science and artificial intelligence research. Approach: The paper discusses (a) von Foerster’s contributions to understanding the limitations of the computer metaphor which has long dominated cognitive science, and (b) his theories concerning how reality is constructed in organizationally closed organisms, and what the underlying neural mechanisms are. The latter is exemplified with a simple neuro-robotic model that illustrates the constructive and anticipatory nature of memory. Findings: von Foerster’s work on the integration of a radical constructivist philosophy of knowledge construction with models of the underlying neurophysiological and sensorimotor mechanisms is still highly relevant to the understanding of embodied cognition and robotic models thereof. Value: This paper identifies conceptual contributions that von Foerster’s constructivist cybernetics can make to cognitive science’s still limited understanding of the embodiment of cognition and “representation.” Relevance: The paper addresses the relevance of radical constructivism in general, and von Foerster’s work in particular, to modern embodied cognitive science and artificial intelligence research.

Key words: Cognition, cybernetics, artificial intelligence, reality.

Situation and problem: a course‐writing team needs to converge rapidly to what it regards as: an agreed topic structure, which is keyed to agreed learning objectives, which specify relevant assessment questions in a natural learning sequence. Only then can the team members go away individually to write, knowing that everything that they write will fit together. In normal practice, this convergence is only partial: the topic structure harbors gaps, ambiguities and contradictions; the learning objectives are not keyed explicitly to the concepts in the topic structure; and questions for assessment of learners’ understanding do not directly exemplify conceptually keyed learning objectives. The result is courseware which does not help people to learn as well as it otherwise could, and which has been created with more effort than otherwise would have been needed. This paper shows how systemic methods inspired by Gordon Pask can be used to complete the necessary convergence with ease.

Key words: learning, objectives, assessment, systemic thinking.