«“Heinz von Foerster – An Appreciation” (Revisited) Bernard Scott1 Reproduced below, with a few small edits, is a paper I wrote in 1979 for the ...»
Cybernetics And Human Knowing. Vol. 10, no. 2, pp. xx-xx
“Heinz von Foerster – An Appreciation”
Reproduced below, with a few small edits, is a paper I wrote in 1979 for the International
Cybernetics Newsletter, as part of its ‘founding fathers of cybernetics’ series (Scott, 1979). I offer it
here as a contribution to this special issue of C&HK because (i) I believe my younger self did quite a
good job of overviewing von Foerster’s career (recall, he ‘retired’ in 1976) and (ii) the article is not readily available to the current generation of scholars who, I hope, will be interested in ﬁnding out about cybernetics and exploring its history. I preface the 1979 paper with a Prologue in which I set the paper in context and draw attention to aspects of von Foerster’s early work that have perhaps been overlooked in recent years, especially with the particular interest given to his later writings on second order cybernetics. There is also an Epilogue in which I comment on developments post-1979 where von Foerster’s inﬂuence has been evident, in particular, work and writings that explicitly draw on von Foerster’s distinction between a ﬁrst and second order cybernetics.
Prologue The paper below, “Heinz von Foerster: An Appreciation,” ﬁrst published in 1979 (Scott, 1979), was written when I was in my early thirties. I had by that time been a student of cybernetics for more than ten years. Like many others before and since, I was fascinated by the coming together within the cybernetics circle of a group of mutually supportive but disparate scholars of astonishing brilliance. As a way of understanding the discipline, I found myself summarizing for myself the main ideas and achievements of those scholars. I then proposed to Paul Hanika, editor of the International Cybernetics Newsletter, that he should commission and publish a series of papers on the ‘founding fathers of cybernetic’. He agreed this was a good idea. My paper about von Foerster was the ﬁrst in this series and was followed by a two part article about Gordon Pask (Scott, 1980, 1982). Sadly, Hanika died not long after and the series came to a halt. In the absence of other contributions, I was preparing short papers about Warren McCulloch, Ross Ashby and Norbert Wiener. Perhaps one day they, too, will see the light of day.
My article about von Foerster covered much ground in a short space. To do this, I found myself writing in a terse, aphoristic manner. To some extent, I took Heinz’s own writings as a model, though I cannot claim that I achieved the same level of masterly clarity.
I have commented elsewhere on von Foerster’s work (Scott, 1996). More recently, I have written a short paper entitled “Second order cybernetics: an historical introduction,” in which I comment on developments that draw their inspiration from
von Foerster’s work (Scott, in press). It is interesting that, although it refers to the text in which von Foerster ﬁrst made his famous distinction between a ﬁrst and a second order cybernetics and clearly appreciates von Foerster’s reﬂexive goal of “explaining the observer to himself,” the 1979 paper does not highlight the ﬁrst and second order distinction and its terminology. It took a while for that distinction and that terminology to become standard usage in cybernetics.
The reader should perhaps also note that the distinction between ﬁrst and second order cybernetics, where ﬁrst order study is of observed systems and second order is of observing systems, was made almost as an aside in 1974, just two years before von Foerster’s retirement. In his later years, von Foerster became celebrated for having made that distinction and continued to write and present papers on that theme. See von Foerster and Poerksen (2002) for an account of von Foerster’s latter day interests and activities.
The reader might like to know that when von Foerster read my tribute to him, he sent me a telegram, saying, “Congratulations! You have reversed evolution. Now the sons are inventing the fathers!” I am now wondering if there is some wise truth in this comment of Heinz. In the 1979 article, I refer to Heinz’s bringing a European sensibility to the more pragmatic American scene. In von Foerster and Poerksen (2002), he makes similar comment about himself (pp. 109-110).
My 1979 article covers much of von Foerster’s work prior to 1974. As von Foerster makes clear, his early papers on molecular computation draw directly on Schroedinger’s (1944, 1958) classic texts What is Life and Mind and Matter. His later ethical work also draws upon the same source. Von Foerster’s ethical principle that the observer should “enter the domain of his own descriptions” is an inverse reading of Schroedinger’s statement of the Principle of Objectivation, the principle that the scientist in pursuit of objectivity necessarily does not include himself in his domain of descriptions (Schroedinger, 1967, pp. 126-137).
In the early papers, we also see very clear evidence of von Foerster’s liberal view of what is ‘computation’, in particular ‘biological computation’, a topic also discussed in von Foerster and Poerksen (2002, pp. 54-59 & pp. 106-109). He adopts a far broader view than that used, for example, to characterize the operation of the general purpose digital computer. This broader view of computation very much parallels that of Warren McCulloch. Indeed it is evident that the early cyberneticians as a group spoke of computation in this broad biological way. See, for example, the fascinating discussions in Yovits and Cameron (1960) about the computations carried out by slime moulds as they migrate from one location to another. See especially the paper by Gordon Pask on “The natural history of networks” (Pask, 1960).
What is also evident in von Foerster’s early work are parallel and complementary interests in the dynamics of self-organization and the forms of computation that constitute cognition in such systems. Following Pask’s usage, I like to refer to the former as “macro-level theory” and the latter as “micro-level theory.” The important point to appreciate is that functional models of cognition with information ﬂows between parts of the system are static representations of structures that are continually 154 Bernard Scott being reconstituted, reproduced within the overall dynamics of self-organization.
These relationships between the macro, the global dynamics, and the micro, the structure and functions, are fully brought out in the cybernetic literature by Maturana and Varela, with their conceptions of autopoiesis and organizational closure, and, especially by Gordon Pask.
Gordon Pask, over a period of years, developed cybernetic models of learning processes which explicitly had macro and micro aspects. In his later writing, these models form the basis for his extended development of conversation theory. I was a collaborator with Pask on much of this work (see Scott, 1993). My own Ph.D. work carried out under Pask’s supervision, includes a complex model of skill acquisition.
Micro-level structures are concept schemas and perceptual-motor procedural ‘operators’ that interact and transform as part of the evolution of accurate, fast-acting automatically applied subskills. Macro-level dynamic procedures allocate processing resources to sub-processes and manage the selection processes whereby successfully applied operators ‘survive’ and unsuccessful ones ‘die’ (Scott, 1976). (As an aside, John Holland’s model of ‘robust adaptation’, with which the reader may be familiar, is homomorphic to this model and similarly motivated; see, Holland, 1975.) Von Foerster clearly appreciates that cognition as a process is holistic. This is evidenced by his often repeated explication that ‘memory’ should not be confused with mere ‘storage’—that, in fact, perception, cognition and memory are indissoluble aspects of one integral cognitive system. In later years, following Varela and others, the term ‘enactive’ has been adopted to capture this view of cognition as an embodied holistic process, rather than as a static set of structures where information in various “representational forms” is passed around. I have commented elsewhere on how, as the ‘new’ paradigm of cognitive science was set up, the AI “symbolic representationalist” view of computation, as expounded by Minsky, Boden and others, threw out the baby of self-organization with the bath water of stuff deemed to be ‘oldfashioned’. In Scott (2000), in a discussion of Boden’s critique of the work of Jean
Piaget, I expressed my concerns thus:
Piaget, on many occasions, acknowledges that his underlying model is cybernetic in character. (See Boden, 1979, Chapter 7, for an extended discussion of Piaget’s afﬁliations with cybernetics). Other than acknowledge the existence of self-regulative properties as a necessary precursor to cognitive development, he gives no satisfactory account of them. Indeed, so often has his concept of “equilibration” been criticized that Boden (op. cit.) argues that it would be better if he cast his theory in computational, information processing form, where procedures may be spelled out in explicit detail. Although there is merit in such a recommendation, Boden confuses the issue by not only making a distinction between the “dynamics” of systems (i.e. their self-organizing properties) and the “structure” of systems (their forms of computation) but argues that the study of the latter (as evidenced by research in Artiﬁcial Intelligence, v. Boden, 1977) is the “new” cybernetics that has superseded the “classic” cybernetics. Whilst it is true historically that there has been a shift in emphasis, it is not true that the older problems and approaches have been ﬁnally resolved or found wanting. Certainly, the former is not the case. Implicit in Boden’s distinction is exactly the distinction one seeks to get to grips with: what sort of entity is it whose properties make it self organizing and lead it to evolve sophisticated cognitive methodologies for problem solving, pattern recognition, language comprehension and the like?” (Scott, 2000, pp. 980-981).
An Appreciation (Revisited) 155
The 1979 Paper: Abstract
An overview is given of Heinz von Foerster’s career to his retirement in 1976. For convenience of exposition, four phases are distinguished: (i) early work on large molecules as a medium for computation (ii) work on self-organizing systems (iii) work on memory as computation (iv) work on self-referential observing systems that gave rise to the distinction between ﬁrst and second order cybernetics. Finally, there are some thoughts on praxis: what it means, in von Foerster’s view, to be a cybernetician.
The 1979 paper: Introduction
In a short article it is not possible to present a comprehensive account of von Foerster’s work. As an alternative, I will attempt a summary of what I see as the main themes of his thought.
Reference is made to key papers where appropriate. A larger bibliography is redundant since von Foerster’s own department, the Biological Computer Laboratory at the University of Illinois, has made all his papers available on microﬁche (Wilson, 1976).
His published papers are only a partial monument to von Foerster’s work in cybernetics. He has been a major catalyst in bringing together many original thinkers in cybernetics and has helped to provide both moral and ﬁnancial support (some examples of those who have passed through the BCL are W. Ross Ashby, Gotthard Gunther, Gordon Pask, Lars Loefgren and Humberto Maturana).
In the sense of Lakatos (1970), von Foerster has helped create and sustain a research program whose aims and ideas have been embraced by a whole generation of cyberneticians. That the research program is still alive and full of promise is afﬁrmed by the increasing contributions of younger cyberneticians (Varela, Glanville and Kallikourdis are outstanding in this respect).
A mark of the esteem in which von Foerster is held by the scientiﬁc community at large is the special edition of Forum, the journal of the American Society for Cybernetics, which, to honor him on his retirement, contains a collection of papers and articles paying tribute to von Foerster’s life and works.
The decision to write this article was made before I became aware of the special issue of Forum and discovered that the task I had set myself had been anticipated by those with greater claim to be authoritative and discerning. I have had the opportunity to read two of the articles in pre-print form (Pask, 1978; Beer, 1978) and recommend them to all lovers of cybernetics. Although I have reservations about the relevance and possible redundancy of the words I write, I am persuaded by von Foerster’s own example that cybernetics is both and personal and public endeavor and welcome the opportunity and challenge of articulating my view of his view.
156 Bernard Scott
The 1979 Paper: Beginnings: Molecular Computation
Biologist, physicist, mathematician and philosopher were well met in von Foerster.
His ﬁrst papers (von Foerster, 1948, 1949) present a treatment of biological computation where he argues that the medium or fabric for cognition is available in the quantal changes and stabilities of large molecules. This was at a time when most thinkers were still working towards an understanding of the brain’s potential for computation and organization in terms of neural nets (McCulloch & Pitts, 1943) and cell assemblies (Hebb, 1949).
Already, von Foerster recognized as few before him had (a parallel contribution is Schroedinger, 1944) that the roots of communication, control and computation in biological systems are to be found in the properties and behaviors of large molecules.
Two traits of von Foerster’s style of thought are already in the fore: his elegant, pithy use of formalism and his awareness that behind all particular computational systems (macromolecules, brains and cells) lie forms of computation.
The latter trait is, of course, the hallmark of cybernetic thought, conceived as a discipline in its own right by the vision of Wiener (1948), anticipated by the British psychologist Craik (1943) as a fundamental philosophical stance and inexorably pursued as an
science by Ross Ashby (1956).