An Unfinished Revolution? Heinz von Foerster and the Biological Computer Laboratory – BCL 1958‐1976

Citation

Andrew, A.M. (2008), "An Unfinished Revolution? Heinz von Foerster and the Biological Computer Laboratory – BCL 1958‐1976", Kybernetes, Vol. 37 No. 6, pp. 828-830. https://doi.org/10.1108/03684920810877016

The decision to prepare this work was announced at the First Heinz von Foerster Conference in the Wittgenstein‐House, Vienna, in November 2003 (the conference reported on in Kybernetes, 33/7, 2004, pp. 1211‐15). The content is partly based on contributions to the meeting, along with much additional material including as the penultimate chapter the text of the Heinz von Foerster lecture delivered by Ranulph Glanville at the corresponding event a year later.

Following a general introduction by the editors, and a historical account by the son Tom von Foerster, the book has two parts, of nine chapters each. The first is entitled: “The BCL from inside” and the other: “The BCL from outside.” Both parts, especially the first, present valuable historical data and impressions. The first four chapters consist of interviews with, respectively, Humberto Maturana, Stafford Beer, Alfred Inselberg and Stuart Umpleby. The interview with Inselberg, a former member of Biological Computer Laboratory (BCL), along with material in a chapter by Paul Weston in the first part and another by Peter Asaro in the second, are valuable in providing a record of earlier work in BCL prior to the emergence of second‐order cybernetics (SOC).

Paul Weston briefly reviews early work on artificial neural nets that was not fruitful in itself but paved the way for discussion of more general topics of self‐organisation and self‐reference. He then describes in some detail the “numarete” machine that was able to indicate, using a cellular structure, the number of unconnected excited areas on a model retina. Another project referred to statistical analysis of properties of language, with remarkable agreement between the results for the three target languages of English, French, and German. Yet, another project was the devising of a computational data structure that was argued to correspond more closely to human processing than does standard computing practice. The advantage of the “cylinders” system, described as allowing linkages in 3D where conventional systems are essentially 2D, was demonstrated on challenging problem‐solving tasks.

Although work on artificial neurons was abandoned, a major project by Murray Babcock (who sadly died in 2000), resulted in a “Dynamic signal analyser” whose design is based on known structures of the ear and biological auditory perception. This was at the time when the term “bionics” was in vogue as indicating use of biological prototypes for man‐made devices. The analyser was highly‐parallel in operation, like the ear with numerous hairs on the basilar membrane sensitive to different frequencies, and was a valuable research tool. It is described in some detail in the chapter by Asaro.

In the historical account by Tom von Foerster, and in many other places in the book there is anecdotal material about the event‐filled lives of Heinz and Mai von Foerster also related in the little book by von Foerster and Poerksen (2002). There is much appreciation of their social activities that maintained a happy atmosphere in the group, and of the efforts they made to see that visitors felt welcome and cared for. (I can personally vouch for this.) The final chapter of the first part of the book is a review by Stuart Umpleby of various studies aimed at identifying the conditions under which research groups are most productive, and he is able to show that Heinz's running of the BCL was near‐optimal according to all of them. The historical relevance extends beyond the immediate group, and for instance Stafford Beer reminisces in considerable detail about his exchanges with Warren McCulloch.

The book is, in fact, a mine of valuable information from many points of view, and many people would regard what has been described so far as its less‐interesting aspects. Most of it is concerned with SOC and the associated idea of constructivism. SOC is given the status of a second Copernican revolution in science, and the overall title of the book supports the same general idea with a promise of more to come.

Although it is a currently unpopular view, I feel that the level of significance attached to SOC needs to be questioned. One of its main contentions is that there is no possibility of objectivity in science, but this is also implied by the well‐established view, particularly associated with Karl Popper, that a scientific theory is merely the hypothesis that currently best fits observations. The viewpoint has been epitomized by observing that a theory can never be proved, but can be disproved. There is of course, rather more to it than that, since there is undoubtedly extra subjectivity in deciding which theories are worthy of consideration, and how the goodness‐of‐fit should be judged, and some of the discussion around SOC emphasizes these aspects, for instance by observing how the nature of the universe would be described by adherents of various religions. Support for SOC has come largely from social and management studies, where the observer cannot avoid emotional involvement and must allow for differing perspectives of other participants.

There certainly seems to be a dearth of critical appraisal in discussions of SOC, and it was encouraging to find, in the interview with Stuart Umpleby that is Chapter 4 of the new book, criticism of Heinz for undue dogmatism over some issues. Stuart is an enthusiastic advocate of SOC, at least in the social context, and the fact that he engaged in critical discussions inspires respect for his views.

The book has two chapters by Ranulph Glanville, one in each part, and I found both of them disquieting. That in the first part is on “Personal wonder” and follows Heinz in the apparent suggestion that certain phenomena should inspire wonder, and reference to miracles, apparently to the exclusion of analytical curiosity. That the results of biological evolution are indeed wonderful is readily acknowledged by, for example, Richard Dawkins, but with the assumption that it is possible to at least speculate about the evolutionary processes that produced them.

The other chapter by Ranulph is the 2004 Heinz von Foerster lecture delivered in Vienna. This is entitled “Grounding difference” and develops a quite elaborate theory beginning with reference to the making of a distinction, following George Spencer‐Brown's “Laws of form.” The adoption of this as a starting‐point for consideration of intelligence seems to me to be contrary to Ashby's view of the brain as a “specialised organ of survival” according to which the means of making a distinction would only evolve if it was advantageous to the organism. This means that the distinction must be correlated with some variable having a bearing on survival, and perception of a correlation (which can be between continuous measures) can be argued to be more fundamental than the forming of a distinction.

These, of course, are issues that could be debated at length, and in commenting I am airing a favorite hobby‐horse of my own, though with the excuse that, as observed earlier, critical discussion is sadly wanting in these areas. Both Heinz von Foerster, in the book by von Foerster and Poerksen (2002), and more emphatically Ranulph Glanville in the above‐mentioned chapter, are dismissive of work under the heading of AI. This is despite the fact that Heinz discussed heuristics at length, and their importance was highlighted by early work on AI. It is certainly true, and admitted even by Marvin Minsky, that progress in AI has been disappointing (though, with characteristic assertiveness he told me a few years ago that he was working on a project that would show a way forward). My own feeling is that the relative failure of the AI effort is for reasons that are not much illuminated by SOC, but that is obviously another topic that needs to be debated at length. I feel it is important, though, to emphasize that many unsolved problems associated with first‐order cybernetics are neither solved not rendered irrelevant by the SOC bandwagon.

The book is indeed a mine of valuable information from many points of view, not least as a source of historical anecdote concerning the members of BCL (including Gordon Pask and Gotthard Günther as well as others already mentioned) and other people having interaction with it. The chapter in the first part that is based on an interview with Humberto Maturana gives a very full account of events involving him as well as Stafford Beer and Heinz at the time of the disastrous Chilean revolution in 1973. All this is in addition to the scientific content relating mainly to SOC, which should be read and judged despite my negative comments since it is quite possible that I have misinterpreted it.