19. There is yet another angle on Ashby's spirituality, which I discovered as I was completing this book and have been unable to take further. In his journal on 5 August 1930 Ashby wrote: "Today, after reading de Brath (ref 88) I have made my decision, and will accept spiritualism. I think that to refuse to take the plunge when one is convinced is mere cowardice" (164). The book he cites is de Brath (1925), Psychical Research, Science, and Religion.The entry continues, "I have just read a statement by Oliver Lodge," and goes on to discuss the interaction of spirit with matter, and the brain as an "elaborate and exceedingly delicate trigger mechanism . . . the "obvious place for spirits . . . to step in and seize control" (165). On 18 October 1931, Ashby made several pages of journal notes from his readings on clairvoyance and hypnotism, reproducing quotations such as "Experiment seems to show that thoughts and words leave traces in the super-individual mind which can be recognised by the seer" (280). Thirty pages later he made more notes on his readings on clairvoyance, mediums, possession, psychiatric patients, and "supernormal knowledge" (318–23). This interest evidently persisted. On 18 February 1950 Ashby sent John Bates a list of thirty-four possible topics for discussion at future meetings of the Ratio Club, which included "18. Telepathy, extra-sensory perception, and telekinesis." (The list ended with "If all else fails: 34. The effect of alcohol on control and communication with practical work.") On 26 April Ashby sent Bates a revised list of twenty-eight suggestions, which included "12. Is 'mind' a physical 'unobservable'? If so, what corollaries may be drawn? . . . 27.Can the members agree on a conclusion about extra-sensory perception? 28. What would be the properties of a machine whose 'time' was not a real but a complex variable? Has such a system any application to certain obscure, ie spiritualistic, properties of the brain?" (Wellcome Library, Bates papers, GC/179/B.5). I thank Mick Ashby for pointing out that in a 1972 biographical sketch Ashby listed a consultancy with the Institute for Psychophysical Research in Oxford (www .rossashby.info/index.html), and that a 1968 book on out-of-the-body experiences published by the Institute thanks Ashby,Walter, and many others for advice and assistance (C. Green 1968). With spiritualism, mediums, telekinesis, and so on, we are back in the space of strange performances and altered states, as discussed in the previous chapter onWalter. The idea of the brain as a material detector for a universal field of mind and spirit runs through the work of John Smythies (chap. 3, nn. 62, 69). In the present context, we can also note that such a view can hang together with a notion of biological/spiritual evolution in which the brain latches onto and adapts to the universal field of mind, as in the work of MichaelMurphy (1992) (mentioned in chap. 6, n. 69, below).
20. Conant's obituary rightly conveys the amateur flavor of Ashby's early cybernetics, though the details should be taken with a pinch of salt. I am grateful to Jill Ashby for consulting Denis Bannister on her father's work and letting me know what she found (email, 15 April 2005). Bannister worked for Ashby at Barnwood House for just over a year and married his daughter Sally (subsequently divorced). Bannister's opinion is that the kitchen table story was one of Ashby's jokes, though he recalls that some parts of the homeostat were purchased from the post office. Bannister and Ashby's principal project was "researching enzymes concerned with ECT therapy. Only when they had spare time did they build the Homeostat which was done definitely as a sideline and definitely at Barnwood House. It took about three months after which they took it round to meetings. It was very bulky and heavy. They mixed frequently with those at the Burden and Grahame White from the Burden (also mentioned in the notebooks with Denis) also helped with the building." The journal entry in question is dated 3March 1948 and reads: "My two lads [White and Bannister] have been most useful & did a good deal of the wiring up" (p. 2433). The only other reference I have found to Ashby's source of parts for the homeostat is to the detritus of WorldWar II: "It has four ex RAF bomb control switch gear kits as its base, with four cubical aluminium boxes" (3 March 1948, p. 2341).
21. Jill Ashby (email, 8March 2005) pointed out to me a journal entry for 1 February 1943, pp. 1181–82, which refers to Ashby's key 1943 essay that was eventually published in 1945 (Ashby 1945a): "I have rewritten the previous paper, restricting it to an explanation of 'adaptation by trial and error,' keeping the maths in Part II, & have sent it to the Brit. J. Gen. Psychol. And I just don't give a damn what happens. I'm just losing interest. (Am doing astronomy at the moment). Now that my previous childish phantasies have been shown to be untrue I have lost my drive. And without something to give a drive one just can't move." The continuation of this passage nicely documents the professional precariousness of Ashby's cybernetics: "Tennent has given me to understand quite clearly that he wants nothing to do with it. It is now impossible for me to move further unless I can feel that there will be some benefit somewhere. As it is, it merely leads to trouble. Summary:Theory has been submitted for publication for the third time." Jill Ashby adds that "Dr Rambout sadly suddenly died and his position was filled by Dr Tennent who I can remember my mother saying did not like my father. Dr Tennent did not mention my father's work in the next four annual reports." Rambout was the medical superintendent when Ashby was appointed to St. Andrew's Hospital in Northampton, where he worked from 1936 to 1947. Number 12 in the "Letters and Documents" section of the Ashby archive is a letter from N. Rashevsky, 8 January 1947, rejecting a paper Ashby had submitted to the Bulletin of Mathematical Biophysics,entitled "Dynamics of the Cerebral Cortex, VII."
22. It is interesting to contrast cybernetics' institutional travails in the West with its enormous success in the Soviet Union (Gerovitch 2002). Although much remains to be explored about the substance of Soviet cybernetics, it is tempting to reflect that, ontologically, cybernetics was much more congenial to a materialistMarxist ideology than to the taken-for-granted dualism of the Western academy.
23. On the history of the BCL and its closure, see Müller and Müller (2007); Umpleby (2003); and Franchi, Güzeldere, and Minch (1995), where Franchi recalls that when he visited the University of Illinois in the mid-1980s, "I couldn't find a single person who knew of the BCL" (301). See also the BCL website, which includes valuable information and links to other sites, including several festschrifts for von Foerster: bcl.ece.uiuc.edu. On the distinctive research program of the BCL, see Asaro (2007); on von Foerster, see Brier and Glanville (2003).
24. As far as I can make out, Ashby makes the mistake of dividing by 2 instead of 1/2 when he evaluates T2 and T3, but nothing hinges on this.
25. One gets a feeling for this kind of world in some of Stanislaw Lem's fiction.
26. This last example strikes me as assuming an unwarranted Platonism about the structure of mathematics, but it still counts as a nice example of Ashby's train of thought.
27. Think, for example, of Newtonian physics or quantum field theory. All of the possible entities and their interrelations are already prescribed by the theory; there is no room even to raise the sort of question that Ashby addresses. Of course, questions of weak and strong couplings do arise in such sciences, but as matters of epistemological convenience and possibility. Coupling constants are what they are. Weakly coupled systems can be analyzed by the methods of perturbation theory; strongly coupled ones are resistant to such methods (and this gets us back, of course, to questions of complexity and predictive transparency). From another angle, Ashby's analysis reminds me of arguments in science and technology studies. There the humanist critique is that posthumanist analyses are irrevocably useless because, lacking the principle of humanist centering, they amount to a brute assertion that everything is coupled to everything else. My own reply to this critique is that units of empirical analysis are not given but have to be found in any specific instance. It turns out that one finds that some cultural elements are strongly coupled to each other—constituting the object of analysis—while indefinitely many are weakly coupled at most. For more on this, see Pickering (1997, 2005b, 2005c).
28. Ashby seems to have finally concluded that the
task was hopeless by 1957. In a journal entry dated 23 February 1957 he recorded, "Now we come to the (most unpleasant) deduction of p. 5539 [seven pages earlier]—that there is no secret trick to be discovered, no gimmick, no Ashby's principle, no ultimate ruling law:—there is only information, knowledge of what is best in each particular case. So the idea that has been nagging me ever since the 'Red mass' of p. 2957—find once for all how the discriminative feedback is to find the right step-mechanism—is now shown to be mistaken. The feedback can be directed with discrimination only after the necessary information has been collected. If then I (or other designing factor) am to get the feedback right without personal attention, I must achieve a supplementation. This is not systematic: it just means looking for any trick, using the peculiarites of the material at hand, that will meet the case. Thus, if I build a new machine & want discriminative feedback, I may use any trick that happens to be convenient (for there is no way better), but I must admit that it is ad hoc, and not claim any wider validity for it. Ingenuity—it is now up to you! Summary: 'Thinking things over' in a multi-stable system. Discriminative feedback requires mere opportunism [marginal notes forward to pp. 5549, 5584, 5601]" (pp. 5546–47, and see also following pages). "Summary:No general principle can be sufficient guide when selection must be done; some actual channel is necessary" (27 February 1957, p. 5549).
29. Ashby's journal makes clear the extent to which his thinking was importantly informed by trying to build DAMS and finding that his expectations were frustrated. An entry from 15 May 1951: "Saw Wiener today at the B.N.I., & asked him whether he could give any help towards getting a mathematical grasp of DAMS, towards solving the problem of p. 3151 (lower half). He was firm in his opinion. There is no branch of mathematics that can, at present, be of much use. The work I am doing is quite new, and the mathematician has no experience to guide his intuitions. Therefore, says Wiener, I must develop my machine empirically until I have established a clear & solid structure of factual material. It will then become clearer what are the important factors, & what are the significant questions. The mathematical attack must come after the empirical exploration" (p. 3304).
30. In the discussion of the homeostat I emphasized that Ashby's approach required to him to model the world as well as the brain, and the same went for DAMS. The earliest explicit discussion I have found of DAMS's environment comes from a journal entry on 23 September 1951. "A much more practical idea for the 'env[ironmen]t' of DAMS. My previous idea was to have parts that can be joined together in a variety of ways. I now see a simpler way of getting variety:—Build a 'machine' of, say, coils and magnets, making the relations permanent. But provide it with, say, eight inputs & eight outputs that can be joined to the machine [i.e., DAMS] at random." But by the end of this note, Ashby had gone around in a circle and concluded, "This leads naturally to the conception of just thinking of the last row of valves [in DAMS] as the 'envt' [of the rest of DAMS]" (pp. 2482–83). Just as the world of the homeostat was more homeostats, so the world of DAMS was a subset of DAMS. The brainworld symmetry is even more evident in this latter setup.
31. More than a year later, on 20 October 1951 (p. 3152) Ashby recorded: "DAMS is working quite well now with 40 valves for simple jobs. It could probably be made to work like a simple ultrastable system by just feeding back some disturbance to shatter the neons. . . . On the other hand it is showing little sign of the characteristic multistable system's power of adapting by parts and of accumulating patterns. I presume that if the machine demonstrates clearly that something can not be done with 100 valves, then that will be worth knowing!"
32. A version of the idea that DAMS should evolve by trial and error reappears in Ashby's Thalèsessay under the heading " 'Darwinian' Processes in Machinery" (Ashby 1951, 5), though there the emphasis is on spontaneous reconfigurations within the machine (and Ashby acknowledges a private communication from Wiener for his use of "Darwinian"). The perspective of the human designer returns in Ashby's notes for 20 October 1951 (pp. 3512–17). "Here is a simple and, I think, potential method for improving DAMS—a sort of sequential analysis" (p. 3512). Ashby then discusses regulating a watch by fiddling with its "index" (while not understanding how that enters into the mechanism). "The method proposed is not new: it has been used in endless trades throughout the history of civilisation for the improvement of products and processes when the conditions were too complex to allow of scientific analysis" (21 October 1951, p. 3516). He gave as examples Morocco leather, white lead, linseed oil, the motor car, and different ignition systems. "Summary: Improvement by the purely empirical is as old as industry. Corollary: If I can make substantial improvements in DAMS by purely empirical process, I shall provide facts that will give a solid basis on which later workers may found a science—just as the rubber technologists enabled the scientists eventually to build a science of high polymers" (p. 3517). Ashby's remark about the necessity of new techniques in new developments in science, which seem "plain crazy" in the old framework, are echoes of Kuhn's (1962) thoughts on the incommensurability of paradigms more than a decade avant la lettreand expressed in a more thoroughgoingly performative register. Ashby's recognition of the problems raised by machines like DAMS for notions of understanding and explanation likewise presage much later philosophical discussions of complex systems more generally (e.g., Kellert 1993, Pickering 1995).
33. In the quoted passages from both Thalèsand Design for a Brain,Ashby cites Humphrey (1933), the work discussed in note 12 above in connection with Ashby's earlier steps toward the homeostat.
34. Habituation actually surfaced in DAMS as a block toward the overall goal of accumulating adaptations. Ashby wrote in a journal note dated 1 May 1951, "During the last two or three months I have been testing the first 10-valve block of DAMS Mark 13. Whereas I expected it to be remarkably unstable I have found it to be remarkably, almost undesirably, tenacious; so much so that I had doubts whether I could ever make it learn."
35. Ashby seems to have saved this argument up for six years, before presenting it in public at the 1958 "Mechanisation of Thought Processes" conference discussed below.
36. In the next chapter we will find Stafford Beer eloquently lamenting the same situation a few years later.
37. The only such machine that I have seen written up is the one Ashby demonstrated at the "Mechanisation of Thought Processes" conference in 1958 which simulated the production and cure of mental pathologies, discussed in the next section.
38. This review appeared as an undated special issue of the Journal of Mental Science,but internal evidence suggests that it appeared in 1950. It is a measure of the visibility of cybernetics in British psychiatry by this date that both Ashby and Walter were invited to contribute to it (Walter with the entry "Electro-Encephalography"). Frederick Golla's entry "Physiological Psychology" begins with the remark "The past five years have seen an attempt to reformulate the basic conceptions . . . in terms of neural mechanics . . . Cybernetics," and then discusses Ashby's work at length, followed by that of McCulloch, Pitts, and Walter on "scanning" and Masserman's cats, before moving on to more prosaic topics (Golla 1950, 132–34).
39. One might read Ashby as asserting that the undoing of habituation by any large and random disturbance is discussed in Design for a Brain,but it is not. As far as I know, Ashby's only attempt to argue this result in public was at the "Mechanisation of Thought Processes" conference in 1958, in a paper titled "The Mechanism of Habituation" (Ashby 1959a). This includes an extended discussion of "de-habituation" (109ff.) without ever mentioning electroshock. At the same meeting, Ashby also exhibited "A Simple Computer for Demonstrating Behaviour" (Ashby 1959b). According to its settings, Ashby claimed that it could display "over forty well known pieces of biological behaviour," including various simple reflexes, "accumulation of drive," "displacement activity," and "conflict leading to oscillation" or "compromise." As exhibited, it was set to show "conflict leading to catatonia, with protection and cure" (947–48). It might be significant that
in the passage quoted Ashby uses the memory loss clinically associated with ECT to make a bridge to his cybernetic analysis of dehabituation. Though I will not explore them here, Ashby developed some very interesting cybernetic ideas on memory (Bowker 2005). It was in this period that the first discussions of memory appear in Ashby's journal. The suggestion is, then, that the concern with the mode of action of ECT might have been a surface of emergence also for Ashby's cybernetics of memory (and see, for example, the long discussion of memory that appears in Ashby 1954).
40. Walter's article on CORA appeared in the August 1951 issue of Scientific American,just one month before Ashby made this note.
The Cybernetic Brain Page 54