When Norbert Wiener died a few months ago, one of the most original and significant—yet idiosyncratic—of contemporary scientists disappeared from the intellectual scene. He was entitled to go happy. He had lived fully and stylishly, with a flair not often found in the academic cloister. (But then M.I.T., on which he was based for forty years of his life, is nowadays scarcely a cloister.) He had the legitimate gratification of leaving behind him, not just a world-wide circle of admiring colleagues and pupils, but also a corpus of personal anecdotes motivated by the affection of those who knew him. And finally, he had reason to believe that his major innovations, both in thought and in mathematics, were by now securely established in the foundations of whole new sciences: cybernetics, information-theory, biophysics, and the rest.
There was only one fly in the ointment. By the time Wiener was approaching seventy, some people were already having second thoughts about his contribution. Had it really warranted all the earlier fuss? Were not the first claims about the significance and implications of cybernetics decidedly overdone? Had this initial enthusiasm perhaps been a temporary fad—something born of the scientists’ understandable recoil away from the military preoccupations of World War II and back to their chosen enterprise of extending human understanding? Like Edward Elgar (say) or John Galsworthy, Norbert Wiener lived long enough to sample the pains as well as the pleasures of becoming an historical figure in his own lifetime. The caravan was already moving on.
Those of us who believe in the permanent value of Wiener’s ideas are accordingly compelled to think over his achievement once again, and must try to present it more judiciously in the form of a balance-sheet, rather than a manifesto or company-prospectus. The time for hopes and promises is past. Now we must ask ourselves seriously just how many of those earlier I.O.U.s are in fact likely to be redeemed.
But first, the man: as an Englishman, I myself had little enough chance of meeting him, and I saw him in action not more than half-a-dozen times. Yet this was enough to leave in memory the impression of a unique and extraordinary man. He was (to use the word in an entirely innocent sense) the most peculiar American in my experience, and even in England I can liken him only to the late Sir Thomas Beecham. The similarities between the two men were no accident. True: they had a certain physical resemblance. Both of them were short, myopic, tubby. But their shared rotundity was more than a genetic coincidence: in both men, it marked out the cosmopolitan, the bon vivant. With it, there went a rotundity of expression in public conversation—I nearly said, monologue—which was too puckish to be called pompous, and an assumed air of prejudice and self-importance so extreme that it became a joy to observe. For so many of the barbs in which both men indulged were penetrating and well-placed; and, after all, on these occasions their tongues were never far from their cheeks:
The education of the average American child of the upper middle class is such as to guard him solicitously against the awareness of death and doom. He is brought up in an atmosphere of Santa Claus; and when he learns that Santa Claus is a myth, he cries bitterly. Indeed, he never fully accepts the removal of this deity from his Pantheon, and spends much of his later life in the search for some emotional substitute.
Wiener or Beecham? It could be either; and it in fact comes from the essay on “Progress and Entropy” in Wiener’s book, The Human Use of Human Beings.
Behind this common sententiousness lay similarities in childhood and upbringing also. Even more than Beecham, Norbert Wiener was a prodigy, whose talents and inclinations were cultivated by his father. The elder Wiener was Professor of Slavic Languages at Harvard and spared his son the hampering liberties of “life-enhancement” education. As a result, the boy was ready to enter Tufts College at the age of eleven, and he graduated Ph.D. from Harvard at eighteen. After seven more years of study and teaching in Europe and the US, he found his life’s niche at the Massachusetts Institute of Technology and there he took root, gradually building up a solid though unspectacular reputation with papers on a variety of mathematical subjects—notably, on Fourier integrais, about which he published a book in 1933.
The Second World War opened Wiener’s eyes to the wider potentialities of his subject. He was set the task of analyzing the mathematical problems involved in linking guns directly to radar target-seekers; and, on carrying through this analysis, he at once saw that he was on the track of an entirely general theory of control-linkages, whose relevance would extend far beyond the military demands of fire-control. (This urge to pursue the possibilities of abstraction and generalization to their very limit is one mark of the great mathematical originator.) The result was “cybernetics”—i.e., the general theory of control and communication—and it was in this field that Wiener was to spend the second, and more striking half of his professional career.
Cybernetics is a subject with many ramifications. Its concepts lie behind the design of all those self-monitoring factory assemblies that we refer to collectively by the term “automation”; it has important connections with the operation of electronic computers; it has made significant inroads into physiology, notably in explaining the failures of muscular control occurring in ataxia, and in permitting the construction of artificial limbs controlled by the actual nerves of the amputated stump; and it is widely hoped—indeed, expected—to throw much light on brain-functioning, and on the operations of sensory perception and discrimination. (Those who are prepared to undertake a little vigorous mental gymnastics will find an excellent brief survey of the whole field in the new book on Brains, Machines, and Mathematics by Michael A. Arbib.)
How is one to isolate the precise novelties at the heart of the theory of cybernetics, or indicate justly the true implications of Wiener’s achievement? Naturally, one turns first to the printed documents for enlightenment; but these by themselves do little to help us. To the uninitiated, Wiener’s classic papers are unusually opaque. From Galileo’s Sidereus Nuncius on, plenty of scientific classics have carried their importance on their faces: both the questions from which they start and the conclusions at which they arrive have a general significance which, being manifest, does not need to be underlined or explained. Wiener’s work is in a somewhat different case. His standard treatise on Time Series (1949) and the monograph on Nonlinear Problems in Random Theory (1958) will reveal nothing except to a trained mathematician. Even his one major book on Cybernetics (1948, with a second amplified edition in 1961), though far less technical in its general treatment, touches only intermittently on subjects which are self-evidently “important.” With nothing but these to go on, the average reader might be tempted to put Wiener’s work aside with a hidden suspicion that his reputation was inflated.
But to do this would be to miss the point. Some intellectual revolutions announce their arrival with trumpets and cymbals; but others creep in incognito (so to speak), so that it takes men some time to awaken to what has been done under their very eyes. One can cite as an example Sadi Carnot—another favored son of a famous father—that early-nineteenth-century prodigy who had just time to create a new style in physical thought before dying of cholera at the age of thirty-six. When Carnot’s Réflexions sur la Puissance Motrice du Feu (1824) were first published, those who noticed the book at all apparently took it to be merely one more contribution to the technology of steam-locomotives: after all, was it not specifically about heat-engines? Indeed, it was only ten years later, after Carnot’s death, that his fellow-scientists began to realize the theoretical implications of his ideas. Yet now, looking back from the mid-twentieth century, we can recognize the book as one of the founding papers of thermodynamics, and as the earliest clear origin of the famous “second law.” Carnot’s thermodynamics, developed subsequently by men like Clausius and Boltzmann, has long since ceased to be the mere theory of steam-traction: it has become a completely general façon de penser, part of the language of physical science. In the case of cybernetics, too, we must take care not to be misled, either by the historical origins of the new science, or by its mathematical technicalities. For the deeper significance of Wiener’s ideas (as I shall argue) lies not so much in what he explained, as in how he framed his explanations.
If we turn from Wiener’s mathematical treatises to his more popular books, how far do they help us to understand his long-term contribution? Speaking for myself, I have always found them something of a disappointment. As a popular writer, Wiener was continually tripping over his own public persona. If you want to relish him as an individual, well and good: his essays and memoirs are all highly personal—rich, unpredictable, opinionated. But if it is understanding you want, they are rather less than perfect: on the cold page, the idiosyncrasies which were so engaging in the living man become irritants. At his best, he was capable of a luminous intelligence and insight, but he rarely managed to stay at his best for long. Quirks and cranks keep breaking in. Where he could be magisterial, he becomes imperious; close argument lapses into argumentativeness; cogent illustrations turn into hobby-horse. The trouble was that the habits of monologue got the better of him, so that he lost the power, or the inclination, to present his sober authoritative judgements separately from a whole lot of irrelevant obiter dicta. To use a phrase of Matthew Arnold’s, his essays conspicuously lack that “central tone without which no philosopher can engage at the same time our interest, our affection, and our full respect.”
Where, then, are we to turn? There is a hint to be got (I believe) by looking at Wiener’s only novel, The Tempter (1959). This is a work of his last years: a sideline, comparable with Bertrand Russell’s solitary excursion into fiction. True: there is little in Wiener’s novel about the philosophical implications of cybernetics or anything like that. Yet in quite another way it is a very revealing document, for it enables us to see behind the grandiloquence of Wiener’s public manner to a more endearing personality within. As a novelist, he shows himself to be a nice-natured, good-hearted, but slightly naive human being, whose outstanding capacity as a mathematician is harnessed to about as much or as little moral profundity as most academics would claim. The story he tells is an improving one. It concerns an electrical engineer who contrives, for reasons of commercial advantage connected with a patent claim, to divert intellectual credit (and cash) from a crotchety English recluse to a Johnny-come-lately Mexican-born professor at a pretentious New England college—and who is subsequently stricken with remorse. The book carries a touching dedication:
To those inventors who have preferred the claims of truth to the gifts of fortune.
But this scorn for Mammon is at variance with the message of the actual story, which argues (as I read it), that original mathematicians—however remote they hold themselves from commerce—are entitled, not just to intellectual satisfaction, a decent living and public recognition of their talents, but also to a financial cut from the profits that other men win by the industrial application of their results. This doctrine is certainly an arguable one—it was in fact argued pretty extensively in the 1930s—but it is not the wholly unwordly conclusion to which we expected to be led, and Wiener’s moral trumpet-call is correspondingly muffled. (There is something of the same ambiguity about the ethical and political attitudes expressed in Wiener’s essays. In theory, he was a liberal cosmopolitan, as ready as Russell or Linus Pauling to denounce unthinking US orthodoxies, or to proclaim nuclear angst. But he was always too much of a Brahmin to let himself be swept up body and soul into a crusade, and his vehemence remains too theoretical to be wholly convincing.)
So let us ignore, for a moment, our natural assumption that the secret of cybernetics is necessarily complex and profound. May it not, one the contrary, be something very simple and straightforward—something as unsophisticated and easy to overlook as the secret of the Emperor’s New Clothes? And let us come at the subject from quite another direction, putting Wiener’s books aside and viewing his work against an historical background. To begin with, we may cast our minds back to the seventeenth-century situation in which the program of modern physical science was first promulgated. The “new philosophy” which “called all in doubt” was advertised at the time, sometimes by referring to its intellectual methods—as the “mathematical and experimental” philosophy—at other times by appealing to its basic explanatory analogies, as the “mechanical” philosophy. The controversies which the new philosophers provoked were in no sense a protest against their intellectual methods: despite all the current legends, mediaeval science was, when occasion offered, both mathematical and experimental. Rather, the difficulties sprang from the declared ambition to explain all the phenomena of Nature in mechanical terms. This program invited accusations of Epicureanism, and so of infidelity, against which seventeenthcentury scientists took great care to guard themselves. As the price of their freedom to develop a “mechanistic” physics, they had to compromise, and para of this compromise is symbolized in Descartes’ clear-cut distinction between Mind and Matter as two distinct “substances.” It was the physical world of material objects which would be explained in terms of analogies with machines. Physiological and psychological processes might have a mechanical aspect, in so far as they were influenced by the working of the bodily frame; but the rational soul, at any rate, must escape the grasp of all such theories.
Given this starting-point, natural science developed for two hundred years in a way which took it for granted that mental procedures were not just conceptually but absolutely distinct from mechanical processes. No doubt there were “laws of thought,” and perhaps there could even be expressed (as the Irish logician George Boole demonstrated) in algebraical form; but the analogy between the algebra of logic and the algebra of—say—Newton’s dynamics was only a passing one. Nobody could seriously think of identifying the supreme activities of Man, the roseau pensant, with the workings of any conceivable machine, and surely the very idea of a “thinking machine” involved a contradiction in terms. From this same starting-point, there began also the debate in epistemology which has continued up to our own day, based on an apparently self-evident opposition between Mind and Matter, reasoning and mechanism, noumena and phenomena, sense-data and physical objects. If the world of thought and the world of things interacted at all—it seemed—they did so only in the sensorium: that mysterious dark-room within the recesses of the human brain, linked to the external world through the television-receivers of the sense-organs, where the spiritual Ghost sat monitoring the operations of the bodily Machine.
In the early years, few people challenged the presuppositions underlying this debate. The philosophers were apparently confident that they knew just what “machines” were, or could become; and they assumed almost unanimously that “matter”—or “physical nature”—was intrinsically capable of nothing more than crude pushes-and-pulls. Nobody followed up Laurence Sterne’s satirical comment on the sensorium theory: that, if it were correct, then
nothing would have been wanting, in order to have taken a man’s character, but to have taken a chair and gone softly, as you would to a dioptrical beehive, and looked in,—viewed the soul stark-naked…then taken your pen and ink and set down nothing else but what you had seen, and could have sworn to.
“In the planet Mercury (belike),” he added, “it may be so.” The “tenements of their souls” may be “one fine body of transparent glass (bating the umbilical knot;)” but our own terrestrial minds
are wrapt up here in a dark covering of unchrystalized flesh and blood, so that if we would come to the specific characters of them, we must go some other way to work.
(Tristram Shandy’s Uncle Toby found his own sensorium as opaque to introspection as a “smoaky lantern.”) The only man, in fact, to denounce the whole dichotomy openly as a pure prejudice was that scandalous and heretical figure, Julien de la Mettrie, in his notorious essay L’Homme Machine (1747):
Non, la matière n’a rien de vil, qu’aux yeux grossiers qui la méconnoissent dans ses plus brillans Ouvrages; & la Nature n’est point une Ouvrière bornée…Je crois la pensée si peu incompatible avec la matière organisée qu’elle semble en être une propriété, telle que l’Electricité, la Faculté motrice, l’Impénétrabilité, l’Etendüe, &c.
Yet how did La Mettrie think one was to overcome this prejudice? “Look and see,” he cried—
Brisez la chaîne de vos préjugés; armez-yous du flambeau de l’Expérience, & vous ferez à la Nature l’Honneur qu’elle mérite.
We need only recognize, in all honesty, that observation presented us indifferently with a world of natural objects which might be animal, vegetable, or mineral, and might be capable of thinking, feeling, growing, multiplying, or merely moving.
Yet unaided observation has little power against an axiom of metaphysics. Nor was it enough for La Mettrie to declare, as a matter of principle, that a machine of sufficient complexity could reproduce any of those operations which, in Man, were declared to be characteristically “mental.” If the idea of a “thinking machine” was to escape from being a contradiction in terms, two further things had to be done. First, it was necessary to develop a systematic mathematical theory, capable of representing, without distinction, both the interactions of brain and limbs involved in rational behavior, and also the networks and linkages required in order to construct an artefact capable of simulating that behavior; and secondly, it was necessary to build such an artefact. (After all, the power of the “new philosophy” lay not in bare appeals to mechanical analogies, but in its method of proof by way of mathematics and experiment.) The second of these achievements, the technological one, may be the glory of IBM; but the more fundamental mathematical victory is owing above all to Norbert Wiener.
In stating Wiener’s achievement like this, I have used the word “artefact” instead of “machine” quite deliberately. For certainly, Descartes would have found difficulty in recognizing an IBM computer as being a “machine” at all. When he dismissed the idea of explaining Mind and Matter in the same mechanistic terms, “matter” chiefly meant to him iron bars and streams of water, not electric fields in transistors; while “machines” were collections of levers, springs, and pulleys operating on one another in accordance with the laws of impact. Thus, it is not just the technology of machines which has changed during the last three hundred years: our very concept of what constitutes a “machine” has been vastly expanded in the process, and as a result the old philosophical questions take on quite a new look. In his popular books, of course, Wiener himself is quite happy to play on our philosophical susceptibilities—referring offhand to “the special sort of machine known as a human being”; and even Michael Arbib claims it as a merit of cybernetics that it helps to “banish the ghost from the machine.” But to put the point this way only intensified the initial prejudice which it is the virtue of cybernetics to circumvent, and aroases our philosophical sense of loss needlessly. Once the ghost is banished from the machine, it seems, we shall be left with nothing but a machine. Yet the very reverse is the point of the matter. We should say, instead: until now, we did not realize what artefacts were capable of—that even man-made systems can have a “mentality.”
The essence of the matter lies, after all, not in any semantic debate over the question, whether an IBM computer is or is not to be called a thinking machine, but rather in Wiener’s mathematical analysis. There still remain a dozen reasons for not calling human beings “machines”—even “machines” of a very special kind. For machines are characteristically instruments: they are used as substitutes for humans or animals and are incapable of living lives of their own. Unless the world of R.U.R. comes upon us, and the artefacts set up an independent society, we can continue to regard even a supremely good chess-playing machine as something less than an autonomous creature. No: the essential step forward consisted in ignoring the whole of the traditional argument about Mind and Matter, and in simply doing what mathematicians had previously assumed could not be done.
When we study the operation of the nervous system (declared Berzelius in 1813) “the chain of our experience must always end in something inconceivable.” Wiener blandly conceived the inconceivable. A common pattern of ideas and inferences could be built up, he showed, which was capable of elucidating both the design and operation of complex electronic networks and also the interconnections of the nervous system, the brain and the muscles. It was not a question of identifying brains and machines, or even of explaining brain-processes on the analogy of mechanical processes previously understood. It was a question of working out a single system of concepts whose implications were capable of representing, equally well, the operations of the brain and the workings of a conceivable artefact. As Arbib rightly puts it,
We may not yet have modeled the mechanisms that the brain employs, but we have at least modeled possible mechanisms, and that in itself is a great stride forward.
The long-term consequences of cybernetics, both technological and intellectual, are bound to be far-reaching, but they are a matter for prophecy not prediction. Philosophically speaking, however, something can already be concluded. What Norbert Wiener did was not to banish the Ghost from the Machine. Rather, he demonstrated that, in the formal pattern of their activities, the Ghost and the Machine are one.
Note: The books referred to in this essay are: Norbert Wiener, a Time Series (M.I.T. Press, 1949; paperback reprint, 1964), Nonlinear Problems in Random Theory (M.I.T. Press 1958), Cybernetics, or Control and Communication in the Animal and the Machine (Paris: Hermann, and M.I.T. Press, 1948; revised edition, 1961), The Human Use of Human Beings (Houghton Mifflin, 1950; Anchor Books paperback, 1954), God and Golem, Inc. (M.I.T. Press, 1964), and The Tempter (Random House, 1959); and also Michael A. Arbib, Brains, Machines, and Mathematics (McGraw-Hill, 1964).
September 24, 1964