Progress PublishersMoscow

Generalisation and Cognition

The book

This introduction was written for Generalisation and Cognition, Dmitry Gorsky.

Very few books in this series carry quantifiers. Generalisation and Cognition is full of them: formulas for identity and equivalence, proofs set out from three axioms, a notation for "relevant" implication set against Bertrand Russell's notation for a law of nature, and a long passage on Lotfi Zadeh's fuzzy sets, in which a person can be tall "to an extent 0.3". It is a work of logic and the methodology of science by a specialist. D. P. Gorsky headed the department of dialectical materialism and the logic sector at the Institute of Philosophy of the USSR Academy of Sciences, as the jacket says. His Problems of Abstraction and the Formation of Concepts came out from the Academy's press in 1961, and Progress had published his Definition in English in 1981. A reader expecting a catechism will be surprised. So will a reader expecting a Soviet philosopher free to follow his own methods wherever they lead.

Start with the date, because the spine misleads. The English edition, translated by Campbell Creighton, is dated 1987, the high tide of glasnost. The Russian original was published by Mysl (the name means "Thought"; it was the state house for political and philosophical books) in 1985, the year Gorbachev took office. The text is older than either. Its references run thick through the 1970s and stop in 1982, with an article by Lazarev and Novoselov in a Nauka collection on materialist dialectics and a paper by N. Ts. Zhambaldagbaev in Filosofskie nauki (Philosophical Sciences) on models of explanation. The manuscript was finished in the last Brezhnev years or under Andropov. It contains no perestroika and no uskoreniye (acceleration); the only Party congress it cites is one Lenin addressed in 1919; it never names the CPSU. The Foreword says whom the book is for (natural scientists, sociologists, linguists, psychologists, "specialists in aesthetics and the creative intelligentsia") and states its polemical aim: the book "contains a critique of neopositivism on the problems of generalisation ... and demonstrates the illegitimacy of trying to make an absolute of logic." The commission was to carry Soviet logic of science into the international argument over Carnap, Reichenbach, Kuhn and Goodman, and to take part in it rather than preach at it.

To see why that matters, recall where Soviet logic had come from. For two decades after the purge of the Deborin school in 1931, formal logic in the USSR lived under suspicion: restored to the schools after the war, distrusted as a rival to dialectics, and in its mathematical form tainted by its origins in Russell, Carnap and the Vienna Circle. As late as the early 1950s Soviet reference works called cybernetics a reactionary pseudo-science. Gorsky names one of the people who kept the discipline alive: "The first attempts in the Soviet Union to employ definitions by abstraction outside mathematics were made by Yanovskaya in a work published in 1936." Sofya Yanovskaya, the Moscow mathematician who did more than anyone to keep mathematical logic teachable in the USSR, had applied the technique to Marx's concept of value. Gorsky takes up her programme half a century later.

Measured against that history, the book's dealings with Western thinkers are a real achievement. Carnap is cited to Meaning and Necessity and Philosophical Foundations of Physics by page. Reichenbach, Quine's "Two Dogmas of Empiricism", Stegmüller, Ackermann's 1956 paper in the Journal of Symbolic Logic, Polya, Cassirer and Husserl are all read in the original. Kuhn's Structure of Scientific Revolutions is quoted at length and with care. Gorsky sets out the thesis that rival theories are incommensurable, which he attributes to Feyerabend, Kuhn and Toulmin; grants that Newton's terms acquire new meanings inside relativity; and argues that the formulas recovered in the limiting case are extensionally identical with Newton's, so that something survives the revolution. His verdict reads like a seminar: "The arguments adduced by Kuhn to substantiate his thesis about the absence of continuity in the development of theories in a revolutionary way seem unconvincing to me." The answer is debatable; the register cannot be faulted. Nobody is called an idealist for holding a view.

The most interesting absence fits this pattern. Karl Popper is not named anywhere in the book, yet the chapter on how generalisations change is built around the illustration Popper made famous: "all swans are white", refuted by the black swans of Australia. Gorsky's conclusion is his own, and good: "The refutation of general statements of a law character by contradictory examples thus means their concretisation (the introduction of differences into the identical, and the limiting of concepts) and not a cancelling out of the results of research." The old law is not thrown away; it is restricted to the domain where it holds (swans of the genus Cygnus), and a new class is drawn for the exception. That is an answer to naive falsificationism, given without naming the man it answers. Other Progress authors argued with Popper at length in these years (Lektorsky's Subject, Object, Cognition, issued in English in 1984, is one), so the silence cannot be put down to a ban. It is the author's choice, and an odd one in a chapter on Popper's own subject.

What is best in the book is technical, and it holds up. The account of "identification abstraction", which groups objects into a class by a shared property while remembering that each member could be individualised again, gets past the circle in Locke's theory of abstraction (you must already know which things are men before you can abstract the idea of man from them) by tying the formation of concepts to practical comparison with a chosen standard; the worked cases are the discoveries of oxygen and of X-rays, each identified first by how it was produced and only later by its properties. The formal treatment of "definitions by abstraction" builds the concept of value from the relation of exchangeability and is honest about its limits: a footnote concedes that exchange is not strictly transitive, because goods spoil while they wait, so the construction holds only under an idealisation that "can be justified by pragmatic considerations". The discussion of idealisation in Capital rests on one of Lenin's best methodological sentences, from a letter to Potresov of 1899: "But the impossibility of something being realised in a pure form is not a refutation." And on the hardest question in the logic of laws, how to tell a law of nature from an accidental truth such as "All my friend's cats are black", Gorsky tries relevant logic, finds that it solves some difficulties and creates others, and concludes that "it is impossible to solve this problem fully by purely formal means." That is roughly where the international discussion stood, honestly reported.

The weakest pages are the ones where the polemic outruns the evidence. The critique of neopositivism is serious when it deals with Carnap's own texts on analytic and synthetic statements. Elsewhere it falls into caricature. Gorsky attributes to "Neopositivists (for example, Carnap)" a recommendation to merge two inconsistent experimental results into one equation by adding a variable for the number of the experiment, and extends it to absurdity: "If a person is operated on on Wednesday he will recover, but if on Friday, he will die". No reference to Carnap is attached to this passage, though he is footnoted by page everywhere else. The one place where the opponent is made ridiculous is the one place he is not cited. The doctrine's ritual shows in smaller ways too. Hume gets a fair hearing and is then dismissed with a formula: dialectical materialism "does not make an absolute of the stability of its regularities and interactions. But there is no hint of scepticism in that." The second sentence states a conclusion; it does not argue for it.

One sentence does more than ritual, and it needs correcting. In the section on applying theories in practice, after pages of careful argument that a social theory must be fitted to the particular country and period, Gorsky writes that Lenin "creatively developed the Marxian theory of socialist revolution in new conditions and drew a conclusion about the possibility of victory of the socialist revolution in one, relatively backward capitalist country taken by itself." In a section where Lenin is footnoted to volume and page, this sentence has no footnote. It is the founding claim of the doctrine of socialism in one country, and it rests on an article of 1915 that Leon Trotsky examined line by line in The Third International After Lenin (1928). His point was textual: Lenin "wrote this passage not in application to Russia. He was speaking of Europe in contrast to Russia." In the same months, Trotsky showed, Lenin was writing that Russia's coming revolution would be democratic, while the conditions for socialist revolution had matured in Western Europe. The official reading carried a statement about one field of objects to another without justification. Gorsky's third chapter has a name for that move, extrapolation, and he says it gives results close to the truth only when the connections between the two domains are known. By his own rule, the sentence fails.

That brings us to the test this series applies to every Soviet volume. Could the book's own categories be turned on the society that produced it? This book has unusually sharp categories, and every one of them stops at the Soviet border.

Take the concept of "type", which Gorsky builds from fuzzy sets and offers as the characteristic concept of the social sciences. In a classical class every member belongs completely or not at all; in a type, members belong to different degrees around a central model. "The concept of countries that belong to one socio-economic formation or another is a concept of type," he writes. Greece and Rome are the models of the slave-owning formation, and "Other states and countries could be included in this formation only to a certain extent." It is a genuinely useful idea, and the obvious next question goes unasked: to what degree did the USSR belong to the type "socialist society"? Nor does the book notice that the question had been answered, in nearly these terms, half a century earlier. In chapter 9 of The Revolution Betrayed (1936), Trotsky declined to file the Soviet Union under any finished category. He called it "a contradictory society halfway between capitalism and socialism", and anticipated the objection: "Doctrinaires will doubtless not be satisfied with this hypothetical definition. They would like categorical formulae: yes – yes, and no – no." What Gorsky describes as graded membership in a social type, Trotsky had practised as political analysis, in the one place Gorsky leaves the method unused.

The same holds for the book's other rules. "When applying a social theory in practice we renew and develop it, as soon as the object studied itself alters and develops." Gorsky's example is Lenin's theory of imperialism; no comparable renewal is offered for the stage the Soviet Union itself had reached, sixty-five years after the revolution. "The point is that what we discard may subsequently prove essential to explain the essence of the matter." What Soviet social science discarded above all was the bureaucracy as an object of study. On that question, Trotsky wrote in chapter 6 of the same book, "the Soviet statisticians, economists and publicists preserve, as we have said, an especially concentrated silence." Gorsky's statistical examples concern electric razors and the proportion of men on a collective farm. His example of a constructively defined class is the electorate, cut off at age 18, and he pauses to note that "many Soviet youths who had not reached 18 during the Civil War and the Great Patriotic War performed heroic deeds for the Motherland". He does not pause to note that until 1989 that electorate was offered one candidate per seat. His first methodological rule forbids generalising from parameters irrelevant to the essence of the matter, because "Irrelevant techniques of generalisation free workers from causal explanations of the phenomena studied". It is a good rule, and no one writing under Soviet conditions in 1982 could have turned it on Soviet society.

None of this charges Gorsky with failing to write a different book. The logic sector of the Institute of Philosophy was not a place from which anyone could publish an analysis of the Soviet state. The finding is narrower: the tools are general, the examples are chosen, and the choice marks out exactly what could not be discussed. Gorsky quotes Lenin approvingly, from "Our Programme" (1899): "We do not regard Marx's theory as something completed and inviolable". Inside the sciences, this book lives up to that sentence. Turned on Soviet society, it does not.

The final section, on individualisation and generalisation in fiction, is where the book comes closest to the state's aesthetic doctrine and where it moves most clearly away from it. The phrase "socialist realism" does not occur. Gorsky's subject is the classic Soviet problem of the "typical" character, and he treats it with real independence. He distinguishes typification, in which the general shows through a fully individual character (Pechorin, Raskolnikov, Anna Karenina), from "artistic idealisation", in which the general is brought forward and the individual simplified, and among his examples of the second are Mayakovsky, Brecht and Bulgakov's The Master and Margarita. He defends Kafka, whose world he describes exactly: "Poorly organised life, and bureaucratic morals grow here to a prodigious scale and are converted into a real phantasmagoria." He corrects Lukács, who had held that "where there is no realism and typification the artistic image disintegrates", and quotes with approval the remark that "Only in some cases does art conceal its conventionality, in others it flaunts it." Against the doctrine of the 1930s, which made realism compulsory and modernism decadent, this is a considerable loosening.

The older doctrine survives in the definitions. Artistic idealisation works "by eliminating everything from them that does not conform to the writer's ideas, aspirations, and ideals, and at the same time by adding to them what does correspond to them", which is an accurate account, offered neutrally as technique, of how the positive hero of official Soviet fiction was manufactured. A work succeeds when it meets "high ideological-artistic criteria". And "the typicalness of the persons, doings, and events described in an artistic work is determined first and foremost by the author's plan". In science the essential is fixed by the object; in art, on this account, the typical is fixed by intention, so a writer could make typical whatever the plan required. Trotsky had described where that licence led in "Art and Politics in Our Epoch" (June 1938): the "socialist" character of official Soviet painting "apparently consists in representing, in the manner of pretentious photography, events which never took place." Gorsky's theory has room for great literature and for that kind of picture alike, and no way of telling them apart. Kafka is credited with seeing bureaucratic morals grow into nightmare; the book applies the word "bureaucratic" to nothing in the Soviet Union.

What did the wider anti-Stalinist left say on these questions? The honest report begins with a negative. The local corpora (the complete English Trotsky, the Fourth International run of 1940 to 1956, the collected Novack) contain no mention of Gorsky, Yanovskaya, Zadeh or Soviet formal logic as a discipline. That press followed Zhdanov's 1947 intervention in philosophy and the Lysenko affair closely; it did not follow the Soviet logicians.

On the logic itself, though, the two sides agree more than either would have expected. Trotsky's "The ABC of Materialist Dialectics" (December 1939) concerns the very problem that opens Gorsky's book, the limits of the law of identity. A pound of sugar, Trotsky wrote, is never equal to a pound of sugar, because "a more delicate scale always discloses a difference"; his answer was the machinist's idea of tolerance, which he said applies to concepts too. Gorsky makes the same point: Leibniz's strict identity applies only to the abstract objects of mathematics, while real objects are identified by "concrete identity" within a chosen interval, and his fence of 500 metres and 2 millimetres, truly described as 500 metres for practical purposes, is Trotsky's cone bearing in a Soviet setting. George Novack's lectures to members and sympathisers of the Socialist Workers Party, delivered in 1942 and published as he revised them in 1963 in An Introduction to the Logic of Marxism (5th ed., New York: Pathfinder Press, 1971), go further. In Lecture II, "The Limitations of Formal Logic", the law of identity "gives correct results only within certain limits", set by the development of the object and "by the practical purpose in view". His example is the Hudson, plainly itself between Albany and New York, which "At its mouth beyond New York harbour ... loses its identity". A river with no sharp edge is just what Zadeh's fuzzy sets were invented to handle.

The difference lies in what each side did with the agreement. Novack treated formal logic as the logic of identity, contradiction and excluded middle, a "petrified shell" that dialectics had broken through, and in 1942, writing as William F. Warde in Fourth International (vol. 3 no. 6, June 1942), he called "mathematical rationalism, logical positivism" the "fashionable, pseudo-scientific costumes" of a relapse into reaction. Gorsky knows that formal logic had not stood still. Relevant logic, many-valued logic and fuzzy sets were attempts from inside formal logic to deal with exactly the limits the dialectical critique had pointed to, and he takes them seriously as tools. On this the Soviet logician has the better of the American philosopher: one was a specialist reading the technical literature, the other wrote popular expositions for a persecuted party. The story has its own irony. The first issue of Fourth International (vol. 1 no. 1, May 1940) carried "The Algebra of Revolution" by Jean van Heijenoort, Trotsky's former secretary, writing as J. Gerland: "Formal logic, only too often obliged to capitulate before reality, leaves the field open to mysticism." Within a decade van Heijenoort had left politics for mathematical logic, and his anthology From Frege to Gödel (1967) became a standard source for its history.

Where the two traditions part company is not over the logic at all. Novack, in Lecture I, made the law of identity do political work. He argued that the petty-bourgeois opposition in the SWP had gone wrong about the Soviet Union because "they couldn't correctly apply the law of identity in the process of the development of the Soviet Union". Trotsky ended his own lesson on tolerance with the same object: "Not a workers' state in general, but a given workers' state in a backward country in an imperialist encirclement". For both men the point of the logic was to classify the society they were fighting over. Gorsky's logic is more refined than theirs. It can classify gases, swans, nucleons and the slave-owning formation of antiquity, but in his hands it is never used to classify the Soviet Union.

So read this book as what it mostly is: a serious work on how concepts and laws are formed, by a logician who had read his opponents and could argue with them. Its chapters on abstraction, idealisation and the logic of laws can still teach a student something, and its section on fiction is more open than the official aesthetics of the previous generation allowed. Read it also for the line it would not cross. Its best sentence says that a refuted generalisation should be concretised, not abandoned. The generalisation the Soviet Union most needed to concretise was its claim to have built socialism, and that was the one generalisation this book could not examine.

Download the whole number as a PDF