Much of this attitude must have been bound up with contemporary images of the putatively universal nature of the “scientific method,” which itself arose out of the physicists’ participation in the war effort. “Only after the inference revolution, when inferential statistics finally were considered an indispensable instrument and had to some extent mechanized inductive reasoning and decisions, did it become possible to reconsider the instruments as a theory of how the mind works” (Gigerenzer & Murray, 1987, p. 58). In many ways, the version of information processing so favored at Cowles initially bore a closer resemblance to “old-fashioned” British OR than it did to any innovations linked to von Neumann’s computer and American OR. The trick in managing the straddle between Walras and the cyborgs was to hint at the brave new world of information processing — namely, the tradition sketched in Chapter 2 from thermodynamics to information to command and control and connectionist theories of the brain — but, when writing down the model, hew as closely to previous econometric formalisms as possible. 53 There was, however, at least one salient wartime precedent for “guessing machines” being conflated with information processors, derived primarily from large-scale mechanized cryptanalysis, such as Turing’s Colossus or Bush’s Navy machine. Shannon’s theory of “information” clearly bore the birthmarks of this genesis, as described in Chapter 2. Nonetheless, few cyborgs long rested content with statistics as a freestanding paradigm of machine cognition. If a lesson was to be learned from wartime experience with cryptanalysis, it was that inferential statistics alone were never sufficient to constitute an effective engine of effective induction. At Bletchley Park the British came to appreciate that the problem was “to capture not just the messages but the whole enemy communication system. . . . The Hut 3 filing system, therefore, had to mirror the German system as a whole. Only when this had been done could the Enigma decrypts yield real value — not so much in juicy secret messages, but in giving general knowledge of the enemy mind” (Hodges, 1983, pp. 196– 97). This lesson was brought home time and again in areas such as pattern recognition, machine translation, and the whole range of military intelligence activities. The premature tendency to mistake a very small part of the problem of information processing for the whole, if only to reify some convenient mechanical procedure as a tidy “proof” of some optimality result or another, was stubbornly prevalent in America and became a noticeable weakness of Cowles’s intellectual trajectory from this point forward.
Here the fascination with the mechanics of statistical inference tended to defocus attention on the need for decryption to deal with issues of strategic deception, cognitive limitations, interpersonal semantic systems, and an accounting of the differential sources of randomness. But cryptanalysis and parochial notions of the scientific method are still not sufficient to account for the fascination with intuitive statistics across the disciplinary board in the 1950s and 1960s. Here, following some hints (Goldstein & Hogarth, 1997), we trace the spread of the human bet- ting machine back to — you guessed it! — some offhand ideas of John von Neumann. Briefly, it seems that he had influenced the course of postwar American psychology at least as much as American economics, and its consequences, especially at RAND, had an unanticipated feedback upon the course of events at Cowles. A number of recent works now argue that World War II was a critical turning point in the history of American psychology. 54 Although there are many facets to this transformation, the one that concerns us here is the projection of the economistic metaphor of life as a “gamble” onto all manner of previously mentalistic phenomena such as decision making, judgment, and ratiocination.
Mirowski, Philip (2001–12–03). Machine Dreams: Economics Becomes a Cyborg Science (pp. 276–277). Cambridge University Press. Kindle Edition.