The Hedgehog Review

The Hedgehog Review: Vol. 20 No. 1 (Spring 2018)

Digital Metaphysics: The Cybernetic Idealism of Warren McCulloch

Leif Weatherby

The Hedgehog Review

The Hedgehog Review: Spring 2018

(Volume 20 | Issue 1)

In the autumn of 1948, Warren McCulloch, neurophysiologist, bohemian cold warrior, and a founder of machine learning, stood before a gathering of brain scientists at the California Institute of Technology. The occasion was the inaugural Hixon Symposium, and the topic, cognitive behavior—more specifically, “Cerebral Mechanisms in Behavior.” John von Neumann, for whom modern computer architecture is named, was in the audience. Questions about the new digital machines hung in the air, even as the brain remained the dominant topic. McCulloch decided to talk metaphysics.

He divided the world into “mind” and “body,” noting that the physicist claims to study only the latter, unless we compel him to include himself, as physicist, in his account of matter. Then he is faced with a choice: refuse, and remain a physicist, or assent, and become a metaphysician.1 McCulloch thought he could do this dilemma one better, by developing what he called an “experimental epistemology.” Every path forward, as he would later claim, lay “through the den of the metaphysician.”2 The physicist, in reality, has no choice but to assent, because the “synthetic a priori is the theme of all our physiological psychology,” McCulloch concluded.3

The term “synthetic a priori” is taken from the philosophy of Immanuel Kant.4 To know something a priori is to know it in isolation from experience. The category of “unity,” for instance, is something we don’t derive from experience, but rather take to experience. (Similarly, we apply causality to the world, but can’t learn about cause and effect from it.) “Synthetic,” in Kant’s vocabulary, means that what we know does not proceed from its mere concept: This kind of knowledge counts as “about” something outside itself. One example of a synthetic a priori judgment would be addition (7 + 5 = 12). Another would be the Newtonian principle that every action causes an equal and opposite reaction.

You’d hardly expect to hear technical terms from Kant’s philosophy in a setting like the Hixon Symposium, but McCulloch was telling a roomful of scientists that he was searching for the basis of any possible knowledge that is really informative about the world. Somehow, he was going to do that “experimentally,” even mathematically. The “synthetic a priori” was to be found somewhere between the brain and the new digital “automata.” By placing the physicist’s dilemma at the center of his project, McCulloch was reanimating the program of German Idealism—the philosophical movement that began with Kant and ended with Hegel. In doing so, he produced the beginning of a metaphysics we urgently need in the era of Big Data and machine learning, as the digital fades from the horizon of our control, and even our ability to grasp it conceptually.

McCulloch’s name is again in the news, in the wake of the recent “explosion” in artificial intelligence. “Machine learning”—in which software programs called “neural nets” (McCulloch had called them “nervous nets”) are exposed to millions of iterations of specified processes and build layered “knowledge” of those processes—has moved from twentieth-century fantasy to twenty-first-century fact.5 The New York Times Magazine devoted a cover story to the “new machine learning” last year, recounting parts of this history even while burying McCulloch in a hyperlink.6 But his influence is felt everywhere, in our artifacts and our algorithms. Hardware and software alike, it turns out, were part of a metaphysics McCulloch drew from German Idealism.

At the beginning of the digital revolution, there existed a speculative energy that we could use now. It was put at the service not of innovation or disruption but of maintenance and politics, of establishing categories to put our digital world on a better course. McCulloch’s evocation of Kant can show us a way to think through the balance of the digitization of our world while avoiding the extremes of digital utopianism and digital denialism.

To read the full article, please subscribe to our print ($30 yearly) or digital ($25 yearly) editions or purchase a copy at select Barnes & Noble bookstores.


  1. Warren Sturgis McCulloch, Embodiments of Mind (Cambridge, MA: MIT Press, 1965), 73.
  2. Ibid., 156.
  3. Ibid., 74.
  4. Immanuel Kant, Critique of Pure Reason, trans. Paul Guyer and Allen W. Wood (Cambridge, England: Cambridge University Press, 1998), 142–43.
  5. Ethem Alpaydin gives an excellent short overview in Machine Learning: The New AI (Cambridge, MA: MIT University Press, 2016).
  6. Gideon Lewis-Kraus, “The Great A.I. Awakening,” The New York Times Magazine, December 14, 2016,

Leif Weatherby is assistant professor of German at New York University and the author of Transplanting the Metaphysical Organ: German Romanticism between Leibniz and Marx.

Reprinted from The Hedgehog Review 20.1 (Spring 2018). This essay may not be resold, reprinted, or redistributed for compensation of any kind without prior written permission. Please contact The Hedgehog Review for further details.

Who We Are

Published three times a year by the Institute for Advanced Studies in Culture, The Hedgehog Review offers critical reflections on contemporary culture—how we shape it, and how it shapes us.

IASC Home | Research | Scholars | Events | Support

IASC Newsletter Signup

First Name Last Name Email Address

Follow Us . . . FacebookTwitter