> The First Neuron_

In 1943, at the height of World War II, two unlikely collaborators produced a paper that would quietly become the foundation of artificial intelligence. Warren McCulloch was a 45-year-old neurophysiologist at the University of Illinois, a pipe-smoking philosopher obsessed with the question of how the brain creates the mind. Walter Pitts was an 18-year-old self-taught mathematical prodigy who had run away from an abusive home in Detroit at the age of 15. Pitts had wandered into a library, found Bertrand Russell's Principia Mathematica, and mailed Russell a letter pointing out errors in it. Russell was so impressed he invited the teenager to study at Cambridge — but Pitts, penniless and undocumented, ended up in Chicago instead, sleeping under desks and attending lectures uninvited.

Their paper, "A Logical Calculus of the Ideas Immanent in Nervous Activity," proposed something radical: that the brain's neurons could be modeled as simple binary switches, firing or not firing, and that networks of these switches could compute any logical function. Each neuron received inputs, applied a threshold, and produced an output — on or off, 1 or 0. By connecting these artificial neurons together, McCulloch and Pitts showed that such networks could, in principle, compute anything a Turing machine could compute. It was the first mathematical model of a neural network, and it bridged the gap between biology and computation in a way no one had attempted before.

The timing was both perfect and terrible. Perfect because the wartime urgency around computation — code-breaking, ballistics tables, the Manhattan Project — meant that people were thinking hard about what machines could do. Terrible because in 1943, nobody had the technology to actually build what McCulloch and Pitts described. Their paper was published in the Bulletin of Mathematical Biophysics, a journal so obscure that most computer scientists and engineers never encountered it. For years, the work was largely ignored by the mainstream.

The personal aftermath was tragic. Pitts, who never earned a formal degree, fell into a deep depression after a falling-out with Norbert Wiener, the father of cybernetics, who had been his mentor at MIT. Wiener abruptly cut off all contact with McCulloch's circle in 1952 due to a personal dispute, and Pitts — fragile, brilliant, and deeply dependent on these intellectual relationships — began drinking heavily. He burned his unpublished manuscripts and withdrew from academic life. He died in 1969 at the age of 46, largely forgotten. Yet the McCulloch-Pitts neuron remains the ancestral seed of every neural network running today, from the simplest perceptron to GPT-4. Every node in every deep learning model is a direct descendant of that 1943 paper written by a middle-aged philosopher and a teenage runaway.