Sign in

McCulloch-Pitts artificial neuron

1943Big moment

Warren McCulloch and Walter Pitts model neurons as simple logic units. Networks of them can in principle compute, the first idea of a neural network.

The story

In 1943 the neurophysiologist Warren McCulloch and the young logician Walter Pitts published 'A Logical Calculus of the Ideas Immanent in Nervous Activity' in the Bulletin of Mathematical Biophysics. They proposed a simplified model of the neuron as a switch: it receives signals from other neurons, and it fires, outputting a 1, only if enough excitatory inputs arrive and no inhibitory input blocks it. They showed that such units could act as the logic gates AND, OR and NOT, and that networks of them could in principle compute any logical expression, and, with a memory tape, anything a Turing machine could. It was the first mathematical model of a neural network.

McCulloch was a psychiatrist and brain researcher in Chicago who wanted a logic of the nervous system. Pitts was about 20, a self-taught prodigy who had studied logic with Rudolf Carnap. The idea drew on three earlier threads: Santiago Ramon y Cajal's view of the brain as a network of separate nerve cells, the finding that nerve impulses are all-or-none events, and the logic of Boole, Russell and Whitehead. Turing's 1936 paper had defined computation in the abstract, and the pair suggested that the brain might be a kind of computing machine. The paper was difficult and was written in the symbolism of Carnap's logic, so few biologists read it closely.

Its influence was large anyway. John von Neumann cited it in his 1945 report on the design of stored-program computers, and it helped start the field of cybernetics around Norbert Wiener. Its limits were also clear: real neurons are not simple binary switches, and the model had no way to learn. Donald Hebb proposed a learning rule in 1949, and Frank Rosenblatt's perceptron of 1958 added adjustable connection weights, the step toward the networks that now power machine learning. Pitts became estranged from his mentors in 1952 and withdrew from research, and he died in 1969.

Why it mattered

  • The perceptron of 1958 took the McCulloch-Pitts unit and added learning by adjusting weights.
  • Modern artificial neural networks and deep learning descend from these simple threshold units.
  • The paper encouraged a view of the brain as a computing system, which shaped cognitive science and the field of artificial intelligence.
  • Von Neumann's computer design used its language of logic and switching elements, tying biology to computer architecture.

Drafted by AI (Claude) from the sources above and checked against them by AI; no historian has reviewed this page. Spot a mistake? Suggest a correction. How pages are made

Community notes

Loading notes…

Learn more ranked by votes

BookTerrence J. Sejnowski
The Deep Learning Revolution
Search Open Library
VideoSearch YouTube
McCulloch Pitts neuron explained
Search YouTube
VideoSearch YouTube
history of neural networks 1943
Search YouTube