Wikinventia — Atlas of discoveries and inventions · Industrial Age

Mathematical model of the neuron — foundations of neural networks — McCulloch and Pitts

1943 AD · Transmission: Global
AITheoryNorth American

In 1943 Warren McCulloch and Walter Pitts published the article that founded artificial neural networks. They demonstrated that formal neurons — simple binary units that activate when the sum of their inputs exceeds a threshold — could be organized into networks capable of computing any logical function. The implication was radical: computation and nervous activity share a common mathematical structure. Pitts, only 19, had been essentially self-taught and was homeless when he wrote the article. The McCulloch-Pitts model does not learn — its weights are fixed — but it established the conceptual architecture on which all later neural networks would be built.

InstitutionUniversity of Chicago
Historical regionChicago, United States
Primary sourceMcCulloch, W.S. & Pitts, W. — "A Logical Calculus of the Ideas Immanent in Nervous Activity" (Bulletin of Mathematical Biophysics, 5, 115–133, 1943)
Secondary sourceRosenblatt, F. — The Perceptron: A Probabilistic Model for Information Storage and Organization in the Brain (Psychological Review, 1958)
Original languageEnglish
View this entry in the interactive atlas → View in graph →