Warren McCulloch and Walter Pitts published their paper titled A logical calculus of the ideas immanent in nervous activity in 1943. This theoretical work laid the foundation for what would become the perceptron algorithm decades later.
The machine was publicly demonstrated on the 23rd of June 1960 during a press conference organized by the US Navy. Rosenblatt described his three-layered network structure in a 1958 paper presented at Mechanisation of Thought Processes symposium held in November 1958.
Marvin Minsky and Seymour Papert published their book Perceptrons in 1969 which proved single-layer networks could not learn XOR functions. Their analysis caused neural network research funding to decline significantly for approximately ten years until the 1980s resurgence.
The Mark I Perceptron featured 400 cadmium sulfide photocells arranged in a 20x20 grid called sensory units. These S-units connected randomly to 512 association units through a plugboard designed to eliminate intentional bias.
The original Mark I now resides in the Smithsonian National Museum of American History after being transferred there in 1967 under government administration. Tobermory, built between 1961 and 1967, occupied an entire room with 12,000 weights implemented using toroidal magnetic cores.