Artificial neuron — shared thought trail

Historical context

The first widely recognized mathematical model of an artificial neuron was introduced by Warren McCulloch and Walter Pitts in 1943. It reduced a biological neuron to a simple logical unit and showed that networks of such units could represent logical operations. This work helped establish the conceptual foundations of neural networks. Frank Rosenblatt’s perceptron, developed in the late 1950s, later extended the idea into a trainable model.

Original operation

A McCulloch–Pitts neuron received binary inputs representing whether connected neurons were active. It combined excitatory inputs and compared them with a fixed threshold; if enough were active, the neuron produced a binary output of 1, otherwise 0. An active inhibitory input could prevent firing altogether. The original model used fixed connections rather than learning its parameters from data, but networks of these neurons could implement logical functions such as AND, OR, and NOT.