Neuron
Chapter 2 · page 40
Definition
An artificial neuron is a computing unit whose name and general principle are borrowed, in simplified form, from biological neurons. In the human brain, a biological neuron receives signals (through its dendrites), combines them (in the cell body) and passes on, or does not pass on, a signal to other neurons. In a similar way, an artificial neuron receives one or more inputs, performs a calculation (usually a weighted sum), applies an activation function, then produces an output. The calculation it performs makes it possible to detect patterns.
Example
In an image recognition model, thousands of artificial neurons are organized into layers. The neurons in the first layers detect simple shapes (lines, colours), those in the middle layers recognize patterns (eyes, wheels), and those in the final layers identify the whole object (cat, car). Each neuron “activates” or “inhibits” the passing of information (to the next neuron) according to what it has learned.
Why it matters
The artificial neuron is a fundamental building block of deep learning. It is the arrangement and the adjustment of millions (even billions) of neurons, organized into networks, that give a model its capacity to analyze, to predict or to generate.
How to cite this text
Vial, S. (2026). Neuron. In A Living Lexicon of Artificial Intelligence (p. 40). Stéphane Vial, publisher, Montréal.
