Inspired by the Brain

In 1943, two scientists — a neuroscientist and a mathematician — sat down and asked a strange question: what if you could model a brain cell with math? That question launched an idea that took sixty years to mature, and today powers the systems that recognize your face, translate your words, and recommend what you watch next. Neural networks are named after brain cells called neurons, and that origin story is worth understanding — not only because the name stuck, but because knowing where the analogy fits (and where it falls apart) will help you understand exactly what these systems can and cannot do.

How the Brain Inspired the Idea

Your brain contains roughly 86 billion neurons. Each neuron is a cell that receives signals through branch-like structures called dendrites, processes those signals in its cell body, and fires an output signal down a long fiber called an axon to the next neuron. Whether a neuron fires depends on how strong the combined incoming signals are. If they cross a threshold, the neuron fires. If they do not, it stays quiet. This firing-or-not-firing property is what the early researchers found interesting. It seemed almost like a logic gate — a switch that could be on or off. What if you built a mathematical object that worked the same way? Receive some numbers, add them up, and produce an output? That idea became the artificial neuron, and collections of artificial neurons became artificial neural networks. The name stuck. The inspiration was real. But from here, the paths of biology and computer science diverge sharply.

The 'loosely inspired' part is doing real work in that definition. Biological neurons use electrochemical signals; artificial neurons use floating-point numbers. Biological neurons can have tens of thousands of connections and change their structure physically over a lifetime; artificial neurons have fixed numbers of inputs and adjust only their weights (numbers that we will explore in the next lesson). A biological brain runs on about 20 watts of power and is embodied in a living organism with senses, hunger, and emotions. A large artificial neural network trains on thousands of specialized chips drawing megawatts of power, and it has none of those things. The analogy is useful as a starting point — it explains the vocabulary. But do not let it mislead you into thinking neural networks work the way brains work, or that they understand the world the way you do.

Where the Analogy Breaks Down

Here are the most important places where the brain comparison stops being accurate: First, scale and complexity. The human brain is not just a very large neural network. Its neurons have vastly different types, shapes, and functions. Some fire in rhythmic patterns; some are inhibitory; some are involved in attention and emotion in ways that have no counterpart in a feedforward neural network. Second, learning. The brain learns continuously throughout life through dozens of mechanisms. Artificial neural networks learn by a specific mathematical algorithm (called backpropagation, which you will study in Lesson 6) applied over a fixed training period. They do not keep learning after deployment without being retrained. Third, energy and embodiment. A brain evolved over millions of years to survive in a physical world. An artificial neural network is a text file of numbers on a server. The 'thinking' it does is matrix multiplication, not lived experience. None of this makes neural networks less impressive — their achievements are remarkable. It just means the brain analogy is a metaphor, not a blueprint.