The Artificial Neuron

Every face your phone recognizes, every sentence a chatbot understands, every move a game-playing AI makes — all of it traces back to a single, humble building block called an artificial neuron. It is simpler than you might expect, and understanding it completely is the foundation of everything else in this module.

Inspired by Biology — but Not a Copy

Your brain contains about 86 billion neurons. Each one receives chemical signals from thousands of other neurons, weighs those signals somehow, and either fires an electrical pulse or stays quiet. Artificial neurons borrow that idea at a high level, but they are made of math, not biology. An artificial neuron is a mathematical function. It takes in a list of numbers (its inputs), multiplies each one by a corresponding weight, adds all those products together, and produces a single output number. That is it. No mystery.

Here is a concrete example. Imagine a neuron that tries to decide how likely a student is to finish their homework tonight. It receives three inputs: Input 1: Hours of free time today. Value = 3. Input 2: How interesting the homework is (1-10 scale). Value = 7. Input 3: How tired the student is (1-10, higher = more tired). Value = 4. The neuron has three weights, one per input: Weight 1 = 0.5 (free time matters moderately) Weight 2 = 0.4 (interest matters quite a bit) Weight 3 = -0.6 (being tired reduces the chance — so the weight is negative) Weighted sum = (3 × 0.5) + (7 × 0.4) + (4 × -0.6) = 1.5 + 2.8 + (-2.4) = 1.9 The neuron outputs 1.9. A higher number signals a stronger prediction that homework gets done. A negative weight simply reduces the total when its input is large — elegant and powerful.

Inputs, Weights, and the Weighted Sum

Every piece of information entering a neuron is an input. Inputs might be pixel brightness values in an image, word scores in a sentence, or sensor readings from a robot — anything reduced to a number. Each input has exactly one weight. The weight determines that input's importance. A weight near zero means 'almost ignore this input.' A large positive weight means 'pay close attention.' A negative weight means 'this input pushes the output in the opposite direction.' The weighted sum is the dot product of the inputs and weights: multiply each pair, then add. The result is a single number that blends all the information together in a weighted way. In a real neural network, this operation happens millions of times per second across thousands of neurons.

Why One Neuron Is Not Enough

A single neuron can only draw one straight-line boundary in its input space. It can answer 'is this value above or below a threshold?' but it cannot recognize a handwritten digit, understand a sentence, or distinguish a cat from a dog. Those tasks require many neurons working together in layers — which is exactly where this module goes next. But every layer, every network, every breakthrough AI system starts here: inputs times weights, summed up, forwarded on. Master this and the rest unfolds logically.