Also known as a neurode, or artificial neuron when used in the context of an ANN. In essence, a neuron's primary function is to receive a multitude of input signals that are connected to its synapses from external sources, or from other neurons in a neural network, and produce an output signal. A typical neuron in an ANN produces a single output value (called an axon level in Netlab) that roughly represents the weighted combination of the values on its inputs. It can be connected to the inputs of other neurons, or to outside processes. The synapses modulate, or gate, the input signals connected to them by weight values before combining them with the neuron's output. In floating point math, the input values are simply multiplied by the weight values to gate them. Because these weight values can be adjusted in response to stimuli, the output represented on the neuron's axon is further modified by changes the weight values undergo during training.

In biological nervous systems, a neuron is a single cell with exaggerated signaling capabilities. The output values are represented by all-or-nothing pulses, called action potentials. In artificial neural networks a neuron is a process element that mimics some aspects and characteristics of a biological neuron. In essence, a neuron, whether biological or simulated, comprises inputs called synapses which are connection-points that connect signals from other neurons and external sources. Neurons also have outputs called axons, which carry the neuron's output signals to other neurons and external sources.