An axon carries the output of a neuron. A neuron outputs the result of its processes on its axon for connection to other neurons, or to output devices. A neuron's axon may be divergent. That is, it may fan out and connect to multiple inputs of multiple post-synaptic entities.

Neurons receive inputs from axons originating at other neurons, input devices, and themselves. These axons connect to neurons via synapses, which attach via the surface of the neuron body, its dendrite, or even its own axon.

Biological axons carry the output of a neuron (or input device) as action potentials which they convey in similar fashion to electrical signals over a wire. In biological axons, action potentials are pulses which represent levels using frequency and/or pulse width modulation.

In Netlab axonal action potentials are not represented directly (at this time). The outputs of Netlab's axons are called axon-levels and are generally considered the analog values that are conveyed via modulated pulse streams. Netlab's design is open to representing pulse mechanisms intrinsically in a future release, if such a capability is thought to be important.

In Netlab, axon connections are sometimes expressed schematically as connections to the neuron's synapse-space, either as individually connected axons (a single line), or as a bundle of axons.

Schematic notation:

       Axon 
     __   \ 
    |  |   \ 
    |  |    \ 
    |   >---------
    |  | 
    |__| 

In Schematic notational context axons are drawn as single thin lines. Color of an axon (if multiple colors are used) may be different from bundle-colors.

In Netlab's Noodle™ source code, axons are implicit in connections to other super-types, such as neurons, bundles, trainers, or output devices.