The Model

The CANT model (Cell Assembly Network Technology) itself is based on things we know about biological neurons. It is a simplification, but hopefully only minor.

It has the following properties:
• Neurons have unidirectional connections (axons) to other neurons.
• Each neuron has activation which it has gotten from other neurons or the environment. If the activation passes a threshold, it fires and sends activation to the neurons it is connected to.
• Neurons have distance-biased connectivity. A neuron is more likely to have an axon to a neuron near it.
• There is neural activation decay. This is the leaky integrator model.
• If a Neuron fires often it fatigues.
• Axonal strength can change via Hebbian learning and Anti-Hebbian learning.