In this chapter we examine a dendritic model that generates arbitrary Boolean logic and then applies it to learning. In the model, pulses are recognized to be electrical solitons, pulses that propagate with neither dispersion nor amplitude reduction, owing to an assumed continuity of the dendritic membrane. Electrical solitons can be activated by excitatory neurotransmitters, but inhibitory neurotransmitters are able to stop their propagation toward a soma.

Simulations indicate that solitons can be reflected by dendritic junctions and by the soma itself. Reflected pulses “back propagate” toward the dendritic receivers but do not pass through oncoming solitons as solitons in electrical wires generally do. Colliding dendritic solitons annihilate each other, clearing the way for additional oncoming pulses. Back-propagating solitons arriving at a receiver help to regulate neurotransmitters. Like recently generated pulses, they are positively charged and effectively repel positive neurotransmitter ions. This, in combination with a negative voltage in the presynaptic bouton as its action potential finishes, pushes and pulls excitatory neurotransmitter ions away from a postsynaptic receiver and terminates an action potential.

Artificial intelligence (AI) today is beginning to accomplish scene and sound recognition, but this is only one aspect of AI as defined in this book, the ability to learn being paramount. The ready availability ...

Get Human Memory Modeled with Standard Analog and Digital Circuits: Inspiration for Man-made Computers now with the O’Reilly learning platform.

O’Reilly members experience books, live events, courses curated by job role, and more from O’Reilly and nearly 200 top publishers.