An analog memory circuit for spiking silicon neurons

Area of Science:

• Neuroscience
• Electrical Engineering
• Computational Neuroscience

Background:

• Traditional artificial neurons often lack dynamic memory capabilities.
• Controlling neuron excitability is crucial for simulating complex neural network functions.
• Pulsatile inputs offer a method for modulating neural circuit states.

Purpose of the Study:

• To develop a simple analog memory circuit for silicon neurons.
• To investigate the control of neuron spike firing threshold using this memory circuit.
• To demonstrate applications in neural network dynamics and learning.

Main Methods:

• Designed and implemented a novel analog memory circuit.
• Integrated the circuit into a silicon neuron model with a dendritic tree.
• Utilized pulsatile inputs to control the neuron's excitability.
• Performed experiments to evaluate circuit performance and applications.

Main Results:

• The analog memory circuit successfully controlled the silicon neuron's spike firing threshold.
• Demonstrated regulation of neuron excitability over millisecond to minute timescales.
• Showcased applications in temporal edge sharpening and bistable behavior.
• Implemented a neural network exhibiting classical conditioning-like learning.

Conclusions:

• The developed analog memory circuit provides a versatile mechanism for controlling silicon neuron dynamics.
• This circuit enables adaptive excitability, crucial for advanced neural computation.
• The findings support the development of more sophisticated neuromorphic systems capable of learning and complex temporal processing.