Dynamics and bifurcations in a silicon neuron.

Area of Science:

• Neuroscience
• Electrical Engineering
• Computational Neuroscience

Background:

• Biological neurons exhibit complex nonlinear dynamics crucial for information processing.
• Existing silicon neuron models often lack compactness or fail to replicate specific excitability classes.
• Hodgkin-Huxley (H-H) models describe neuronal dynamics but are computationally intensive.

Purpose of the Study:

• To describe the nonlinear dynamical phenomena of a novel silicon neuron.
• To demonstrate class 2 excitability and excitation block in a compact silicon neuron.
• To enable large-scale integration of artificial neurons for network studies.

Main Methods:

• Designed a silicon neuron with one transient sodium and one activating potassium channel.
• Mimicked voltage clamp responses directly, avoiding complex Hodgkin-Huxley equations.
• Utilized floating-gate (FG) transistors for tuning channel parameters.
• Analyzed bifurcation conditions to identify optimal circuit biasing regimes.

Main Results:

• Achieved a compact silicon neuron design using six transistors and three capacitors.
• Demonstrated a subcritical Hopf-bifurcation, characteristic of class 2 excitability.
• Observed a Hopf bifurcation at higher currents, corresponding to excitation block.
• Showcased post-inhibitory rebound and frequency preference phenomena.

Conclusions:

• The silicon neuron successfully replicates key dynamical behaviors of biological neurons.
• The compact and low-power design facilitates integration of numerous neurons on a single chip.
• This work advances the development of neuromorphic computing systems for complex network simulations.