Take away lessons

Point neuron: A neuron model with no spatial structure or morphology.

The leaky integrate and fire model: A model representing a neuron with a combination of a ”leaky” resistor and a capacitor. The cell membrane acts as a leaky capacitor which models the membrane capability to separate ions. A resistor is used to model the ion channels. The most abundant neuronal model for large-scale simulations.

The Izhikevich neuron model: A quadratic LIF type model with a recovery variable, able to replicate several characteristics of biological neurons while remaining computationally efficient.

The Hodgkin-Huxley model: The most famous and influential biological plausible neuron model. It models a neuron’s ion channels’ conductances in terms of activation and inactivation variables. The cell membrane is represented by a capacitor, and voltage-gated and leak ion channels. The electrochemical gradients driving the flow of ions are represented by batteries (E). The model takes into account Na+, K+ ion channels, and a leakage channel.

Model comparison. Learn the tradeoff between the number of neuronal features (e.g. spike frequency adaptation and phasic spiking) and the required computational resources (needed to simulate the model for 1ms) for the LIF, HH, and the Izhikevich models.