Neural Models
- A further complication is that we don't really know what neurons
do.
- There are a lot of models that have some accuracy.
- Integrate and Fire models are simple and long standing.
- Leaky IF models are relatively popular. Boltzmann machines
don't seem very popular, but work.
- The best models are compartmental models (Hodgkin and Huxley 52)
and I don't know lots about them.
- There are also a range of neural simulators (e.g. Neuron).
- There are lots (47?) of different types of neurons.
- Connectivity is sparse, and not well understood (though you can
map it.)
- It's relatively stable though there are transient changes
in synapses at a variety of scales.
- There are also gross topological features like laminar architecture,
and brain areas, and of course individual differences.
