I guess this research and information is more than 15 years old, but it’s new to me and may be of interest to others in the community here.
As I’ve been looking for efficiency gains for HTM, automated input encoding via SNN spike trains, and just general spiking neural network information, one of the authors I kept coming across was Izhikevich, who has created an extremely flexible and easily computable spiking neural network model to enables the simulation of different neurons in the human brain. Many of his papers and published works are fully available on his website.
Of particular interest is a simulation from 2005 where he simulated 1 second of what amounts to a cortical column, including the various neurotransmitters, synapses, both in the neocortex and connections from the reticular thalamic nuclei.
Even though simulating that 1 second period took 50 days in 2005, pretty sure we can cut down that time… but even closer to the point, simply analyzing those interaction in simulation may shed insight into potential areas of growth in HTM, as well as ways to make the whole system more efficient while approximating biological patterns.
Happy for any thoughts or feedback.