Ok, why not go off topic... matthew thinks this thread is dumb anyway. In simulating neural activity (which I know little about), how detailed are the models of dendritic arbor and spines?
Warning, jargon below:
I mean do they attempt to model all the postsynaptic feedback factors (AMPA and NMDA receptors, various GPCRs and G-proteins, etc) to level of the cytoskeleton, or is it just calcium influx -> greater weight? Also, how is the weight of each synapse adjusted according to distance and number of branch points from the soma, is this corrected for dendritic thickness?
What would the effect on firing rate be if a certain treatment increased the proportion of distal dendritic branches while decreasing the number proximal to the soma (i.e. results of sholl analysis)? Assume same input firing rate, homogenous spine density, etc.
Warning, jargon below:
I mean do they attempt to model all the postsynaptic feedback factors (AMPA and NMDA receptors, various GPCRs and G-proteins, etc) to level of the cytoskeleton, or is it just calcium influx -> greater weight? Also, how is the weight of each synapse adjusted according to distance and number of branch points from the soma, is this corrected for dendritic thickness?
What would the effect on firing rate be if a certain treatment increased the proportion of distal dendritic branches while decreasing the number proximal to the soma (i.e. results of sholl analysis)? Assume same input firing rate, homogenous spine density, etc.
If you're genuinely interested, read everything you can about the Bluebrain project: http://bluebrain.epfl.ch/
They are using an IBM supercomputer which has enough resources to dedicate the equivalent of one laptop per neuron, each having perhaps 5,000 synaptic connections for the simulation of one cortical column. Watch the simulation of a cortical column completely built from scans of slices of a rat's brain: http://www.youtube.com/watch?v=xHi9oLzvD8E
Google PDF documents where Henry Markram is a coauthor. Watch the TED video he gave. Their goal is build a complete human brain. As for some of your technical questions, I don't have the answers.
I personally was implementing STDP and an empirical model to simulate a neuron's action potential, with very efficient storage of synaptic connections, simulating the length of axons and dendrites by storing them in a list, sorted by length, such that the program could traverse the list, adding in the delay from one to the next, and thus the program would always know the next synapse which would fire, and these pulses would accumulate to each respective receiving neuron, such that it could be calculated which neuron would fire next, thus distributing pulses further on down the line.
Ah, I actually have never heard about this project (or at least didn't pay attention if I did) but have consistently come across great papers out of Henry Markram's lab.
