CST 4157 Page The Previous Page The Next Page
Middlesex Logo

Lab 2: Izhekivich Neurons

A Reverberate and Fire Model

  1. Let's just try it from the shell.
    1. python3
    2. Now in python from the >>> prompt and all the way to the left import pyNN.nest as sim #or some other variable say pine
    3. sim.list_standard_models() #This shows the models that are in #your running system. You can define your own.
  2. import a python file and call
    1. . While still running python you can
      import lab2
    2. Now, walk through the commented out main. You need to have called pyNN.nest sim as the functions assume that.
    3. The lab2 functions need to be called by lab2.func like lab2.init()
    4. Once you get the png file (in today's directory), you can quit() python, uncomment the calls, and do python3 lab2.py to get a new png file like the old one.
  3. Now, we're just going to run from the lab2.py file. Go to the step 3 comment. Comment in the cells[1].b=0.3 line. Run it again. Notice this png file has the same line for 0 as the earlier ones, just a change in the voltage indices. Also note that neuron 1 gets no input.
  4. Now we're going to try to get an intrinsically bursting neuron from Izhekivich's Simple Model of Spiking Neurons. Comment in the c and d paramaters for cells[2], and run it. It looks just the same. I just picked the b parameter for the first one, and you actually get some dynamism from it, but none of the three neurons gets any external input.
  5. Comment in the useDCSource call from main. See how the IB neuron gets several spikes, then just one. The default neuron is a regularly spiking neuron. Also the one that I futzed with now spikes more (unsurprisingly).
  6. I was using Izhikevich neurons a few years back. However, I ran into a problem. Comment out the dcSource from step5 and comment in makeSynapses from 6b. It's now got inhibition and works quite reasonably if you run it. Now change the time step from the step 6a comment. Oddly, you actually get the neuron to spike. Try the printPklSpikes if you like to see. It's not an integrate and spike model but a reverberate and spike model. If it gets out of balance enough, it will spike. This was a problem for me as I was inhibitting my neurons and this was making them spike.
  7. There are lots of other types of neurons. Try to get all the ones from Izhikevich's paper.