Is it possible to simulate this?

[hunbe...@gmail.com]

Hi, I have attached 2 pngs here. Is it possible to simulate changes in conductaces of synapses using  such voltage-time waveform shapes as pre and post spikes (sample.png and sample2.png)? Are there any reference tutorials that  can illustrate this? Thanks

Regards,
Samuel

[Felix Kern]

Hi Samuel,
This looks like straightforward STDP to me, for which a good starting point is at the end of the second tutorial (https://brian2.readthedocs.io/en/stable/resources/tutorials/2-intro-to-brian-synapses.html).
Best, Felix

[hunbe...@gmail.com]

Hi Felix, thanks for the reply!

I dont know if my interpretation is correct. Does this mean that in such simulations, we only need the value of the dapre and dapost and the time constants from our fitted experimental data (a result of those spike-time waveforms), so when we use the Poisson neurons and LIF neuron as input and output neurons respectively, we DONT need to feed in the values of spike-time waveform into the simulation anymore since the spikes in simulation are analogous to sending in spike-time waveforms in real life experiments? 

Much thanks! Take care and stay safe!


Does this mean that in the 'simulation' we can just use poisson neuron inputs to generate the 'spikes' which is analogous to generating real voltage-time spikes in reality, meaning that we do not feed in the values of that spike waveform in the 'simulation'? 

[Felix Kern]

I'm not sure what you mean by spike-time waveform. If you mean the delta_G vs delta_t curve in your sample2.png, then yes, that's approximated by the example, though with a simplification for small delta_t. This shouldn't make a big difference unless you expect a significant proportion of your relative spike timings to fall between the two peaks.

Conversely, if you mean the action potential shape, then, welcome to the world of spiking models, where action potentials are modelled as instantaneous blips. You could, of course, model neurons with more physiological realism, but I'm not aware of any STDP models that avoid abstracting spike time to a single point in time anyway.