## Random synaptic distribution and activation

The basics of how to develop, test, and use models.
afc
Posts: 23
Joined: Tue Jul 08, 2014 1:31 pm

### Random synaptic distribution and activation

Dear all,

I would like to know:

1) if there is an out-of-the-box solution, either using the GUI or a chunk of .hoc code, that enables one to randomly distribute a pre-determined number of synapses over a cell’s dendritic tree, and randomly activate those synapses;

or,

2) If such solution does not exists, is there a step-by-step/easy guide to solve the previous problem?

Thank you all for the availability!
ted
Posts: 5796
Joined: Wed May 18, 2005 4:50 pm
Location: Yale University School of Medicine
Contact:

### Re: Random synaptic distribution and activation

afc wrote:if there is an out-of-the-box solution, either using the GUI or a chunk of .hoc code, that enables one to randomly distribute a pre-determined number of synapses over a cell’s dendritic tree, and randomly activate those synapses
No, no, and no.
If such solution does not exists, is there a step-by-step/easy guide to solve the previous problem?
No again. The problem lies partly in the slovenliness of human language, which allows any two people to say the same words but mean quite different things, and partly in the boundless imagination and creativity of modelers who, given any supposedly "clear" goal to achieve, can often conjure up many alternative solutions that produce quite different results.

But if you provide an unambiguous statement of your particular target, I can suggest how you might reach it. With regard to model construction, do you mean that the probability that a particular segment will have a synaptic attachment should be proportional to its length, or to its surface area, or to something else? Is it permissible for a given segment to have more than one synapse attached to it? What is your definition of "random activation"--do you have a particular statistical model in mind? Negative exponential ISIs is easy; other temporal patterns may be more difficult.
afc
Posts: 23
Joined: Tue Jul 08, 2014 1:31 pm

### Re: Random synaptic distribution and activation

Dr. Carnevale,

thank you for the prompt response! The question I made “asked for a lot”, but I was just curious, and my goals require less:

Basically, I am modelling single neurons and I would like to run a few simulation where the simulated neurons have (1) a evenly (or quasi-even) distribution of a fixed number synapses along the dendritic surface, but (1.1) the surface area may vary between neurons, so the density of the synapses will be bigger or smaller depending on the dendritic surface. Something like this is easily achieved with density mechanisms, e.g. the dsyn mechanism (viewtopic.php?f=16&t=2613), but I do not know it is possible to achieve something that resembles random independent synaptic activation with such density mechanism.

What I had in mind when I said “random activation”, it was something like (1) the distributed synapses would be activated independently from each other, (2) negative exponential ISIs will definitely do the trick! (3) The overall number of synaptic activations over the stimulation interval would have to be a fixed number, that could not be exceeded.

Thank you!
ted
Posts: 5796
Joined: Wed May 18, 2005 4:50 pm
Location: Yale University School of Medicine
Contact:

### Re: Random synaptic distribution and activation

afc wrote:the surface area may vary between neurons, so the density of the synapses will be bigger or smaller depending on the dendritic surface.
Why not base synaptic density on empirical observations? Often it is assumed that spine density is ~1/um dendritic length, but close examination of some published data suggests that spine density is best specified in units of "number/um2" (e.g. 1/um2).

What do you plan to do about inhibitory inputs? There is much evidence that, at rest, there is a balance between excitatory and inhibitory synaptic currents.

What mean ISI do you propose to use for synaptic activation? The mean firing rate that comes to mind for principal neurons in mammalian cortex in vivo is about 7 Hz but I forget whether that was recorded in waking or anesthetized animals.

Whatever the real mean ISI is, if you decide that 1 synapse/um (or /um2) is way too many for a model (a reasonable decision, because otherwise your model cells will have many hundreds or thousands of synapses), your simulations may have to use ISIs that are a factor of k shorter, where k = (actual number of synapses attached to a cell) /(number that you use in your models).
Something like this is easily achieved with density mechanisms, e.g. the dsyn mechanism (viewtopic.php?f=16&t=2613), but I do not know it is possible to achieve something that resembles random independent synaptic activation with such density mechanism.
It would not be possible.
The overall number of synaptic activations over the stimulation interval would have to be a fixed number, that could not be exceeded.
And the justification for the limit is?