Inhibition analysis

[Sandrine]

Hi Ted, Michael,

I am trying to look at the inhibition effect (the effect of different weights) from an Inhibitory cell to a target cell (excitatory or inhibitory). There is an inhibitory (GABA synapses) connection between the two cells. And i am looking at the membrane potential mean of the target cell after action potential emission of the inhibitory cell.
(i put also an IClamp in the target cell to active the cell)


And my problem is to choose the synaptic weight?? At which weight should i stop? I thought that the weight should be between 0 and 1...is it true?? What is the real meaning of w?

I thought that when increasing the weight of the connection, the firing rate of my target cell and its postsynaptic activity would decrease. This is the case, but even is i put w = 10000.

Does w = 10000 mean something for a biophysical model??

Could you explain how do you choose the synaptic weight of your connections?
I know that it depends on the number of neurons, synapses and groups of neurons we want to connect, but isn't there any "rule" to start, and more important to have a correct dynamic of cells!!

Thanks in advance for your help,

Sandrine.

[ted]

And my problem is to choose the synaptic weight?? At which weight should i stop?

That's a bit like asking "How should I choose what to wear today?" and "How far should I
drive?" The first answer depends on what you intend to do, and the second depends on
where you are starting from and where you want to go to.

I thought that the weight should be between 0 and 1...is it true??

Depends entirely on the definition of synaptic weight. In the context of artificial spiking
cells or current-based synapses onto biophysical model cells, positive weight means
excitation and negative weight means inhibition. In the context of conductance change
synapses onto biophysical model cells, weights should always be >= 0--the synaptic
equilibrium potential determines whether the synapse is excitatory or inhibitory.

What is the real meaning of w?

For NEURON's built-in classes of artificial spiking cells, weight affects the "activation
variable" m (m is analogous to membrane potential). For the IntFire1 class, a synapse
with weight w causes an instantaneous jump from m to m + weight. For the IntFire2 and
IntFire4 classes, a synapse with weight w causes a smooth "bump" in the time course
of m which reaches a peak some time after the arrival of the synaptic even; if the cell is
initially resting, the magnitude of the peak is given by the absolute value of weight, and
its sign depends on the sign of weight.

For NEURON's built-in conductance change synaptic mechanisms ExpSyn and Exp2Syn,
weight specifies the maximum conductance change (in microsiemens) caused by a
synaptic event.

For more details, see chapter 10 of The NEURON Book, and of course the relevant
entries in the Programmer's Reference
http://www.neuron.yale.edu/neuron/stati ... l#IntFire1 etc.
and
http://www.neuron.yale.edu/neuron/stati ... tml#ExpSyn etc.

I thought that when increasing the weight of the connection, the firing rate of my target cell and its postsynaptic activity would decrease. This is the case, but even is i put w = 10000.

Does w = 10000 mean something for a biophysical model??

The effect of a synaptic conductance change on a postsynaptic cell depends on many
factors--time course and equilibrium potential of the conductance change, location of the
synapse on the postsynaptic cell, anatomy and biophysical properties of the postsynaptic
cell. 1e4 uS is a huge peak conductance, but it won't depolarize the cell much if the
equilibrium potential is near resting potential, if the decay time constant is very fast,
if the synapse is located on a small neurite far away from the spike trigger zone
(especially if cytoplasmic resistivity and/or membrane conductance is very large).

Could you explain how do you choose the synaptic weight of your connections?
I know that it depends on the number of neurons, synapses and groups of neurons we want to connect, but isn't there any "rule" to start, and more important to have a correct dynamic of cells!!

The assumptions and decisions that must be made in constructing a model depend
primarily on the hypothesis that is to be tested. WIthout any specific knowledge of your
hypothesis, I can only suggest that you examine the methods sections of network
modeling papers that address hypotheses on topics related to synaptic weight. You
might also read
11.3.4 Scaling Issues
on pages 297-298 of the chapter "Modeling of Large Networks" by Hasselmo and Kapur,
in the book "Computational Neuroscience: Realistic Modeling for Experimentalists" edited
by Erik de Schutter. Google Books has scanned this book so you may be able to read
these pages on line, for free.