www.neuron.yale.edu

I am a bit confused on the extracellular mechanism and after much searching I think clarification of a few questions would help greatly on moving forward.

1) Does e_extracellular need to be set for each segment, or does it act on the entire section? Based on another thread I gather it is for each segment, but I am unsure.

2) Based on the extracellular_stim_and_rec example, it seems like I should be determining a resistance from a stim source to the external part of a segment, then passing this into the "xtra" mechanism. I can not find any documentation for this mechanism, although I am not sure I need it.

3) Am I correct in my understanding that vext[2] & i_membrane are both outputs, and that the rest of the properties are inputs?

4) Ultimately, I am unclear how to impose different Istims at each segment, where each is calculated based on the difference between an external potential at a given node and it's neighboring nodes (due to a stimulus source) divided by the intracellular resistivity

JimH wrote:1) Does e_extracellular need to be set for each segment, or does it act on the entire section?

It is a range variable. As with any other range variable, simple assignment that does not
specify a particular location on the section -- e.g. secname.varname = value -- will set
varname to value along the entire length of secname

2) Based on the extracellular_stim_and_rec example, it seems like I should be determining a resistance from a stim source to the external part of a segment, then passing this into the "xtra" mechanism.

True, if you want to implement extracellular stimulation with the strategy demonstrated by
that example.

I can not find any documentation for this mechanism, although I am not sure I need it.

The documentation of xtra is in the file that defines the xtra mechanism, i.e. xtra.mod

3) Am I correct in my understanding that vext[2] & i_membrane are both outputs, and that the rest of the properties are inputs?

The values of each are computed by NEURON in the course of a simulation.

4) Ultimately, I am unclear how to impose different Istims at each segment, where each is calculated based on the difference between an external potential at a given node and it's neighboring nodes (due to a stimulus source) divided by the intracellular resistivity

I don't understand your question. There is no Istim or istim in the example you cite.

Thanks.

I am most familiar with extracellular stimulations in terms of ultimately inserting a Istim into the dV/dt equation (dV/dt = [-Iions+Istim]/c. This Istim typically varies along the segments, and with time. I thought perhaps the "extracellular" mechanism was the best way to accomplish this, although I am unsure how e_extracellular ultimately translates into the Istim parameter I am more familiar with.

Is there a better approach than "extracellular" in this case or how do I translate between the Istim I am thinking of and e_extracellular?

JimH wrote:I am most familiar with extracellular stimulations in terms of ultimately inserting a Istim into the dV/dt equation (dV/dt = [-Iions+Istim]/c. This Istim typically varies along the segments, and with time. I thought perhaps the "extracellular" mechanism was the best way to accomplish this, although I am unsure how e_extracellular ultimately translates into the Istim parameter I am more familiar with.

Extracellular stimulation affects real cells by virtue of what it does to extracellular
potential. The most direct representation of extracellular stimulation is to force
extracellular potential throughout the model to follow a user-specified time course,
and let the properties of the model cell and the extracellular field automatically take care
of the resulting transmembrane currents. This can be done in NEURON because the
extracellular mechanism gives you direct access to extracellular potential, and it is
demonstrated by the models presented in extracellular_stim_and_rec.zip . What makes
it attractive, aside from the conceptually clear parallel between the physical reality and
the computational model, are the facts that extracellular potential is easy to calculate,
and is independent of the membrane properties of the model cell.

If you don't have control over a model's extracellular potential--and most user-written
simulation software does not offer such control--you have to explicitly calculate the
stimulus-generated transmembrane currents. These currents are what you are calling
"Istim." This is a frequently used dodge that can produce perfectly valid results; for
example, it has been used in NEURON by McIntyre and Grill.

It would make no sense to combine both approaches in one model.

Great. I think I got it figured out. Just to clarify a few things that I was confused on.

For an unmyelinated model, or a myelinated model with only one layer and changing sections of myelin and nodes, the only variable that needs to be set is e_extracellular. The rest of the variables are such that setting e_extracellular and nothing else by default is equivalent to setting vext[0] or simply vext (same thing), even though this is an output of the system. This is because the defaults of the conductances are really high (low/no voltage drop from e_extracellular to vext) as well as having the capacitive values = 0, and the raxial values really high as well.

This vext is the V(n) given in the paper by McIntyre and Grill, 2000, equation 1, which they subsequently use to find an equivalent "Iint" or my "Istim" for applying to each of the segments.

JimH wrote:For an unmyelinated model, or a myelinated model with only one layer

Doesn't matter how many layers of myelin. Each layer reduces membrane capacitance
and conductance.

the only variable that needs to be set is e_extracellular

You've got the idea, although terminology could be improved. Here's a definition of terms
for the sake of clarity.

It makes sense to speak in terms of parameters and variables.
A parameter specifies properties of a model. Examples:
diameter
length
cytoplasmic resistivity
specific membrane capacitance
the various resistive and capacitive elements that are components of the extracellular
mechanism

The value of a variable is computed from the equations that describe a model. Examples:
v
i_na
the extracellular mechanism's i_membrane, vext, vext[1]

Parameters generally remain constant in the course of a simulation, but sometimes it is
useful to make a paramter be a function of time, e.g. as an input or "forcing function"
that drives a model. Such parameters are sometimes called variable parameters.
Examples:
current delivered by an IClamp
voltage applied by an SEClamp
the extracellular mechanism's e_extracellular

Back to the main discussion:
The only parameter that needs to be changed in the course of a simulation is
e_extracellular. The various xc, xg, and xraxial parameters are such that variables
vext and vext[1] will be essentially equal to e_extracellular.

This vext is the V(n) given in the paper by McIntyre and Grill, 2000, equation 1, which they subsequently use to find an equivalent "Iint" or my "Istim" for applying to each of the segments.

Yep. Which approach you use is up to you. If you were doing something closely related
to their work, especially if it required starting with one of their models, it would make
perfect sense to continue to use their strategy. No need to reinvent the wheel. But if you
were starting from scratch, you might prefer to use the transfer resistance based
approach that uses the extracellular mechanism. Both approaches will yield identical
results.