Difference between weight vs. reversal potential for synaptic connections

Post Reply

Difference between weight vs. reversal potential for synaptic connections

Post by sean.powell »

Good afternoon,

This is less of a question about any particular code and more just pertaining to definitions. When creating a NetCon connection using either ExpSyn or Exp2Syn, what is the difference between the .e (reversal potential) as compared to the weight (second to last parameter in the NetCon)? Intuitively, an inhibitory neuron should have a negative weight, but what does that mean for the corresponding reversal potential? To me, it would make sense that if weight=1 and e=-75, then one should get the same results when weight=-1 and e=75.
Sorry if that was a bit drawn out. For the code shown below, you can ignore everything besides the aforementioned weight.

Code: Select all

OLMCells[0].soma PyramidalCells[0].nclist.append(new NetCon(&v(0.5), AMPA, -40, 1, 0.5))
Where AMPA is represented by:

Code: Select all

objref GABA
OLMCells[0].soma {
    GABA = new Exp2Syn(0.5)
    GABA.tau1 = 30
    GABA.tau2 = 100
    GABA.e = -75
Thank you as always,
Sean Powell
Site Admin
Posts: 5919
Joined: Wed May 18, 2005 4:50 pm
Location: Yale University School of Medicine

Re: Difference between weight vs. reversal potential for synaptic connections

Post by ted »

In its most general usage, the term "synaptic weight" refers to the magnitude of the effect elicited in a postsynaptic cell by activity in a presynaptic cell. The larger the synaptic weight, the greater will be the effect on the postsynaptic cell. In the context of certain kinds of networks of artificial spiking cells, there are no membrane potentials, synaptic reversal potentials, or synaptic currents. Instead, each neural element ("cell") in the network has its own "activation state variable" which is perturbed by activity of presynaptic cells, and the sign and magnitude of the perturbation elicited by a "spike" in a particular presynaptic cell depend on the weight of the connection between that cell and its postsynaptic target cell.

(Here endeth the general discussion of synaptic weight. From this point forward everything is specific to NEURON.)

The above paragraph describes what happens in NEURON when the postsynaptic cell is an instance of an artificial spiking cell class such as one of the IntFire classes (see the Programmer's Reference documentation of IntFire1, 2, and 4).

However, if the postsynaptic cell is a biophysical model cell, i.e. constructed out of one or more sections, then the postsynaptic target must be a point process that is attached to a section, and the NMODL code that defines the properties of that point process must contain a NET_RECEIVE block that specifies what happens when a NetCon delivers an event to it. The simplest examples are provided by the ExpSyn and Exp2Syn classes, in which an event elicits a conductance transient governed by a mono- or biexponential time course. The reversal potential associated with the conductance transient is a parameter, and is a RANGE variable so each instance of the synaptic mechanism can have its own reversal potential. Read about these classes in the Programmer's Reference. The peak amplitude of the conductance change elicited by a single presynaptic spike is specified by the weight parameter of the NetCon that conveys the spike event to the synaptic target. The weight parameter of a NetCon that targets a conductance change synaptic mechanism should never be negative, unless one is modeling a cell whose synaptic ion channels violate the laws of thermodynamics.*

*--Ion flux in a channel with negative conductance would run opposite to the direction of the electrochemical gradient. The sort of thing one might hear about at the White Hart or maybe Gavagan's Bar.
Post Reply