Page 1 of 1

No capacitance in cortical cells from Bazhenov (2002)

Posted: Tue Sep 25, 2018 12:23 pm
by pascal
I am interested in coding Bazhenov's sleep model ("Sleep-wake transitions in corticothalamic system," 2002) in NEURON. (ModelDB link: https://senselab.med.yale.edu/ModelDB/s ... 189#tabs-2)

In getting started, however, I immediately stumbled upon a curious feature of the model: the cortical pyramidal cell they use has two compartments (dendritic and somatic), and the somatic compartment has no capacitance! Here are their equations for the dendritic and somatic compartments (subscripts "D" and "S," respectively):

Image

The authors of the 2002 paper state that the somatic compartment was characterized by such high conductance densities that including capacitance necessitated very small time steps for accurate integration. They therefore eliminated the capacitance in order to permit longer time steps.

My question is, can NEURON handle this? And if so, how? The Programmer's Reference says that capacitance is automatically inserted into every section, and I suspect that simply setting cm=0 will have dire consequences.

Re: No capacitance in cortical cells from Bazhenov (2002)

Posted: Tue Sep 25, 2018 1:13 pm
by ted
The equations that describe multicompartmental models of neurons can be quite stiff. Bazhenov et al. used 4th and 6th order Runge-Kutta integration methods, which are not robust in the face of stiff equations. NEURON's default integrator uses the fully implicit Euler method by design because it handles stiff equations quite well.

Re: No capacitance in cortical cells from Bazhenov (2002)

Posted: Tue Sep 25, 2018 4:29 pm
by pascal
Good to know, thanks, Ted. I will use the same specific capacitance for the soma as for the dendrite, then compare to Bazhenov et al's results.