www.neuron.yale.edu

Hi!

I was just curious about this implementation I found on ModelDB: Fronto-parietal visuospatial WM model with HH cells (Edin et al 2007)

They use gap junctions to connect the soma and dendrites instead of the standard implemetation "connect" (nseg = 1 in their case).

Is more efficient, totally equivalent, or something else entirely? Just to be sure I'm not missing something...

If motivation isn't spelled out in the source code or in the paper itself, the surest way to find out what's in the mind of the implementer is . . . to ask the implementer. This one is a nice guy, and would probably tell you. My guess is that they were using NEURON to replicate a previous non-NEURON implementation, and wanted explicit control over the resistances between adjacent nodes. Fine for their purposes, but from a purely computational standpoint it is not preferable to using NEURON's "connect" in terms of stability, accuracy, or speed, and not something that I would recommend for anyone who is developing new code.