In article <2lguiv$96b at usenet.INS.CWRU.Edu>, ery2 at po.CWRU.Edu (Edwin R. Yeh) writes:
|>|> Dear bionet.neuroscience readers,
|>|> Mahowald and Douglas described and implemented a silicon neuron
|> in Nature. (1991: vol 354: 515-518) By using transistor components,
|> they were able to model the property of neuron firing frequency.
|> As in Mead's Analog VLSI and neural systems they implemented the
|> transistors circuits in CMOS VLSI.
|>|> I was wondering if anyone out here in bionet.neuroscience have tried to
|> implement it using discrete components to model a "single" neuron.
|> Specifically, does anyone know of any commercial transistors having
|> the same I-V characteristics (i.e. similar sigmoidal shape) as the
|> sodium and potassium conductance described in the paper.
|>|> I will appreciate any information.
The use of discrete components to model single neurons has been around
a long time. The work of Jerry Letvin in the late 50s and early 60s and
Edwin Lewis in the mid 60s involves just that. They used discrete
components to model the HH neuron and included lots of knobs and dials
to allow changing things like sodium and potassium concentrations. Since
transistors (bipolars, anyway) have an exponential relationship between
Ic and Vbe, these can be matched quite well to the exponential depencences
of the alphas and betas in HH.
Sorry I don't have any explicit references.
dr bruce parnas
brp at psychomo.arc.nasa.gov
/usr/local/Std.Disclaimer: The opinions expressed here are mine and
not those of NASA, but you probably could have guessed that.
It's not my fault. Not all of us here at NASA are Rocket Scientists