Neural_Computation
------------------

Simmulations that illustrate the application of simple kinetic models
for excitatory and inhibitory synaptic currents.  These simulations
are related to the following paper:

   Destexhe, A., Mainen, Z.F. and Sejnowski, T.J.  An efficient method for
   computing synaptic conductances based on a kinetic model of receptor binding
   Neural Computation 6: 10-14, 1994.  

   (see postscript file synapse.ps.Z)

   ( or pdf at http://cns.iaf.cnrs-gif.fr/abstracts/synapse.html )

This directory contains all the files needed to run the simulations
using the Interviews version of NEURON.  These files are commented and
should run straighforwardly, provided the NEURON simulator is
installed properly.



The kinetic synapse mechanism
-----------------------------

This mechanisms has the following properties:

1. It is based on a simple kinetic scheme of binding of transmitter on
   postsynaptic receptors.  This description has the advantage that
   it is fully compatible with the level of description used for other
   mechanisms (Hodgkin-Huxley currents, calcium diffusion, etc).

2. The mechanism gives EPSP's or IPSP's from a pulse of transmitter.
   The waveform of these PSP's is very close to EPSP's or IPSP's measured 
   experimentally, and the decay is monoexponential.  The user can set all
   the parameters corresponding to the rising phase, decay, amplitude, etc...
   (see .mod files)

3. Summation of consecutive PSP's is handled automatically by the 
   mechanism without need for an explicit event cue.  

4. Each synapse has a state variable corresponding to the fraction of 
   postsynaptic receptors in the open state.  However, the kinetics are
   first order, and so can be solved exactly.  This has the important
   advantage that it can be fit very easily to experimental recordings
   (see J. Computational Neurosci. paper).

5. Finally, this mechanism is very fast to compute. It does not require 
   solving any differential equations; at any given time only one exponential
   is calculated per synapse.  Thus, the mechanism is as fast to compute as 
   optimized versions of alpha function-based models.



How to run the simulation
-------------------------

This directory contains the files necessary to run a simulation such as
illustrated in Fig.1 of the Neural Computation paper.

To compile the demo, NEURON and INTERVIEWS must be installed and working on
the machine you are using.  Just type "nrnivmodl" to compile the mechanisms
given in the mod files (glutamate.mod and gaba.mod are the mechanisms
for glutamate and gaba synapses, and HH.mod is the Hodgkin-Huxley kinetics). 

Then, execute one of the two example files by typing:

special demo_glutamate_neuralcomputation.oc -

or

special demo_gaba_neuralcomputation.oc - 

Once the menu and graphics interface has appeared, click on "Init and Run"
button to start the simulation...






All these simulations were done using the NEURON simulator written by
Michael Hines, and which is available freely on internet via anonymous
ftp from neuron.neuro.duke.edu:/neuron.  For more information about how
to get NEURON and how to install it, please refer to the following sites:
  http://www.neuro.duke.edu/neuron/home.html
  http://www.nnc.yale.edu/HTML/YALE/NNC/neuron/neuron.html





For further information, please contact:

Alain Destexhe
The Salk Institute
Computational Neurobiology Laboratory
10010 North Torrey Pines Road
La Jolla CA 92037, USA

Department of Physiology
Laval University
Quebec G1K 7P4
Canada

http://www.cnl.salk.edu/~alain/
email: alain@fmed.ulaval.ca