Weighting of synaptic responses

Matt Jones jonesmat at ohsu.edu
Thu Aug 29 13:04:14 EST 1996


In article <503cja$i7e at reader1.reader.news.ozemail.net> Josh Reich,
rogan at auswired.net writes:
>Is it possible that things like cholinesterases can be
>moderated for LTP ?

LTP is almost always studied at glutamate synapses, not ACh synapses. I
don't know whether cholinergic LTP exists. On the other hand, there is
ACh being released into places like the hippocampus where most LTP
studies have been done, and its actions there promote rhythmicity and
excitability through muscarinic receptors, which would favor LTP (someone
correct me if I'm screwing this up). So if there's cholinesterase at
those muscarinic synapses, yes altering it somehow could conceivably have
an effect on LTP through a very inderect mechanism. 

At glutamate synapses, glutamate is cleared by diffusion out of the
synapse, and by glutamate uptake transporters. The exact role of the
transporters in shaping synaptic efficacy is the subject of some pretty
interesting research (see papers by D. Attwell, C. Jahr, E. Schwartz,  or
C. Zorumski labs), but is still debated. A lot of things could
potentially modulate their function: pH, arachidonic acid, teperature,
etc.

The main debate about the mechanism of LTP continues to revolve around
the question: Is LTP a pre- or postsynaptic modification? Proposed
presynaptic mechanisms usually involve an increase in the probability
that a vesicle will be released in response to an action potential, and
postsynaptic candidates involve either insertion of new receptors or some
sort of enhancement in the function of existing ones. 

For the purpose of ANN modelling, my guess is that to a first
approximation it won't matter whether you imlplement it as pre- or
postsynaptic, because either way you'll be increasing the weight of a
local synaptic connection.  However, because of voltage-dependent
properties of the NMDA-type glutamate receptors, triggering LTP requires
that the postsynaptic cell is depolarized when the presynaptic input
arrives: Depolarization alone won't cause LTP, but simultaneous
depolarization and excitatory input can. M. Bear wrote an interesting
review in Neuron recently about a sliding threshold for LTP induction
that you might find useful when thinking about how to implement the
weighting in an ANN.

Cheers,
-Matt



More information about the Neur-sci mailing list