Two new papers

[Ted Carnevale]

The following two articles may be of interest to those
who are interested in synaptic integration, either from
the theoretical or experimental standpoint, and/or
empirically-based modeling of neurons.


The first paper will be most relevant to those who are
interested in modeling the roles of biophysical mechanisms
and use-dependent plasticity in the operation of
individual neurons and neural networks.
    Carnevale, N.T., and Hines, M.L.
    Expanding NEURON=92s repertoire of mechanisms with NMODL.
    Neural Computation 12:839-851, 2000
This is an "executive summary"; for those who need to
know the details, a preprint of this paper that includes
many more figures, examples, and an extensive index is
available from our WWW site at
  http://www.neuron.yale.edu/neuron/papers/nc99/nmodl.htm


The second paper shows that many classes of neurons
(especially nonpyramidal cells in vertebrate CNS)
are fundamentally similar to the processing elements
of artificial neural nets, in the sense that synaptic
impact at the spike trigger zone is determined by the
properties of the synapse itself, and not by the
anatomical location of the synapse.  It also challenges
the widely-held notion that active currents are necessary
to overcome location-dependent attenuation of synaptic
inputs (with one important exception, as noted below).

Passive normalization of synaptic integration influenced
by dendritic architecture.
David B. Jaffe and Nicholas T. Carnevale
J. Neurophysiol. 1999 82(6): p. 3268-3285
If you or your institutuion subscribes to Journal of 
Neurophysiology, you can get their PDF file of the 
article from a link at
  http://jn.physiology.org/cgi/content/abstract/82/6/3268
Otherwise, you may pick up a preprint from
  http://www.neuron.yale.edu/neuron/papers/jnp99/pasnorm.pdf

The two most significant findings reported in this paper
are:
1.  The peak amplitude of individual PSPs as a function
    of synaptic location is best predicted by the spatial
    profile of transfer impedance (Zc), rather than the more
    commonly studied somatopetal voltage transfer ratio
    (Vsoma/Vsynapse).
2.  Active currents are generally NOT necessary to
    overcome location-dependent attenuation of PSP
    amplitudes in real neurons.  Dendritic fields that
    are organized around a central or "primary" dendrite
    were the only exception to this rule.  In other words,
    peak PSP amplitude observed at the spike trigger zone
    is very nearly as large as at the synaptic location,
    and shows little variation with synaptic location in
    cells such as interneurons, granule cells of the
    dentate gyrus, and CA3 pyramidal neurons, even when
    active currents are NOT present.  This also applies
    to synapses onto the basilar branches of CA1 pyramidal
    cells and deep neocortical pyramids.  Since this
    reduction of location-dependent PSP amplitude variance
    does not require active currents, we call this
    phenomenon "passive synaptic normalization."

As noted above, passive synaptic normalization does not
occur in dendritic fields that have terminal branches
organized around a central or "primary" dendrite, e.g.
the apical dendrites of CA1 and deep neocortical
pyramidal cells.  In subsequent work that we presented
at the most recent meeting of the Neuroscience Society
    Carnevale, N.T. and Jaffe, D.B.: Dendritic architecture
    can compensate for synaptic location without active
    currents: the importance of input and transfer impedance
    for synaptic integration. Neuroscience Society Abstracts
    25:1741, 1999.
we found that the lack of passive synaptic normalization in
such dendritic fields is due to the loading effect of terminal
branches, which tend to flatten the spatial profile of input
impedance along the primary dendrite.


--Ted