I spent the last week at WCNN '95. Dan Alkon, Jim Olds, and Tom Vogl (also
working with David Lester) had some very interesting things to discuss
about the work underway at NIH on learning. Alkon indicated that he has
evidence that ltp is not involved in short or long term retention. The
process instead seems to involve membrane biochemistry that goes somewhat
as follows:
1. Glutamate activates a receptor, eventually leading to activation of
PKC, which binds to the dendritic spine membrane and shuts down Ca++
activated P113 K+ channels. Eventually (20-30 minutes) it unbinds and the
channels reopen. Meanwhile the membrane is a bit depolarized and increased
in sensitivity. During this period, if a GABA receptor is activated, it
produces arachnoic acid, which can bind to the PKC, stabilizing it. The
PKC also phosphorylates a small G-protein which interacts with the nucleus
to cause production of mRNA needed to maintain the proteins in the
activated spine.
2. Jim Olds speculated that the mRNA is identified and used by the recently
discovered polyribosomes associated with the base of dendritic spines.
These polyribosomes actually produce the proteins (locally) needed to
maintain the activated spine.
3. As a result of this, learning takes place locally in the dendritic
tree (creating Pribram's 'synapto-dendritic field', although Vogl calls
the s-d field a 'spherical chicken'), not at the cell itself.
It sure looks like rapid convergence to a common model is underway in this
area, particularly as the neural modeling community is starting to
discover the interesting dynamics that can occur in the dendrites.
--
Harry Erwin
Internet: herwin at gmu.edu
Home Page: http://osf1.gmu.edu/~herwin (try a couple of times)
PhD student in comp neurosci: "Glitches happen" & "Meaning is emotional"
