I recently wrote:
:You should know that many of us consider these
:"spikes" to be essential to understanding the computations performed by
:neurons in places like the hippocampus and cerebral cortex, where much
:work on LTP has been done. In fact, if an LTP-like process is involved
in
:neural plasticity, we may one day find proof for such a process through
:single-unit recording methods in behaving animals that demonstrate
changes
:in neural activity that parallel changes in behavioral performance (i.e.,
:learning).
hdvorak at cns.caltech.edu (Hannah Dvorak) responded:
:And, as you surely know, just this has been done by McNaughton and
:Wilson, who showed increases in correlated firing between place cells
:during maze running and again while the rats were sleeping after
:running the maze, and presumably "replaying" the memory of the maze
:in their sleep. However, to the best of my knowledge it has not been
:shown that such a change in firing patterns is directly due to LTP.
:It's probably a more complicated network property (which, in my
:opinion, does make it more interesting).
The paper by McNaugton and Wilson was a wonderful demonstration of
recent developments in many neuron recording. However, I am not sure
that the McNaugton and Wilson article "proved" anything, other
than showing that neurons are capable of firing in a correlated fashion
during multiple behavioral states.
I must admit that I chose the wrong words in my original statement.
I said that "changes in neural activity that parallel changes in
behavioral performance" might be evidence for learning and this
evidence might be based on an LTP-like process. An e-mail I received
in response to my posting pointed out that I forgot that "correlation
is not causation". I agree! Simple correlations between behavior and
neural activity, or between pairs of neurons, before and after some
behavior is learned is not "proof" of learning. What we really need is
proof that the change in neural activity PREDICTS the animal's state of
learning. (This is the best we can due. Statistically speaking, there
is no "proof" of causation.)
Mark Laubach
Wake Forest University
P.S. My intent was simply to ensure that the original poster of this
series was not confused in thinking that action potentials can not
be recorded extracellularly.
P.P.S. I know that there are no direct data that suggest a role for LTP in
mediating learning. I leave it to those who study LTP to tell us all when
they have finally resolved the issue of LTP and its role in brain
processes.
P.P.P.S. I hope that the result IS due to network properties. This is
what
I study!
