d.kelly at NO.ucl.ac.uk.SPAM
Sat Nov 2 11:15:44 EST 2002
The classical model of neurons says that the flow of information is strictly
one-way: a neuron receives inputs at the dendrites and sends its output via
its axon. However, it has been know for many years that this model is far
too over simplistic.
Once important finding is that dendrites have active electrical properties -
they are not simply passive conductors. Dendrites can support action
potentials. These APs can go either towards the cell body or away from the
cell body. Action potentials travelling away from the cell body are known
as 'back propagating' action potentials because they're going 'backwards'
according to the classical view of neurons.
Theories for the function of back-prop are numerous and expect PubMed
(http://www.ncbi.nlm.nih.gov/entrez/query.fcgi) and www.Google.com will give
you many good articles. I'm no expert on the subject but I believe that one
of the most prevalent theories is that back-prop is involved in Hebbian
synaptic plasticity. If the activity of a synapse strongly corresponds to
the firing of a neuron then that synapse will be strengthened. How does the
synapse know when its parent cell has fired? Back-prop.
I hope this is of some help,
"John H." <johnh at faraway.xxx> wrote in message
news:y5Pw9.23132$Sr6.681901 at ozemail.com.au...
> Can anyone here tell me what backward propagation is and its supposed
> functions. That or direct to a source where I can read something about it?
> John H.
More information about the Neur-sci