Back to the original question; although cortical plasticity has been
observed to be far greater than was originally assumed (see Frank's
post below), there is even in this work considerable evidence of
hardwiring. For example, in Merzenich's work,
when a finger is cut off, the sensory area
of the brain that used to represent that finger comes to represent
neighboring fingers. Similar work has been found in visual cortex.
On the other hand, the areas are never found to represent completely
different things. That is, the finger area doesn't come to respond to
stimulation of, say, the nose, nor does it ever come to represent
a different modality.
So back to the original question, does the behavioral function of a
neuron change as the result of plasticity, the answer is, probably
no... that is probably the case for neurons that are motor or
sensory in nature. Probably not much has changed about these neurons
except their weighting of synaptic inputs that already existed
at the time of the trauma...
There is a real problem with our terminology. Words like plasticity, functional reorganization, map changes, representational modifications etc. are all used interchangeably. When a set of cortical neurons begins to respond to inputs from new receptive fields, these terms have been used to describe the changes. However, it is clear from denervation (amputation) studies that neurons may acquire new receptive fields -- hence, plasticity, map changes, etc. -- but still signal (represent) the original skin sites; stimulation of the new fields can result in sensation projected to the original, now-denervated skin fields (phantom). So plasticity, reorganization, etc. has occurred in the sense of neurons responding to new inputs, but when those neurons are activated they still signal (mean) the original skin location -- the function of the neurons has not changed. GSD