Interesting ideas, but I wonder if you see where they take you.
Pleasure and pain are basically reinforcement signals: an action
followed by pleasure tends to be repeated in a similar situation; an
action followed by pains tends not to be repeated. But what are the
actions that a neuron can take? To fire or not to fire, those are the
only things it can do. So here is the logic:
1. You have defined "neural pleasure" as firing, and "neural pain" as
ceasing to fire.
2. Therefore if a neuron takes an action (fires), and this is followed
by continued firing (pleasure), then in a similar situation the
neuron will again tend to fire (repeat the action).
3. Conversely, if a neuron takes an action (fires), and this is
followed by the neuron ceasing to fire (pain), then in a similar
situation the neuron will tend not to fire as much (i.e., not to
repeat the action).
If this is a fair summary of what you're proposing, then I have a
couple of comments:
1. The mechanism outlined above would be called "Hebbian learning with
a delay" by a neural network theorist. It is a useful mechanism,
but not by any means a universal learning rule. That is, this
mechanism by itself is not enough to give a properly functioning
brain.
2. The concept of single-neuron-pleasure may or may not make sense,
but if it does, then firing per se is unlikely to be the pleasure
signal, because if it were, then it seems that neurons would act so
as to maximize their overall firing rates, and they don't. If
anything, they act to keep their long-term average firing rates
within a certain range. It might then be reasonable to say that
departures either above or below the set-point act as single-neuron
pain signals. (I should add, though, that the cellular basis of
whole-organism pleasure and pain are likely to be quite different
in nature.)
-- Bill
