FROM SYNAPSE TO SOUL
*** Reinforcing input neutralization ***
For two years now, I have been exploring (using computer simulations)
reinforcement of all neural growth that results in greater
'neutralization' of a neural network's sensory input. This reinforcement
causes a network to collect and maintain feedback connections that deliver
some sort of inversed imagery to the sensory input channels. A lack of
neutralization results in signals representing prediction error or
'novelty'. Many neural pathways in the brain are known to convey such
novelty signals, therefore neutralization must be a biological reality.
*** Fertilizing the evolution of recurrent networks ***
Each maintained connection transforms a recurrent pattern that has been
formed and transformed by the network already in place. Therefore
sequentially collected neural wiring cannot be turned on as a single unit:
recent connections would inmediately start to transform the incoming
patterns in an, at that stage, undesirable manner. To reboot the system
if it accidentally becomes chaotic its development has to be 'replayed' in
a compressed form. I would suggest that our diffuse arousal networks (that
are woven through our cortex) are responsible for a continuous replay by
means of large scale neuromodulation (to which recent connections could be
more sensitive). Since these networks are associated with different EEG
waves, I would suggest that these waves determine the speed and a depth of
developmental replay in order to match the demands of the environment
(high speed vs. high stability).
*** The fruits of input neutralization ***
o Absence of novelty implies correct neutralization. Neutralization
requires accurate anticipation and timing, not ignorance. These skills are
obviously essential for the survival of any advanced organism.
o Neutralization causes a reduction of the input level. Therefore,
continuous neutralization forces the brain to develop high gain and
internal recurrency (it has to produce more and more inversed imagery from
diminishing input patterns). These are typical human, cortical properties
that neurophysiologists associate with epileptic seizures. I believe that
the last paragraph explains why these seizures are actually more rare than
neurophysiology can explain.
o Finally, I would like to stress that behavior can be interpreted as an
effort to increase the performance of a neutralization seeking network.
Through behavior a network can choose or create an environment that it can
anticipate to a greater extent. Obviously this environment shall contain
many aspects of its early environment (like, hopefully, a physically and
socially healthy state of being). Another property of neutralizing
networks that provides them a large evolutionary advantage!
The picture that arises from these properties is that of a whirlwind of
activity that through its anticipating and controlling efforts remains
relatively input independent and quite conservative in its behavior.
Properties of a soul?
Comments? Mail mervyn at xs4all.nl or reply to the newsgroup!
With Kind Regards,
Mervyn van Kuyen