"Richard Vickery" <Richard.Vickery at unsw.edu.au>
> The EEG is an averaged response from many millions of neurons. What is
lost
> in the averaging is all the detail like what the person is thinking, the
EEG
> just tells you WHETHER they are thinking (their brain state: asleep,
relaxed
> etc)
>> In principle you really would need to know what each neuron is doing -
fMRI
> has time and space resolution problems, EEG has space resolution problems.
It
> is not inconceivable that new technology could allow us to simultaneously
> monitor the activity of billions of neurons, but to interpret the role of
each
> of the neurons would probably take more than thelifetime of each
individual
> that you tried to analyse.
I think someone experienced with fMRI can tell, from looking at brain
activation
patterns, what state of mind the subject is in (roughly). The same with
EEG, it
depends on how many samples you want to provide with it.
Brain development is dependent upon a person's life history and each ones
brain
is different. The number of samples required to train an Artificial Neural
Net must, therefore, be comparable to a whole life's experience. So you are
right.
On the other hand, the "mechanisms" that enable us abstract thinking came
from
evolution, which is slow. which means that they could be simple and few in
number. Like the immune system that can generate huge combinations of
antibodies,
the brain's mechanisms could be similar.
So it might be possible to identify a *particular* brain state with the
exact
group of neurons (or glia for that matter) associated with it. Moreover,
once that
mechanism is fully understood then it might be possible to transfer the
information
from one substrate (cells) to another (silicon?).
To answer the original poster: before such mechanisms were elucidated, it
would be
anyones guess how long it would take. I agree that it is at present,
impossible.