On 7 Jan 2004 07:29:00 -0800, guyscarsbrook at hotmail.com (Guy Lux)
wrote:
>Reading that the brain makes 100 trillion calculations per second
>(relaxed state or at peak?) and that a new Linux supercomputer will be
>able to make 11.1 trillion calcs a second, could anyone hasard a guess
>as to what percentage of the brain's 100 trillion calculations are
>used for pure conciousness as opposed to movement, sensitivity and
>some of the more basic functions like breathing and heart control?
>>If 11.1% or less is used for consciousness, why could this new
>computer not become consious?
>>And would a healthy, developed, conscious part" of the brain be
>considered a conscious entity, if it had never had any sensory input.
>>What sort of input would a brain/supercomputer need in order to become
>conscious?
>
There are several problems dealing with the questions you ask.
First, what is a "calculation" in a computer? Do you know so
precisely that you can say for cetaintly to three significant figures
exactly how many a computer can do? Is it a floating point
calculation? Is it a machine instruction?
Second, what is a "calculation" in a brain? Whatever the answer is
for a computer, there is no analogous concept for a brain. It does
not make any sense to say that a single neuron is making x number of
calculations in one second. Is it related to action potentials?
Probably most neural integration is completely graded in microcircuits
and has nothing to do with action potentials. These are analog
running in continuous time. The concept really doesn't make sense.
Yes, people do throw out numbers like the 100 trillion figure, but
that doesn't mean that the number has any significance. Are
"computations" part of what microcircuits or even individual synapses
do? or do major dendritic branches do it? or whole neurons? or neural
complexes? How could you even begin to count?
Third, what is "consciousness"? We really don't have a clue so it is
impossible to describe "what percentage" of the brain is needed to
make it happen. Is it raw computing power? As far as we know, there
is no molecular or cellular machinery working in human brains (that
supposedl have consciousness) that does not also work in insect or
slug or whatever other type of brain you can think of (that supposedl
don not have consciousness). As a result we conclude that, since the
main difference is the size of the brain, then size must be important.
Still, there are animals with larger brains than our own. Does that
automatically make them conscious? Almost no one believes that.
Furthermore, whatever consciousness is, it is not simply raw machine
computational power, but must be something about how that power is
used. For example your supercomputer, no matter how powerful, is
totally incapable of solving a weather simulation or nuclear reactions
or generating a fully animated video scene or anything else unless it
is programmed to do so. Further, none of these situations requires a
supercomputer in a theoretical sense. With a small, slow computer, it
just takes longer to do the job. Is consciousness a problem that
requires, for some fundamental reason, massive computational power or
is it simply something that runs too slowly to be useful on a small
system? We have no clue.
Most concepts of consciousness -- there is a very large literature in
the realms of neurology, psychology, cognitive science, and philosophy
that you should consult -- indicate that it is necessarily related to
integrating a sense of "self" with the experience of behavior in the
world. That is, you do need to have available sensory information
about what is happening in the world and also how your own actions
modify what happens in the world. So without any sensory input, there
really could't be consciousness in the form we usually consider.
Exactly what it would take to produce machine consciousness, or even
if that is at all possible, is far from being answerable at present.