This posting discusses the paper "The neural servo model"
> While your theory is interesting, your "observations" on sleep are
> somewhat incorrect. People don't "go back to their early childhood"
> during deep sleep.
I agree that "observations" is a strange word to use here in the first place.
I remember reading in KIJK (a dutch popular science mag) that they do,
but I could be wrong of course. I 'll check this weekend what the article
said. (The issues under future work have not been worked out yet ;-)
> Some other sleep-related comments: True, the cortical oscillations
> are slow and of high amplitude. But sleep depth doesn't "shut down" the
> less "primitive" parts of our brain: Those oscillations are from
> cortical cells firing in unison, most likely due to thalamic neurons
> controlling the pattern of cortical firing. [...]
I believe that these neurons that fire in unison (the diffuse systems) are in fact
providing some sort of neuromodulation for large areas of the cortex. If (groups
of) neurons that have been probing for a particular type of association for
a longer time also develop a greater strength (Hebbian learning), then this
neuromodulation could have less effect on them. This is what I mean with
effectively downsizing the network to a more primitive parts of the brain
(not the *evolutionary* primitive parts like the limbic system).
The question Why would we want this? is answered by the paper.
Do you think such a mechanism could be plausible?
> [...] We don't wake up in the mornings because of
> increased sensory input (the "supply of reality") - just ask anybody
> that works a third shift or parties all night.
Actually, I wrote:
"In the morning when there is a sufficient supply of reality pouring into our senses,
the brain can finally launch a successful attack, creating a phantasy world that keeps
matching reality until we are fully 'awake': dreaming correctly or correcting reality."
So, I would suggest that our brain creates a phantasy that becomes more detailed
as more recent connections kick in. On the other hand, if we have been awake too
long, then our mental world could lose touch with reality (for reasons mentioned
in the paper) resulting in a need for relaxation or sleep to reduce this mismatch.
So it is not the supply of reality that matters, but the ability to generate a matching
phantasy (effectively cancelling out the input to a certain degree).
> The idea that the brain corrects "reality" is interesting, and one I've heard of before,
> but I still have the same question - if the brain is involved in fixing
> reality, why can someone be killed without getting the sensory input
> that says he is going to die? [...]
I think that the paper is quite clear here: I mean correcting by means of *physical*
actions, not some sort of solipsism or how it called. It is these actions that make
the network so useful for the survival of an organism, without these actions the
discussed system would not stand a chance to have been selected by evolution
in reality. I am not sure it *is* a biological reality, but it is a computational reality.
> Lack of sleep does not have the same effect as panic. You can get
> hallucinations (not generally a symptom of panic but sometimes a cause)
> with long sleep deprivation. [...]
The effect I mention in the paper is: "We are unable to refresh our phantasy world,
which requires a very long sequence of operations to build (slow brainwaves)."
These slow brainwaves imply lower arousal, so both situations are about a
system that is more aroused than would be useful:
- the sleepy person should be asleep, but is awake
- the panicing person should be alert, but is overaroused
Both (according to my model) will be unable to adapt to changes in their
environment. The sleepy person's mental world drifts away, the panicing
person's world seems to have turned against him: many skills like
operating a telephone turn out to be impossible (cf. the paper).
Have a nice day,
Mervyn van Kuyen