Re: Sleep

Anders Sandberg (
01 Nov 1999 11:41:01 +0100

Jeff Davis <> writes:

> "Robert J. Bradbury" <>,
> on Fri, 29 Oct 1999 09:37:14 -0700 (PDT), wrote:
> >I concur with [Anders'] comments on the primary purpose of sleep
> >being to "integrate" critical experiences into your knowledge
> >database. There may also be a biological purpose for
> >maintenance, repair & recycling to be done when resources
> >(esp. energy) are less needed for "thinking" (though I can't
> >point to much hard biological evidence for this).
> It seems reasonable, though I too, have no evidence, that thinking employs
> the current structural "state" of the brain, and that a backlog of new
> experiences of the latest experiential period would require a certain
> amount of cellular energy to add them--overlay them--onto the
> continuously-revised biostructural substrate. Energy resources might be
> the limiting factor, or it could just be that thinking and revising simply
> can't go on at the same time.

Actually, I don't think so. First, the brain uses a lot of the body's energy all the time, and while memory encoding seems to increase the metabolism (support: PET imaging of people in memory tasks and that higher levels of oxygen, glucose or ALC increases memory encoding success) it doesn't seem *that* expensive (maybe I'm wrong, maybe we sleep because otherwise memory consolidation all at once would cook our brains :-). Second, we seem to do some memory encoding even when we are awake (I can remember the bus trip to my office this morning, and I haven't slept in between).

> So thinking has to be turned off. The
> brain's version of "Please excuse the delay while we update our records."
> This might tend to explain the strikingly different nature of those other
> metal activities that occur while one is asleep. Perhaps they are confined
> to areas of the brain not "out of service for the purpose of revision" , or
> perhaps the nature of these mental activities lack some essential feature
> of awake thinking, a feature that conflicts with "revising"?

Mayhaps. It is rather hard to run a neural network while it is also doing unsupervised learning, it can easily get caught in an attractor and learn it to the exclusion of everything. So I think it makes sense to turn off a lot of functions (especially motor activity) during sleep.

Anders Sandberg                                      Towards Ascension!                  
GCS/M/S/O d++ -p+ c++++ !l u+ e++ m++ s+/+ n--- h+/* f+ g+ w++ t+ r+ !y