As I sat on my living room floor, a strange woman with bedraggled hair came up to the glass back door. She simply opened it without knocking and started striding in. “Who are you?” I demanded, while the woman… wait on, it was a man. I looked more closely at the face but I couldn’t seem to focus on it. What was going on here? I must be dreaming! “Wake up! Wake up!” I yelled. My eyes flicked open and adjusted to my dark bedroom.
Normally, I wouldn’t blog about dreams. No-one really cares. This dream, though, I was proud of. I had had the fastest nightmare ever. The perceived elapsed time from the start of the dream to my eyes being open was only about three or four seconds. If I could efficiently detect bad dreams, then I could avoid any extended nightmares and spend most of sleep-period enjoying more pleasant dreams.
Last night, I had another dream. I was standing in front of the ice-cream counter, and a man asked me what flavour I would like. I replied “Seafood Extender”. My eyes flicked open wide in horror at the thought.
Two seconds! A new personal best!
Comment by Andrew on December 15, 2005
Dreaming is for space compaction and the building of memory indices, like Google desktop for your head. It also compiles learned scripts into more efficient object code. It is not meant to be pleasant or unpleasant.
Interrupting the compilation process may result in corrupted or incomplete index entries, and intermediate build targets consuming neural space.
Bad dreams are not the side-effect. Remembering them is. Strive not to experience your dreams (wow, that sounds like a corrupted “Team Building” poster – or a koan of some kind).
Comment by Julian on December 15, 2005
So REM sleep is just the beginning where the parser strips out the comments? I am programmed in some form of compiled BASIC?
Did I wake up because of an
assert(Flavour.IsValid(E_ICECREAM, request));
statement.How far should this analogy be taken?
Comment by Alan Green on December 16, 2005
Richard Feynman wrote about his experiences with “lucid dreaming” in Surely You’re Joking, Mr Feynman. Keep going in that direction and you can start redirecting dreams while you are asleep, which saves all that hassle of waking up. It works for me 🙂
Comment by Andrew on December 16, 2005
The compiler generates all kinds of bizarre output, and it’s putting parts of your brain into test mode while it does so. In particular, it turns off all the motor outputs so you don’t do anything unfortunate.
You are interpreting the output as real events, because the system that does this is extremely robust, and assumes that whatever data it gets must mean something. Unless you are unusual, you will remember very little of what you see.
Interrupting the process disrupts it. Alan’s suggestion of redirecting any dreams you experience is a superior option and may well be more fun.
And of course you should bear in mind that I’m making this up as I go. No-one knows what it’s really for.
Comment by Chris on December 16, 2005
“No-one knows what it’s really for”
true… but for some of the latest research, you might be intested in “Nature”‘s free supplement on the science of sleep
http://www.nature.com/nature/supplements/insights/sleep/index.html
Comment by Aristotle Pagaltzis on December 16, 2005
Actually, we do have a basic grasp of what it’s for, and it is indeed to solidify and enhance important memories.
The brain works by triggering patterns associated with other patterns, basically. Sensory inputs light up certain patterns, which trigger associated patterns, which in turn trigger associated patterns; after a short cacophony of echoes, this mutual triggering becomes a feedback loop for the most common patterns and amplifies them until they drown out all others. When the strength of a signal crosses a certain threshold, it becomes an “output.” These patterns are strengthened, so that next time a similar input will cause them to activate sooner.
Dreams are simply fast replays of roughly the last waking period’s inputs, in random order, over and over. The purpose is to strengthen important associations. Dreams are when you actually learn (in a neuorological sense) the day’s lessons.
(Lucid dreaming doesn’t much affect this, btw. Dreams are actually extremely short in real time and last only seconds, even when the dream’s events span years in dream time.)
I remember reading about recent research on neuronal networks (in CompSci). The usual problem with neuronal networks is that you need a “training mode” in which an external “teacher” gives the network inputs and them judges its outputs for correctness; then this judgement is fed back into the system and ripples through it, correcting weights. In normal operation, the weights do not change. But living things don’t work this way; living things somehow judge their own outputs and correct weights constantly. Well, these researchers found that this can be achieved by introducing artificial “dreaming.” Electric sheep indeed…