Function of Dreaming: memory reprocessing  My own theory

Copyright George Christos 2003

return to my home page


As almost all mammals have evolved with rapid-eye-movement (REM) sleep, the phase of sleep associated with dreaming,  it clearly serves an important biological function.  During REM sleep the brain is (slightly) more active than when you are awake.  More blood flows into the brain and there is an increased amount of asynchronous activity, compared to other sleep stages.  Since the brain is practically disconnected from the body during REM sleep, that function is likely to involve the internal processing of memory.  
In 1983 Crick and Mitchison proposed that REM sleep involves a reverse-learning or unlearning mechanism that helps the brain to eliminate the spurious memories (or parasitic modes in their terminology) that the brain generates as a consequence of the distributed overlapping storage of memory in the brain.   I have turned their theory on its head.  See below.  Unlearning also helps to subdue obsession, or the dominance of the system by a limited number of strong memories.  I have previously suggested that REM sleep may also help to eliminate (or forget) other unwanted (older and irrelevant) memories. This would offer mammals that dream an evolutionary advantage in a new and changing environment. 
I have performed numerous simulations to test the CM hypotheses, but  recently while writing my book Memory and Dreams I have discovered that instead of reducing the number of spurious memories reverse-learning in dream sleep actually increases their number.  The problem was that earlier simulations were carried out on unrealistic neurobiological models and with memories stored with equal intensity.  Earlier models also had much more symmetry than in a real biological system.  A consequence of this is that there were too many spurious memories to contend with.  In a more realistic neurobiological system the number of spurious memories is limited and it is in this regime that I am suggesting that they are important for creativity and new learning.  I have proposed my own theory of why we dream.  We dream to generate more spurious memories, or that is to roughen up our 'memoryscape'.  This prepares us for a new day of learning.  Without spurious memories the brain would not be able to learn something new.  An interesting consequence of this theory about learning is that whatever we learn the next day is already in our heads.  All we are doing is strengthening those shallow memory states.  In a sense you cannot learn something new if you do not know something about that already.  This is a profound but logically sound proposition.  For a deeper discussion of my dream sleep theory and for a comprehensive review of other ideas about the function of dream sleep I must refer you once again to my book "Memory and Dreams".  
"It seems to me that we agree that creativity is partly due to becoming aware of concepts that are related, but not obviously so. This might be helped by spurious attractors. Mitchison and I suggested that REM sleep diminished these, so we thought that the "unlearning" we had postulated would tend to make a person less creative and more matter-of-fact."
"You and I both think that spurious memories could help "creativity," but we suggested REM sleep decreases them, while you suggest it could increase them."
Francis Crick, Nobel Laureate, co-discoverer of the structure of DNA and the basis of biological life.

The brain is refreshed for new learning, adaptability, and creativity after a good session of dreaming . photo taken 12 Nov. 2002 by Jonathan Christos, Esq.   



radio 3AKA Melbourne, 2 November 2003, live

Sunday Times, Perth, Australia, 2 November 2003

Nonetonoon, Radio New Zealand, 19 November 2003

Herald Sun, 5 March 2004