"Sleep has been shown to be critical for the transfer and consolidation of memories in the cortex. Like memory consolidation, a role for sleep in adaptive forgetting has both historical precedent, as Francis Crick suggested in 1983 that sleep was for “reverse-learning,” and recent empirical support."
Reverse learning is a neurobiological theory of dreams. In 1983, in a paper published in the science journal Nature, Crick and Mitchison's reverse learning model likened the process of dreaming to a computer in that it was "off-line" during dreaming or the REM phase of sleep. During this phase, the brain sifts through information gathered throughout the day and throws out all unwanted material. Ac
"Crick and Mitchison proposed that a reverse learning mechanism in REM sleep removes certain undesirable modes of interaction in neural networks within the cerebral cortex. If their theory is correct then abnormalities of reverse learning might account for some aspects of schizophrenia, mania, and depression."
"In what can be called the eraser theory of REM sleep, Crick and Mitchison have treated its reported absence in the echidna as evidence that it amounts to a mechanism for reverse learning, in which stimulation of the forebrain weakens the synaptic strength of undesirable “parasitic modes” of neuronal activity, thus fine-tuning the brain’s operation ... The echidna, it is said, gets by without REM sleep because its surprisingly large neocortex makes reverse learning unnecessary. If true, an inverse relationship between size of neocortex and REM sleep quotas is to be expected in other species, but supportive data are lacking."
"The idea that sleep might be involved in the erasure or filtering of information has been put forward by several authors ... In particular in 1983, Crick and Mitchison ... proposed, based on a neurocomputational model of associative learning, that dreaming during REM sleep helps to forget “parasitic modes” of activity, thus ensuring an efficient mode of operation of the brain during waking. ... As a solution to this problem, the authors proposed a “reverse learning” mechanism during REM sleep-dreaming that dampens synaptic weights to reduce the probability of these parasitic activity modes and thereby also enhances the efficacy and storage capacity of the network. ... In simulation studies, repeated unlearning procedures indeed improved the learning capability of the network and retrieval of recently learned patterns, but concurrently weakened more remote memories ..."