I've recently became aware of the idea of "reverse learning" that might happen during REM sleep - the brain's attempt to eliminate pathological attractors that might appear in neural networks. The idea was originally expressed by Crick and Mitchison in their 1983 paper on "The function of dream sleep".
I'm interested if there has any recent work that has supported that idea. I'm aware of this wikipedia article on reverse learning. More specifically, I'm interested in how dream recall (writing down or recalling dreams upon awakening) affects the process of "reverse learning"?