It sounds like he was onto something with the waking mind being being prone to finding local maxima, but the background/sleeping mind being good for branching out and finding other solutions.
edit: Actually it looks like he cited scientific american talking about the sleeping brain finding relations between problems, I only remembered the bits about strengthening memory.
thanks to hvass (the link is to a paywalled site), the basic idea is:
- some tasks require "lateral thinking". the example given is: what joins the words "lick", "sprinkle", "mine"? maybe lateral thinking is the wrong expression, but the idea is that the solution requires making connections from unrelated "semantic areas"
- people guess that the brain has a network-like approach where "activation" starts at nodes associated with the different concepts ("sprinkle", etc) and slowly spreads outwards, until, presumably, all three growing (glowing? ;o) areas meet at "salt".
- sleep helps for these tasks, but not for easier ones.
It sounds like he was onto something with the waking mind being being prone to finding local maxima, but the background/sleeping mind being good for branching out and finding other solutions.
edit: Actually it looks like he cited scientific american talking about the sleeping brain finding relations between problems, I only remembered the bits about strengthening memory.