Asking people with amnesia to play Tetris has revealed clues to why dreams are so illogical - and perhaps to why we dream at all.
Most sleep scientists think that in our first dreams of the night we access "declarative" memories of recent events - those we can consciously recall. People with amnesia cannot form this kind of memory because of damage to the hippocampus, a part of the brain.
But Robert Stickgold of Harvard Medical School found that people with amnesia who had no recollection of having played Tetris still "saw" the computer game's falling blocks in their first hour of sleep.
He thinks this shows that dreams during so-called Stage One sleep rely on abstract, subconscious memories, which amnesiacs are still able to access. And he thinks this explains why dreams often seem disordered and bizarre.
"When I saw the results I almost fell of my chair," Stickgold says. "We thought that if there's one part of sleep that depends on declarative memories, which amnesiacs lack, it's sleep onset."
Dream to learn
Declarative memories "are ones you can declare you know: 'I had eggs for breakfast' or 'My brother's name is Ed'," Stickgold says. "Implicit memories are ones that are in your brain but you can't access consciously."
Stickgold thinks the access to implicit memories that dreams give may be essential for learning. "One of the most difficult problems the brain faces is how to put together information from different sources, to see how things fit together.
"By blocking declarative memories and forcing the system to work with these weak associations, the brain is coerced into looking for unexpected, novel and potentially highly creative and useful connections that otherwise we would not notice."
"The study has shown that the mental activity at sleep onset is clearly related to the learning situation," agrees Carlyle Smith, a sleep scientist at Trent University in Peterborough, Ontario.
Stickgold thinks declarative memories are even more likely to be blocked during REM dream sleep. "We're going to look for Tetris images in REM sleep next," he says.
Stickgold asked three groups of people to play Tetris for a total of seven hours over a three-day period. People with amnesia, expert players and novices played the game every morning and evening. Stickgold monitored their dreams as they were drifting into sleep on the first and second evenings.
He found that 17 of his 27 players reported dreaming at least once in the hour after they fell asleep, and all 17 reported seeing falling Tetris pieces. But overall, there were more dream reports on the second night.
This time lag is important, Stickgold says. "It's as if the brain needs more time or more play before it decides 'OK, this is something that I really need to deal with at sleep onset'."
Furthermore, he says, Tetris novices who saw falling blocks had not performed as well in their initial Tetris session compared to the novices who did dream about the blocks. "It's as if the more work you have to do, the more likely you are to get the imagery," he says.
"This is particularly interesting," says Richard Haier of the University of California in Irvine. "It suggests that people who learn the best may use the early dreaming the least for learning."
Source: Science (vol 290, p350)
1010 GMT, 13 October 2000
Maintained by Francis F. Steen, Communication Studies, University of California Los Angeles