PAGE 4 OF 5
This work supports the developing view that memory formation depends on cross talk between hippocampus and cortex, where memories are ultimately stored. The hypothesis is that daytime memories held temporarily in the hippocampus are rebroadcast during sleep from the hippocampus to the cortex, which then replays the signals itself, strengthening the synapses to imprint the new memories. "The hippocampus can learn very, very rapidly, but it doesn't have a very high capacity," says Sejnowski. "The big knowledge storage is in the cortex."
Knowledge in the cortex changes more slowly, though. It has to rehearse many times to learn a new memory. "Through the replaying of the hippocampus, the cortex will gradually figure out which of the new items coming in are the most important ones and how to fit them into existing knowledge," says Sejnowski. "That's the story that's emerging from all the electrical recordings."
Still, many issues about sleep and memory remain to be resolved, such as the relative importance of slow-wave sleep and intermediate sleep. And questions linger about which types of memory are actually stored during sleep. Evidence is strong that memory for procedures and skills is improved by sleep, Sejnowski says. But whether factual (declarative) memories are consolidated by sleep as well has been more controversial.
"Memory is not a monolithic single thing," says Sejnowski. "One of the major discoveries we've made is that there are dozens of different memory systems...that use different mechanisms and different parts of our brain."
Other researchers point out that much of this story, while plausible, awaits rigorous confirmation. For one thing, whether the information in the sleeping brain really flows from hippocampus to cortex needs to be established more definitively, says Giulio Tononi, winner of an NIH Director's Pioneer Award in 2005 for his studies of sleep and consciousness.
"I don't suspect that the hippocampus cannot talk back to the cortex, it just needs to be proven better," says Tononi, a neuroscientist at the University of Wisconsin-Madison who is studying other synapse mechanisms during sleep. He suggests that more work is needed to show whether the replay activity is faithful enough to the original input for strengthening long-term memories effectively.
Tononi also notes that some studies find similar replay signaling during quiet wakefulness, suggesting that sleep itself may not really be needed for memory storage. If that is so, the mystery of why evolution has made sleep so ubiquitous resurfaces.
There is no disagreement, though, about sleep's importance and the need to understand it better. This is a message that today's society seems to be neglecting, according to Magee, and we are paying a price in the reduced productivity and degraded learning capacity that come from sleep deprivation.
Sleep is not simply a time for storing memories, he points out. It's also a time for tune-up and repair to keep the brain in peak operating order. "There is a housekeeping function going on," he says. If you don't get enough sleep tonight, it's not just today's memories that will be less likely to be permanently stored. Your brain won't be working so well tomorrow, either.