Sleeping to Remember—And Forget


by Jim Schnabel

March 7, 2011

The scientific understanding of memory has been growing by leaps and bounds in the past few years. Researchers have found, for example, that a new memory can be destabilized in a person’s brain by a simple reminder—and that they can take advantage of this destabilization to greatly weaken an unwanted traumatic memory. A new behavioral and brain-imaging study in human subjects suggests that this memory-reactivating procedure can have the opposite effect if it occurs during deep, “slow-wave” sleep.

“We think that the reactivation during slow-wave sleep supports the transfer of the memory representation from the hippocampus to long-term storage in the neocortex, and also strengthens it,” says Jan Born, a sleep and memory researcher at the University of Leubeck in Germany, and co-senior author of the study, which appeared online Jan. 23 in Nature Neuroscience.

Born and his colleagues, led by first author Susanne Diekelmann, gave 24 people—mostly medical students—a standard object-location memory task, in which they were to remember the placement of card pairs depicting animals and everyday objects on a small grid. During the task, which was conducted in late evening, the subjects were exposed to a special odor from a nearby incense candle—a memory cue that could be presented later.

Half the subjects then went to sleep, while the other half stayed awake. When the sleeping subjects reached deep, slow-wave sleep, they were exposed again to the candle odor for 20 minutes, which the researchers said would force the reactivation of their card-location memories. At the same time, the awake subjects also were exposed to the candle odor. Next, the sleeping group was awakened and all subjects were given a new set of card pairs, in which the second card of each pair was placed at a new location. The idea was to “interfere” with the previous memories that had just been reactivated, to determine if they were now more vulnerable to disruption.

Both groups were then tested on their recall of the original set of card locations. As a control, they also were tested at different date, using the same methods but without the memory-reactivating odor.

The results showed a striking difference between the two groups. The people who had had their memories reactivated with the candle odor while they were awake scored much worse than in the control situation, indicating that the reactivation followed by the interference had weakened their original memories. The group that had been asleep during the reactivation, by contrast, scored markedly better than controls. The difference did not appear to be due to the effects of sleep loss, for both groups had similar scores on vigilance tests, and a separate experimental run held earlier in the evening produced the same basic results.

In an accompanying functional MRI study, the researchers examined patterns of brain activity in 47 people during sleep-based memory reactivation and waking memory reactivation. The reactivation in awake subjects triggered extra activity in the right lateral prefrontal cortex—hinting that this brain region might be involved in destabilizing memories after reactivation. By contrast, the reactivation in deep-sleep subjects triggered extra activity in the left-hemisphere hippocampus, a key storage area for short-term memories, as well as in parts of the neocortex associated with long-term-memory—hinting that the new memories were being transferred from the former to the latter, like files from a computer’s RAM to its hard disk.

This idea of a hippocampal to neocortical transfer has been the leading hypothesis in the field,” says Diekelmann. “The thinking has been that the transfer process takes days or weeks or even months, but our evidence suggests that it at least begins very quickly, as soon as slow-wave sleep is reached. Our results also suggest that this transfer occurs only during slow-wave sleep, and not in the waking state.”

A study published in 2009 by researchers in the laboratory of Ken Paller at Northwestern University found that a sleep-delivered auditory cue associated with an object-location memory could strengthen the recall of that memory the following morning, while a wakeful cue had no significant effect. That study, however, did not involve an attempt to weaken memories with an interference learning task. “It’s critical to directly test for the stability of memories to see the effects we found in our study,” Diekelmann says.

Paller says that the German researchers’ findings provide “more evidence for sleep-related improvement in memory storage,” but cautions that “further studies are needed to clarify the extent to which this sort of improvement takes place even when external sensory stimulation does not occur during sleep, and to better understand the neural mechanisms that support this improvement.”

Eternal sunshine?

The researchers suggest that one possible application of such findings could be to overwrite unwanted traumatic memories, using the daytime reactivation and new-memory interference technique, followed by deep-sleep reactivation of the new, therapeutic memories. Another application would be to use the deep-sleep reactivation to enhance memories in students, or in elderly people with declining memory skills.

Born cautions, however, that when it comes to practical applications, “the devil is in the details.” For example, although odors can reactivate memories without pulling a person out of slow-wave sleep, people can quickly become habituated to them so that these stimuli lose their memory-jogging effect. The same is true for auditory reactivation cues, which were used in Paller’s study. Auditory cues also must be masked by "white noise" to reduce the chance that they will wake a person, and it is so far unclear how well either of these techniques would work in an ordinary learning situation. Getting past these hurdles to find something that works for everyday use probably will take years, Born says.

His main interest now is in determining the mechanisms and rules by which the apparent transfer between hippocampal and neocortical memory works. “We think, for example, that in this consolidation process, certain detailed features of the memory are forgotten, and the brain keeps in long term storage just more general and abstract aspects of the representation,” he says. “In evolutionary terms, it wouldn’t be adaptive to store complete episodes with all the details that are not really relevant for future behavior.”