New research conducted at Northwestern University showed that sounds can penetrate deep sleep and be used to guide rehearsal of specific information, pushing people’s consolidation of memories in one direction over another.

Twenty-five sounds (such as a teakettle whistle and a cat’s meow) were presented to study participants during a nap as part of the study. The participants were unaware of the sounds as they slept, yet, upon awakening, memory tests showed that spatial memories had changed.

The researchers found that the participants more accurately dragged an object to the correct location on a computer screen for the 25 images whose corresponding sounds were presented during sleep (such as a muffled explosion for a photo of dynamite) than for another 25 matched objects.

"The research strongly suggests that we don’t shut down our minds during deep sleep," said lead author John Rudoy, a neuroscience PhD student, in an announcement about the study. "Rather this is an important time for consolidating memories."

"While asleep, people might process anything that happened during the day—what they ate for breakfast, television shows they watched, anything," said senior author Ken Paller, professor of psychology in the Weinberg College of Arts and Sciences at Northwestern. "But we decided which memories our volunteers would activate, guiding them to rehearse some of the locations they had learned an hour earlier."

The new study focused on memory processing during deep sleep, rather than REM sleep, as much research on the topic does.

"We are beginning to see that deep sleep actually is a key time for memory processing," Paller said.

To conduct the study, the 12 study participants were taught to associate each of 50 images with a random location on a computer screen prior to taking a nap. Each object was paired with a corresponding sound delivered over a speaker. Locations were learned by repeating trials until study participants became proficient at placing all the objects in their assigned places.

Approximately 45 minutes after learning, each participant reclined in a quiet, darkened room. Electrodes attached to their scalp measured their brain activity, indicating when they were asleep. Sleep sounds were presented without waking up anyone. When asked later, none of the participants thought sounds had been played during the naps. Yet, memory testing showed that placements of the objects were more accurate for those cued by their associated sounds during sleep than for those not cued.