Singing in their sleep: Dreaming birds review their songs

Birds, sleep, and learning

Singing in their sleep: At night, birds may review the songs they learn during the day

December 17, 1998

Zebra finches in a University of Chicago lab are providing strong evidence that sleep plays an important role in learning. Researchers have shown that while young birds sleep at night, they may be reviewing the songs they've learned from their parents during the day. The findings were published in the December 18, 1998 issue of Science.

Normally, the brain is desensitized to outside stimuli during sleep--partially because of changing concentrations of a neurotransmitter called norepinephrine.

But Daniel Margoliash, associate professor of organismal biology and anatomy, neurobiology, and psychology, along with two of his graduate students--Amish Dave and Albert Yu--observed an increase in neurological activity during sleep in a region of the zebra finch brain; this region, the robustus archistratalis (RA), is involved in singing. It is generally believed that the activity of the sleeping brain helps to consolidate what was learned during the day, but how this occurs has never been directly shown.

"One would expect this area to be quiescent during sleep," says Margoliash, because the birds don't sing in their sleep. But even when the birds were asleep, Margoliash and his students recorded strong, erratic RA activity. RA neurons in wakeful birds exhibit fast, regularly oscillating patterns.

Margoliash and his colleagues recorded electrical impulses from single neurons in the RA of anesthetized, asleep, and awake birds as they listened to recordings of their own songs played back on a computer.

Without fail, birds that were asleep or anesthetized exhibited reduced regular oscillations but showed occasional bursts of strong activity in their RA neural impulse patterns. When the birds occasionally woke up during the night, the bursting patterns quickly disappeared and were replaced by the steady oscillating pattern seen during the day.

"This is surprising because the same neurons that show no response during the day have these strong responses to the bird's own song when they are asleep. It's possible that songs learned during the day affect the bursting patterns of the RA at night, serving to solidify the newly learned songs in the bird's mind," says Margoliash.

Margoliash wondered whether the same results could be obtained by giving the birds local injections of norepinephrine, a neurochemical expressed at low concentrations during sleep and at higher concentrations in awake animals. "We wanted to know if the patterns we were getting from sleeping and wakeful birds were strongly tied to the brain's chemical composition as it changes from day to night," says Margoliash.

The team decided to inject norepinephrine into the RA and another region called the HVc. The HVc acts as a gateway for incoming auditory signals; it sends messages to the RA and other centers in the brain. Injections of norepinephrine into the HVc abolished responses in the RA, but injections into the RA had no effect on its response. "This suggests that auditory responses in the RA are definitely dependent on the HVc," says Margoliash.

Margoliash's findings run counter to the motor theory, proposed by A.M. Lieberman in 1967. The motor theory states that incoming auditory signals (speech in humans, or songs in birds), are interpreted when the listener subconsciously mimics the sounds they hear, so that regions associated with speaking would be active during listening.

If this theory were true, then the birds should show activity in the RA, a center associated with singing, as they listen to songs while awake. Instead, this region shows weak activity during the day and vigorous activity at night.

"A speculative and provocative theory about song perception has been replaced by one about song learning and sleep," says Margoliash. "This is hardly the turn of events we expected when we started these studies."

Margoliash wants to compare the RA bursting patterns seen in sleeping birds to patterns generated in regions of the brain when the bird is singing its own song. If the patterns match, it would provide strong evidence that the bird's brain is "replaying" its song subconsciously as it sleeps.