In 1964 it seemed as if the avian vocal organ, the syrinx, was a good place to start understanding the relation between brain and song. I started with chaffinches. I found that an alarmed chaffinch, which had had the nerves to the syrinx cut, breathed with difficulty. This problem could be avoided if one cut only the right or left nerve to the syrinx. Song was distorted, but breathing was not. I denervated the left half of the syrinx in many chaffinches and made careful notes of how their song was distorted. Many months later, I operated on one chaffinch's right side. The song of this bird did not change!
After confirming this result I realized I had stumbled across one of those little gems we all dream of finding. Even though the chaffinch syrinx consists of two anatomically similar halves—each with its own air supply, muscular control and innervation— chaffinches preferred to use their left syringeal half to sing. The right half in most cases remained silent, but on occasion both sides chimed in at once, each with a different note. If the left half was denervated before the onset of song learning, then the right half and its innervation developed normal song—much as a person who has lost the right hand in infancy might develop normal skills with the left hand.
As a student at Berkeley I had been taught about handedness and hemispheric dominance for speech, language and manual skills. It was thought that this was a unique feature of the human brain. But here I had a bird that was "left-handed" for song control. The fact that this left-handedness was very consistent, that it affected, as in humans, a learned vocal skill, convinced me that I had now an animal model in which to study handedness—and perhaps hemispheric dominance—for a learned behavior.
Severe quarantine regulations imposed on all avian imports in 1970 made me switch my attention to canaries. Fortunately, left syringeal dominance for song was even more marked in canaries than in chaffinches, and canaries proved to .be superb subjects in which to study the brain pathways used in singing. Canaries had, it turned out, an anatomically well-defined part of forebrain devoted to song. Lesion of this higher vocal center (HVc) in the left hemisphere eliminated virtually all of the song an adult male canary had learned. Lesion of HVc on the right side had a much less marked effect on song. Thus, canaries have not only syringeal dominance for song control, but also left hemispheric dominance—the pathway from forebrain to syrinx is uncrossed. The parallel with the human situation was uncanny.
Other discoveries followed. For example, HVc was several times larger in male canaries than in female canaries, which could be related to the fact that males learn complex songs, whereas female canaries sing much less. This provided the first vertebrate example of a gross brain sexual dimorphism.
Interestingly, adult female canaries treated with testosterone sang in a male-like manner and the size of their HVc doubled. Particularly astonishing was the observation that the HVc of adult males was large in the spring, when testosterone levels were high and the birds were in stable full song, and small in the fall, when testosterone levels were very low and song had become variable and like that of juvenile birds.
Canaries modify their song every year, and most of these changes first occur in late summer and early fall, when HVc becomes smaller. Apparently a yearly recurrence of songlearning is related to an anatomical and physiological rejuvenation of the underlying brain circuitry. In addition, we noticed that male canaries with large song repertoires had large HVcs, whereas other male canaries with small HVcs had small song repertoires.
More surprises were in store. It turns out that when a songbird hears a sound, cells in HVc become excited. Dan Margoliash, working at CalTech, discovered that they get particularly excited if the song played back is the bird's own song. Then Heather Williams at Rockefeller University showed that sound input does not stop at HVc, but cascades down the motor pathway for song production, reaching the muscles of the syrinx. Under those conditions, however, sound is not produced because the activation of the syringeal muscles is not accompanied by an expiratory pulse. Imagine the paradox of a bird perceiving a sound and, with a delay of some 60 milliseconds, converting that sound into a silent syringeal gesture! What for?
There may be a human precedent for this. Alvin Liberman, of the Haskins Laboratory, argues that a conversion of sound into gesture is necessary to achieve a phonetic decoding of speech signals. According to his hypothesis, we know what phonemes we hear because we know what gestures would be required to reproduce them. Birds may provide an animal model to study how sound perception relates to sound production.
The most exciting discovery was yet to come. I had wondered whether a part of the brain such as HVc, that underwent such drastic size changes in adulthood, might not be doing this by adding and discarding neurons. First, Steve Goldman and I showed that neurogenesis continues in adult avian brain, new neurons being added to those already in HVc. Then John Paton and I showed that the new neurons became connected to existing circuits. Later came evidence of matching cell death. HVc provides a lovely example of a brain phenomenon until now thought to occur only in olfactory mucosa: neuronal replacement in adulthood. The periodic replacement of olfactory neurons has been attributed to exposure to environmental wear, so another explanation must be found for neuronal replacement in HVc.
Birds are flying machines and weight is at a premium. Information is stored in brain circuits and such circuits occupy space. Since there is a relation between amount of brain space (HVc) devoted to a learned behavior and how much is learned, one way of economizing on brain weight would be to build circuits with replaceable neurons. If brain parts such as HVc thought to be involved in the perception and production of song are to continue to learn yet remain modest in size, then perhaps the cells that hold memories—and the memories they hold—must be periodically replaced. If true, this is a new principle of brain function.
Twenty years after first tinkering with the songbird syrinx, the story has become rich and perplexing. Is there anything here that could help the welfare of human brains? The answer, of course, is that we do not yet know, but given another 20 years of study the chances are that, as in an election year, there might be something for everyone!