A major tenet of the neurosciences has been that all neurons (nerve cells) in the brains of vertebrate animals are formed early in development. An adult vertebrate, it was believed, must make do with a fixed number of neurons: those lost through disease or injury are not replaced, and adult learning takes place not through generation of new cells but through modification of connections among existing ones. However, new evidence for neurogenesis (the birth of new neurons) has come from the study of canary song. Young canaries and other songbirds learn to sing much as humans learn to speak, by imitating models provided by their elders. Several weeks after birth, a young bird produces its first rudimentary attempts at singing; over the next few months the song becomes more structured and stable, reaching a fully developed state by the time the bird approaches its first breeding season. But this repertoire of song is not permanently learned. After each breeding season, during late summer and fall, the bird loses mastery of its developed “vocabulary,” and its song becomes as unstable as that of a juvenile bird. During the following winter and spring, however, the canary acquires new songs, and by the next breeding season it has developed an entirely new repertoire. Recent neurological research into this learning and relearning process has shown that the two most important regions of the canary's brain related to the learning of songs actually vary in size at different times of the year. In the spring, when the bird's song is highly developed and uniform, the regions are roughly twice as large as they are in the fall. Further experiments tracing individual nerve cells within these regions have shown that the number of neurons drops by about 38 percent after the breeding season, but by the following breeding season, new ones have been generated to replace them. A possible explanation for this continual replacement of nerve cells may have to do with the canary's relatively long life span and the requirements of flight. Its brain would have to be substantially larger and heavier than might be feasible for flying if it had to carry all the brain cells needed to process and retain all the information gathered over a lifetime. Although the idea of neurogenesis in the adult mammalian brain is still not generally accepted, these findings might help uncover a mechanism that would enable the human brain to repair itself through neurogenesis. Whether such replacement of neurons would disrupt complex learning processes or long-term memory is not known, but songbird research challenges scientists to identify the genes or hormones that orchestrate neurogenesis in the young human brain and to learn how to activate them in the adult brain.