For years, nobody had followed up on that question.
"We framed the question in a different way," Fee says of his research with postdoctoral fellow Bence P. Ölveczky and graduate student Aaron Andalman. "We said, this young bird is being creative, exploring many different sounds through trial and error. We hypothesized that the AFP is the source of this creativity, generating the variations, rather than comparing them."
To test this theory, Fee's team studied finches that were just old enough to begin their vocal explorations. The researchers temporarily inactivated the part of the AFP connecting to the motor system used in producing songs. That inactivation shut down all the variability, temporarily stranding the young finch with an immature version of the song.
These results suggested that the AFP circuitry itself causes the juvenile bird's experimentation with various sounds and sequences, and that such explorations are essential to learning songs. Deactivating the AFP after a bird had already learned the correct song had no affect on its continued proficiency.
The researchers then learned that the AFP neurons produce random bursts of activity coinciding with new variations in the practice routine.
"We think the bursts of these neurons 'kick' the motor pathway that is producing a song, jarring it out of the routine and making it sing something new," Fee says. Then another, still unexplained, pathway compares that variation to the bird's memory of the father's song. Gradually, the bird gets it right more often and eventually sings only the songs of its elders.