arts&sciences | Spring 2011

Name That Tune

A neuroscientist at CAS is figuring out how birds know which songs to sing.

By Mark Dwortzan

A picture may be worth a thousand words, but a single image produced by neuroscientist Tim Gardner can capture as many as 1,500. In his case, though, the "words" in question are the distinct vocalizations of a songbird, represented in visual form.

Gardner, an assistant professor of biology, uses these images to understand how birds build and retain the songs they use to communicate with one another. His work could shed light on how the neural circuits of learning and memory are encoded and maintained not only in birds, but also in humans—potentially boosting our understanding of the normal and diseased states of the human brain.

Gardner has chosen to focus his investigations on birds, rather than laboratory mice, or even humans, because few organisms exhibit such a quantifiable behavior. "In the last two years we have succeeded in translating sound into a new kind of image to capture the structure and the variants of birdsong," he says, "and we're now at the point where we can detect subtle changes in specific birdsongs."

Toward that end, Gardner subjects a colony of about 300 zebra finches and canaries—kept in soundproofed cages—to a variety of computer-controlled, quantitative behavioral experiments in BU's Laboratory of Neural Circuit Formation.

In one experiment, Gardner studied canaries raised in isolation from birdsong. While "tutoring" the birds with computer-generated, synthetic songs that depart from species-typical songs, he and his lab recorded every sound the birds uttered through their development. Initially, the subjects imitated the synthetic songs with great accuracy, but as they matured, they reverted to species-typical songs, even in the absence of other canaries.

"There's a complex program that ultimately builds each species-specific song," Gardner says, noting that both genetic and environmental factors contribute to the process. "We're interested in determining the local neuronal rules that govern this amazing process."

To home in on these rules, Gardner is now investigating regions of the brain that encode song patterns. These areas produce a dynamic pattern of song while the birds are singing—and also, surprisingly, when they are asleep. Gardner has produced time-lapsed images of neuron growth in vivo which show the development of new neuronal processes in sleep.

Images courtesy of Tim Gardner

Tim Gardner and his colleagues notated the song patterns of juvenile canaries, top diagram, that had never heard normal species-specific songs, and found that they imitated abnormal synthetic songs with great accuracy, even when the tutor songs lacked phrasing, or what Gardner's team calls "short stereotyped syllables" that are repeated as the bird sings.

A newborn cell tagged with green fluorescent protein, above, shows overnight growth in a song control center of the zebra finch brain as observed in vivo before sleep, left, and after a good night's rest, right.


Gardner hopes to learn more about the growth of these processes by introducing small perturbations to the spontaneous activity that occurs during birds' sleep, and then by observing the impact of those perturbations on their neural networks and songs. As the bird dreams of its own songs, or sings upon awakening, a computer detects the sleeping pattern and triggers a stimulating electrode or implanted optical fiber to induce slight changes in neuronal electrical activity at specific locations in the song pathways.

"If we can increase or decrease the neuronal activity of the bird during sleep, we can see if there's a change in the sequential order—and creativity—of the songs it produces," says Gardner, noting that such studies could help us understand how similar neuronal changes in humans might impact our performance during the day.

This article first appeared in Boston University Research 2010.