Integrated Calendar

Songbirds are a great model for studying how the brain solves the challenges of vocal imitation, since, like human infants, young songbirds learn to produce complex vocal sequences that are exact copies of those of adult conspecifics. This feat is thought to be accomplished by matching the bird's motor performance to a memorized sensory template. To study this process experimentally, we use a computer interface that presents birds with specific vocal imitation tasks and records their entire vocal output during the process.
Applying this methodology to vocal combinatorial learning, we trained juvenile zebra finches to swap syllable order in their song, or insert a new syllable into a string. Surprisingly, solving these tasks required a prolonged stage of learning new transitions between syllables one by one, indicating that the ability to rearrange vocal sounds is not the starting point of vocal learning, but a laboriously achieved endpoint. Analysis of babbling development data of human infants revealed that infants face a similar challenge in acquiring new transitions between syllables, suggesting that birds and humans share a common developmental stage of gradually learning to combine sounds into sequences.
In a current set of experiments, I am testing hypotheses about the computations involved in sensori-motor vocal learning. For example, is the motor output matched to the sensory template as a single unit, or is the match computed independently for different levels of the song hierarchy? Preliminary results suggest that matching vocal performance to the template occurs independently on at least two levels: the level of individual syllables, and the level of syllable sequences, suggesting that learning on these levels is carried out by distinct neural mechanisms.