Behavioral Investigation of the Motor Code Underlying Variable Sequence Generation in Bengalese Finches

Understanding the neural mechanisms underlying sequential behaviors is a critical question in systems neuroscience. Successful execution of a behavioral sequence requires the nervous system to specify the structure and duration of the sequence elements in addition to the order in which they are to be produced. Birdsong is an ideal model system to probe the mechanisms involved in the control of these different aspects of sequential behaviors. Previous studies have shown evidence for the compartmentalization of control of pitch, duration and sequencing in the songbird brain. On the other hand, the recurrent nature of the circuitry and systematic correlations between sequencing, spectral structure, and timing point to linked control.

To investigate possible links in the neural representation of pitch and timing, we exploited the correlation between these features at branch points in the song of Bengalese finches. Using a popular operant conditioning tool, we shifted the pitch of the branch syllable and tested if the subsequent sequencing changes. Our analysis led us to the novel finding that sequencing changes when the pitch is shifted and that these changes persist even after pitch recovers to near-baseline values. Furthermore, the nature of the sequence changes is not well predicted by the baseline correlation between pitch and sequencing.

We also created a new method that can drastically improve the yield of the operant conditioning experiments mentioned previously. A spectral template is used in these experiments to detect target syllable in real-time. The conventional method of making the spectral template by averaging the spectrum of the syllable leads to a lot of false positives for a number potential targets, making it impossible to target them. Our method computes an optimal template for a given syllable that maximizes discriminability. Our method improves the percentage of targetable syllable from 7% to 26%.