“Free Bird:” How to Stay on Key
If you’ve ever done Karaoke, then you know how horribly some people sing, especially when the background music is too loud for them to hear themselves. The ability to adjust to the environment is essential for all sensory systems, which use feedback mechanisms to modify behavior. Usually the environmental cues are contaminated by noise, so your brain has to decide whether to modify behavior based on sensory feedback (and risk “adapting” to signals that do not accurately reflect performance) or to ignore sensory input (and risk leaving errors uncorrected). So, how does your brain deal with this mismatch between the actual and expected sensory feedback, so you can have a better Karaoke performance?
Just like babbling baby humans, baby birds also learn to vocalize by listening to adults. Within days of hatching, Bengalese finches begin to imitate the sounds of adults. At first, their songs are extremely variable and disorganized, just like baby talk. As the birds get older, the variability of mistakes shrinks, until eventually, after a lot of practice, they can sing like their elders! It is hypothesized that this link between variability and learning may explain why youths tend to learn faster while adults are more resistant to change.
To help elucidate this theory, researchers at Emory and UCSF worked to quantify the relationship between the size of a vocal error and the probability of the brain making a sensorimotor correction. They put miniature headphones on Bengalese finches and used sound-processing equipment to change the bird’s pitch in real-time so that the bird thought it sang incorrectly. When the researchers made small pitch changes, the birds quickly corrected their errors, but as the pitch changes were made bigger and bigger, the birds learned less well and were eventually unable to learn at all!
The results of this study show that adult songbirds correct small errors more rapidly and robustly than large errors. While the mechanisms of song acquisition and error correction are probably different, it seems possible that they rely on the same underlying computation. Based on their data, the research team developed a model, and it showed that constraining learning based on variability could explain (1) the large changes necessary during song acquisition (when both error and variability are large) and (2) the stability required during adult song maintenance (when error and variability are low).
Hopefully, their findings will not only help human vocal rehabilitation therapies but also aid our general understanding of how the brain learns!
~Sober S.J. & Brainard M.S. (2012). Vocal learning is constrained by the statistics of sensorimotor experience, Proceedings of the National Academy of Sciences, 109 (51) 21099-21103. DOI:10.1073/pnas.1213622109 Images adapted from Image Source/Corbis and Sober S.J. & Brainard M.S. (2009). Adult birdsong is actively maintained by error correction, Nature Neuroscience, 12 (7) 927-931. DOI: 10.1038/nn.2336