Singing karaoke makes most of us quickly realize how poorly we can carry a tune. Typically, when we're confident enough to sing on stage, we're deaf to the negative feedback. Recently, however, neuroscientist Jesse Goldberg at Cornell University, USA, discovered a brain region that listens and produces a chemical ‘booing’ when birds sing off key. Vikram Gadagkar and colleagues in the Goldberg lab report that a special group of dopamine neurones in bird brains critically evaluate and encourage when birds chirp well.
Like karaoke, birdsong sounds great when performed well. Learning to hit the right notes is hard work, though. To sing properly, songbirds must first memorize a model tweet early in life, continue to repeat and refine their imitation of that song during development, and eventually master the model song reproduction. Practice makes pitch perfect, but how the brain rewards good singing and discourages mangled squawks is unclear. One thought is that dopamine, a chemical messenger important for reward and reinforcement in the brain, may play a role.
To test whether dopamine neurones provide neural feedback on song production, Gadagkar made birds bungle their song. Placing male zebra finches (Taeniopygia guttata) in acoustic isolation chambers equipped with a microphone and speakers, Gadagkar set up a program that automatically detected and disrupted a single note in a bird's song. When the finches belted out the target note, a discordant tone was played over their own rendition, making the birds believe they had tweeted off key. All the while, Gadagkar monitored their activity through delicately implanted recording wires in the ventral tegmental area (VTA), a midbrain dopamine region.
Gadagkar soon discovered that a subset of VTA neurones were keenly aware of the birds’ synthetic song errors. The VTA ‘error neurones’ quickly shut down brain activity whenever the birds perceived that they had sung a note incorrectly. Even more remarkably, when the same target note was uninterrupted, the VTA error cells ‘rewarded’ the correct production and rapidly increased firing during the note's production. Taken together, VTA error neurones appear to reward birds by bathing the brain in dopamine when they hit the right note. What remained to be seen was whether VTA error neurones tracked how often birds performed correctly – did VTA error neurones ‘cheer louder’ (more activity) when a note was accurately sung occasionally versus frequently?
In follow-up experiments, Gadagkar assessed whether VTA error neurones produced graded ‘feedback’ in proportion to how often birds sang a note correctly by interrupting two different notes in the bird's song with different probabilities. Birds perceived they sang one note incorrectly 50% of the time (note A), whereas they perceived singing a different note wrong only 20% of the time (note B). Singing either note incorrectly suppressed VTA firing to a similar degree, irrespective of how often it was mispronounced. However, when the notes were sung correctly, tweeting note A ramped up neural activity three times higher than when birds sang note B. Therefore, when a bird correctly utters a note it struggles with more often (note A), VTA error neurones provide more neural reinforcement to encourage the accurate imitation.
Taken together, Gadagkar and colleagues’ findings demonstrate an error-responsive group of neurones that dishes out dopamine activity as a reward when birds sing well. While everyone is their own worst critic, it turns out the VTA is the brain's harshest audience but most supportive instructor.
