Erich Jarvis investigates the neurobiology of learned vocal communication in the rare group of animals that have this ability, as a model for the study of how the brain generates, perceives, and learns complex behaviors, such as spoken language. His specific quest is to determine the molecular mechanisms that construct, modify, and maintain neural circuits for vocal learning and then engineer brain circuits to repair and enhance those behaviors.
Molecular Mechanisms of Vocal Learning
Vocal learning, a critical behavioral substrate for spoken language, is a rare trait so far found in few species. The vocal-learning species we study include the song-learning birds (songbirds, parrots, and hummingbirds) and humans. We have screened for differences at the DNA, RNA, and proteomic levels for mutations and brain expression in species that can imitate vocalizations relative to those that cannot. We identify candidate genes, and we will be testing their role in developing and maintaining brain circuits for vocal learning. The results are expected to yield insight into the mechanisms and evolution of brain systems for speech and other complex behavioral traits. Techniques used include high-throughput molecular biology and proteomic screening, animal brain surgery, microdissections, gene manipulation, and animal behavior.
Behaviorally Regulated Gene Networks
Sensory processing (e.g., hearing) and performance of learned motor behavior (e.g., singing) are associated with robust activation of gene cascades in different neuron types. We call this phenomenon, which is best studied in songbirds, "behaviorally regulated gene expression." These gene cascades are activated by the neural activity that occurs during sensory processing or behavioral performance. We have now identified more than 2,000 genes that are activated in the vocal-learning system, ~10 percent of the genome. The future goals of this project are to determine the gene interactions in gene-behavior networks and decipher their functions in cellular processes and in learned behaviors. Techniques used include high-throughput molecular biology, computational biology, proteomic screening, gene manipulations, and animal behavior.
Computational Molecular Neurobiology
In this project, we develop and utilize inference algorithms on high-throughput gene microarray data to infer gene regulatory networks. The gene regulatory networks we are currently studying are those regulated by singing behavior in the songbird brain. Our goal is to understand the system of genes and neural activity that becomes activated in the brain during the production and learning of a behavior. The developed algorithms, however, will be useful for many questions in systems biology.