Many animal species communicate vocally, but most use hardwired, innately determined sounds. Songbirds are among the few animals that learn their communication sounds, as humans do, and songbirds are easier to study in the laboratory than other vocal learning species. Just like human infants, young songbirds must hear a tutor during a sensitive period in development and engage in extensive vocal practice that requires auditory feedback and gradually transforms simple sounds into copies of the model. Neuroanatomical studies have shown that song learning and production are controlled by a specialized set of interconnected brain structures highly developed only in species that learn their songs, such as the zebra finch. Among songbirds, the zebra finch is the dominant laboratory model system used to study the vocal imitation process, and a large literature describes the neural bases of song learning in this species. Zebra finches breed well in captivity and sing all year round. Their song is relatively simple and highly stereotyped once learned. Young zebra finches reach adulthood in ~3 months, enabling us to study the whole song development process in a short period of time.

There are many parallels between human and songbird vocal learning. In each case, the youngster must 1) select the conspecific signals to be imitated from a complex acoustic environment (a process that is enhanced by social interaction with the tutor); 2) remember those sounds as a reference to be compared to its own vocal attempts; and 3) make simple vocal sounds that are successively refined, differentiated, and sequenced through a feedback process until they match the tutor’s song or speech. Thus, although birdsong does not have the semantic and syntactic complexity of language, shared principles seem to govern both song and speech learning. Despite neuroanatomical differences between the avian and human brains, we believe that, at the circuit level, similar computational operations must underlie this form of sensorimotor learning.

Understanding vocal learning is important because it may yield insights into brain mechanisms of developmental learning, including human speech acquisition. In addition, by studying the dynamic relationship between multiple levels of motor control, these studies address how hierarchically organized circuits contribute to the production of complex motor behaviors. Song development provides an opportunity to examine how complex behavioral patterns emerge and evolve during development. Although the phenomenology of vocal learning in songbirds and its anatomical substrate have been described, there exist no electrophysiological studies of vocal production in juvenile birds during song learning; thus we know very little about the physiological mechanisms that enable sounds heard by the bird to sculpt its vocal output so exquisitely.

The specific goal of our project is to advance our understanding of the neural mechanisms of vocal learning by providing a quantitative description of the relationship between physiological variables and vocal performance over the course of development in a songbird, the zebra finch. We propose to study vocal learning dynamically across neuronal and peripheral subsystems, using a novel collaborative approach that will harness the combined expertise of several investigators. Our proposed research model will 1) provide simultaneous measurements of acoustic, articulatory and electrophysiological data that will document the detailed dynamics of the vocal imitation process in a standardized learning paradigm; and 2) incorporate these measurements into a theoretical/computational framework that simultaneously provides a phenomenological description and attempts to elucidate the mechanistic basis of the learning process.