Cortical circuit dynamics during vocal learning in a songbird

Abstract

Vocal learning in humans and songbirds occurs during a sensitive period in development. Oscine songbirds, such as zebra finches, memorize a tutor song during the sensory phase and then, using auditory feedback, match their own vocalizations to the tutor song memory during the sensorimotor phase. Songbirds possess a set of anatomically distinct brain nuclei that are dedicated to vocal learning. HVC is a telencephalic song nucleus with auditory and motor activities. The cellular and neural circuit determinants of sensitive periods, which are characterized by enhanced plasticity, are not well understood. Activity in the sensorimotor song control area HVC changes during song learning (Crandall et al., 2007b). For example, neurons in the HVC of juveniles have longer, weaker bursts than those in adults. Changes in the circuitry of HVC that may underlie these developmental changes are not known. We have examined how population bursts in HVC change during sensorimotor learning in the zebra finch. First, we found that bursts of activity in HVC predict stability in song during singing. This led to the hypothesis that bursts in HVC were stronger in the afternoon when song is known to be more stable (Deregnaucourt et al., 2005). We found that bursts in HVC increase each day and during development. To examine changes in HVC burst activity during song learning, we recorded ensembles of HVC neurons using multiple tetrode recording in anesthetized juvenile and adult zebra finches. Arrays of tetrodes, such as those we have used, enable simultaneous recording of many neurons and analysis of their functional interactions. To identify specific cells in the ensemble, we antidromically stimulated projections from HVC to other song nuclei and coclustered antidromic and spontaneous spikes. This method enables the study of identified HVC projection neurons in the context of the functioning circuit. With combined tetrode and antidromic methods, we have begun to investigate interactions among HVC neurons. We have found that both efferent projections of HVC are active in population bursts, and that the class of neurons that project to a brain pathway known to generate song variability increases its bursting activity in adults. We also found that a population of functionally inhibitory neurons exhibits prolonged bursting in juveniles, which are active in developing sensory systems during a sensitive period. Collectively, these data implicate HVC in song learning. We speculate that bursts of activity in HVC, by virtue of a change in inhibition, limit song variability over time during sensorimotor learning.