UCSF

Birdsong is an ideal behavior for testing the neural mechanisms that underlie motor learning, as the dedicated neural circuits that control song production and learning are well-elucidated and amenable to experimental manipulation. Here I investigate how the anterior forebrain pathway (AFP), an avian cortical-basal ganglia circuit long implicated in vocal plasticity, contributes to two hierarchically distinct forms of adaptive song modification in adult songbirds. First, I tested how the AFP contributes to adaptive modifications of the acoustic structure of individual song syllables, both in the context of reinforcement-driven learning and a self-driven recovery process. I found that, in both contexts, transiently interfering with the AFP's output to the primary motor pathway reversed the expression of newly learned changes to syllable acoustic structure, but that over multiple days, the expression of learning consolidated to become AFP-independent. This finding supports an emerging view that cortical-basal ganglia circuits can direct the initial expression of learning via top-down influences on primary motor circuits. Second, I tested how adult Bengalese finches control the stochastic sequence transitions that occur between syllables. I found that the probabilities of transitions are ordinarily stable, yet birds have a previously unrecognized capacity to adaptively modify these probabilities in response to differential reinforcement. Despite this capacity for modification, when reinforcement was terminated, birds gradually restored transition probabilities to their baseline values, indicating that these values are set points that the nervous system has the impetus to restore. These findings suggest that the statistics of sequence variation in a motor skill can reflect an end point of learning that is actively maintained via continual self-monitoring. Finally, I tested the extent to which the adaptive modification of syllable acoustic structure and syllable sequencing rely on the same basal ganglia circuitry. I found that lesions of LMAN, the AFP's cortical outflow nucleus, prevented the adaptive modification of syllable structure but did not affect birds' capacity to adaptively modify syllable sequencing. Therefore, although similar principles of differential reinforcement can be used to drive changes to syllable structure and syllable sequencing, an avian basal ganglia circuit contributes differently to these hierarchically distinct forms of learning.