+
+

Complex motor skills such as speaking a language or performing a violin concerto are learned through experience and practice. In many cases, learned behaviors must be performed accurately and precisely to be effective. They should also be performed reliably over time to stay effective. To acquire new motor skills, we use sensory experience to form an idea of the goal behavior and then alternate between making a movement and assessing the movement. Comparing a movement to the goal behavior and updating the next attempt becomes an iterative process of trial, error, and adjustment that shapes motor output toward the goal. Interplay of sensory input and motor output during practice trains the nervous system’s motor circuits to reliably produce novel sets of motor commands. Mastery of a motor skill indicates that the motor circuits generate the commands for motor performance that reliably matches the goal over time.

+

Once a motor skill has been acquired, the role of sensory experience in motor behavior changes. Motor circuits that produce goal behavior should be more resistant to modifications based on new sensory experience, so that motor output stabilizes and performance is accurate and reliable over time. While mechanisms of neural plasticity facilitate learning, mechanisms of neural stability must allow behavior to become stereotyped by minimizing changes in motor commands. These mechanisms likely decrease the impact of sensory experience on motor circuit function. Such mechanisms that inhibit sensory input from further changing motor circuit output once a skilled behavior is correctly performed are poorly understood.

+

A recent paper by Vallentin and colleagues (2016) reports important findings on motor circuit mechanisms that protect stable motor performance once a skill is learned. The authors studied motor circuits in the songbird brain that subserve song learning in juveniles and production of stereotyped song behavior in adults. Zebra finches were chosen as subjects because they offer unique behavioral and experimental advantages in the study of motor skill learning. First, each bird learns its song by forming an auditory memory of an adult tutor’s song, thus gaining a model of the goal behavior. Second, the bird goes through extensive practice to acquire the skill of producing its own copy of the tutor song. Initial attempts to sing yield rambling, unstructured sounds akin to babbling in human infants. Using auditory feedback to compare vocal motor output with the tutor song memory, the bird gradually develops an accurate copy of the tutor song. This learning process allowed the authors to examine motor circuit activity in conjunction with poor, partial, and complete mastery of song production. Third, song learning is followed by song stabilization; zebra finch adult song performance is one of the most stereotyped complex motor behaviors known to biology. This advantage allowed the authors to test alternative hypotheses for how song motor circuits stabilize and maintain song production. Fourth, song motor circuitry is simple and devoted solely to song behavior. A premotor region called HVC receives auditory input and sends motor commands to the region RA, which projects directly onto the motor neurons controlling the vocal organ. The simplicity and dedication of this circuitry allowed the authors to identify, record from, and manipulate motor circuit neurons with known connections.

+

Focusing on HVC projection neurons that send commands to RA (HVC-RA neurons), the authors recorded single-neuron responses to presentations of tutor song in juveniles still learning song and in adults with stable song. Many HVC-RA neurons in juveniles reliably responded to tutor song with spikes that occurred at the same point in the song each time it was played. No adult HVC-RA neurons responded to tutor song, however. These results indicate that auditory input to HVC drives HVC-RA neurons to fire in learning juveniles but not in adults that have mastered the skill of producing song—suggesting that exposure to the tutor song plays an instructive role in development of the HVC premotor circuit. Because critical periods for the influence of experience on sensory circuit organization close when inhibitory circuitry matures, the authors hypothesized that responses of HVC-RA neurons to auditory input are inhibited in adults. When they blocked inhibition in adult HVC with a GABAA receptor antagonist, they recorded responses in HVC-RA neurons similar to those observed in juveniles, suggesting a role for inhibition in making adult HVC-RA neurons insensitive to auditory input.

+

Vallentin and colleagues next examined inhibition in juvenile HVC, recording the firing patterns of interneurons and inhibitory synaptic currents in HVC-RA neurons, which receive input from interneurons. Interneuron firing rates were higher during presentation of tutor song than during silence. Intracellular recordings of HCV-RA neurons showed significant inhibitory input during tutor song presentation, and inhibitory currents occurred with reliable patterns across song presentation trials in some cells but not others. None of the measures of inhibition correlated with a subject’s age, however. Given that song learning progresses with age to adulthood, the finding that age did not predict differences in inhibitory activity was surprising, given that inhibition contributes to differences in juvenile and adult HVC activity.

+

The authors then hit on the hypothesis that led to their most important findings. They considered four factors: (1) HVC-RA neurons respond to tutor song in juveniles but not adults; (2) tutor song–evoked synaptic inhibition onto HVC-RA neurons varies widely in strength and precision; (3) measures of interneuron responses and synaptic inputs to HVC-RA neurons do not correlate with juvenile age; and (4) individuals develop copies of their tutors’ songs at different rates. These factors inspired the authors to examine their data with a new hypothesis and to conduct two more experiments to test it. The hypothesis was that inhibition in a motor circuit develops as the motor skill produced by that circuit develops. The experimental prediction was that the strength and organization of inhibitory input to HVC-RA neurons correlate with the acoustic similarity between the juvenile’s developing song and the tutor song. Testing the relationship between song learning accuracy and inhibitory input to HVC-RA neurons in each juvenile revealed that the strength and precision of inhibitory currents correlated with acoustic similarity between pupils’ and tutors’ songs. Intracellular recordings of tutor-song responses in multiple HVC-RA neurons from the same bird showed that synchrony of inhibitory inputs across cells scaled with the accuracy of song copying.

+

The final set of experiments manipulated song learning to test whether inhibition was strong during presentation of tutor song syllables that a juvenile accurately copied and weak during presentation of syllables that a juvenile did not accurately copy. The authors tutored juveniles with two different songs. First, juveniles learned to accurately produce a song composed of repeats of the same syllable (e.g., AAAA). Once birds accurately copied the first tutor song, they were tutored with a different song composed of the original syllable and a new syllable (e.g., ABAB). By recording from interneurons and HVC-RA neurons in juveniles that had and had not yet accurately copied the new syllable (B), the authors were able to test whether inhibition was stronger in response to accurately copied syllables than in response to poorly copied syllables. Recordings made during presentation of the new tutor song (ABAB) showed that interneuron firing and synaptic inhibition were significantly greater in response to learned syllables than in response to unlearned syllables.

+

In summary, these results suggest that motor circuit inhibition grows in strength and network coordination according to progress in motor skill learning, suppressing premotor neurons’ sensitivity to sensory input when motor output matches goal behavior. Vallentin and colleagues’ findings advance our understanding of neural mechanisms that facilitate and limit motor learning and provide important clues as to how motor circuits may be controlled to promote development and performance of motor skills.

Reference +
+
Vallentin  D, Kosche  G, Lipkind  D, Long  MA 2016. Inhibition protects acquired song segments during vocal learning in zebra finches Science 351:267–271.
[PubMed: 26816377]