In humans, basal ganglia circuits are involved in the planning and initiation of complex motor programs, as well as in cognitive and affective functions. Two major disorders-Parkinson Disease (PD) and Huntington Disease (HD)- provide dramatic examples of the critical roles of two well-characterized basal ganglia pathways (direct and indirect motor loops) in controlling movement. In PD, dopamine production in the substantia nigra pars compacta is severely reduced, leading to hypokinetic manifestations (slowed movements, rigidity, postural instability). In contrast, HD is a hyperkinetic disorder, characterized by erratic, dancelike movements, resulting from degeneration of striatal neurons in the indirect pathway.
Reinforcement learning of complex motor sequences in vertebrates is dependent on the ability both to explore a variety of motor patterns, and to evaluate and compare the performances. Just as for human mastery of complex motor sequences (e.g. swinging a polo mallet, or playing a musical instrument), reinforcement learning of song in birds requires both exploratory motor behavior, and comparative evaluation of those performances. Ölveczky et al. (2005) examined the role of a basal ganglia component, the lateral magnocellular nucleus (LMAN), in producing vocal variability in juvenile zebra finches.
Two neuroanatomical pathways are involved in song production and song learning: 1) the “motor pathway” involves several premotor nuclei, with the vocal nucleus HVC providing timed inputs, and the robust nucleus of the arcipallium (RA) providing control of brainstem vocal and respiratory motor nuclei, and 2) the “anterior forebrain pathway” (AFP), homologous to basal ganglia circuits through the thalamus and cortex, and critical for song learning. The LMAN component of the AFP does not respond to auditory feedback, and in this paper, the authors address the alternative hypothesis that this nucleus of the nidopallium is involved in generating vocal variability.
In the first set of experiments, the LMAN was temporarily inactivated in juvenile zebra finches, by stereotaxic injection of the sodium channel blocker tetrodotoxin (TTX). This manipulation resulted in a reduction in song variability, even in young birds at the earliest phases of song learning. Song variability was assessed by analyzing song syllables, the basic acoustic units of zebra finch song, and comparing these syllables with the less variable, more stereotyped songs of adult birds. To ensure that LMAN specifically, and not nearby fibers, was inactivated, the investigators stereotaxically injected a GABA receptor agonist, muscimol, bilaterally into LMAN. GABA is an inhibitory neurotransmitter, especially important in the disinhibition mechanisms of basal ganglia circuits. Again, song variability was greatly reduced by this manipulation, and both syllable structure and sequence were affected.
To determine whether LMAN neurons were directly responsible for song variability, through fast synaptic input, the investigators recorded from LMAN neurons. Raster plots of these data indicated that the patterns of spikes and bursts produced by LMAN neurons differed each time the juvenile finches sang; however, LMAN activity, though variable, is not entirely random. In the final series of experiments, the researchers used an NMDA receptor antagonist, AP5, to show that excitatory glutaminergic input from LMAN to RA is responsible for generating song variability.
The results described by Ölveczky et al. (2005) differ from previous LMAN studies in that a reversible effect, on a short time-scale, was demonstrated. The premotor circuit is viewed as a generator of stereotyped song, upon which the AFP acts to increase variability across different song motifs. In addition to its function in song learning in juvenile birds, the LMAN may also generate variability in adult finch song that is not directed at females. The authors suggest that biased firing patterns in the LMAN may increase the efficiency of song learning, and apply their findings in birds to the general question of how the vertebrate brain produces the fluctuations in motor sequences that are required for reinforcement learning.
*First post in the Wednesday Wings series = blogging about peer-reviewed research that involves wingéd animals*
Ã–lveczky, B.P., Andalman, A.S., Fee, M.S. (2005). Vocal Experimentation in the Juvenile Songbird Requires a Basal Ganglia Circuit. PLoS Biology, 3(5), e153. DOI: 10.1371/journal.pbio.0030153