Speech production and control is disrupted in a number of neurological diseases that involve the basal ganglia. Notably, hypophonia and hypokinetic dysarthria (characterized by decreased motor gain) are prevalent in patients with Parkinson's disease (PD). Deep brain stimulation (DBS) of the subthalamic nucleus (STN) produces predictable improvements in other motor symptoms of PD but does not result in consistent improvement in speech and can negatively impact language function. These observations and other accumulating evidence indicate an important role for the basal ganglia in speech. However, a major impediment to developing treatments for speech deficits in movement disorders and reducing speech-related side effects of DBS is the absence of a neurophysiological model for basal ganglia participation in speech production. Testing how general tenets of basal ganglia organization and function apply to the speech motor system presents both unique challenges for clinical neuroscientists and significant opportunities to advance the cognitive neuroscience of speech production. Our overall goals are to determine how motor and linguistic speech information is encoded at multiple levels of granularity within the STN-cortical network, and to determine the relationship between neural activity within the STN-cortical network and the gain of vocal output. Despite the fact that electrophysiological data obtained during DBS surgery offers the unique opportunity to directly assess basal ganglia neuronal activity during speech, this paradigm remains remarkably unexplored. Our central hypothesis is that the STN contributes at multiple levels to the hierarchical control of speech production. Using a completely novel approach, we will rigorously test this hypothesis by simultaneously recording STN units, STN and cortical local field potentials (LFP), and spoken acoustics while PD subjects perform a speech task during DBS surgery. To test for encoding at different levels of granularity, we will explore the extent to which neuronal activity in the STN codes for articulatory and linguistic features associated with different levels of representation within the speech production system (Aim 1). To test for a role in voice modulation, we will explore the extent to which the STN codes for measures of gain, such as volume, pitch and fluency (Aim 2). Additionally, we will directly assess the causal role of STN function in speech production by delivering disruptive stimulation to the STN (Aim 3). A major strength of our project is the complimentary nature of extensive, multi-disciplinary expertise from team members at the University of Pittsburgh, Johns Hopkins University and Carnegie Mellon University. This combined expertise allows us to employ a novel combination of classical analytic methods and more recent machine learning methods for supervised and exploratory analyses to document the neural dynamics of STN and cortical activity during speech production.

Speech production and control is disrupted in a number of neurological diseases that involve the basal ganglia, for instance hypophonia and hypokinetic dysarthria are prevalent in patients with Parkinson's disease (PD). We will use a novel experimental approach and combination of analytic techniques to elucidate the contribution of neural activity in the subthalamic nucleus to the hierarchical control of speech production, in subjects with PD undergoing deep brain stimulation surgery.