Exploration and exploitation of action selection in the motor cortex and basal ganglia

The basal ganglia (BG) have been proposed as a possible neural substrate for action selection in the vertebrate brain. This hypothesis has been developed primarily through pathological observations. Human neurological disorders of the basal ganglia can result in movements that are slowed or eliminated (bradykinesia/akinesia in Parkinson's Disease) or conversely, uncontrolled or unwanted (e.g. in Huntington's Disease and Tourette Syndrome). However, the precise mechanisms by which BG circuits influence behavior remain to be understood. In this thesis, I have focused on determining the role of BG circuits in selection of well-trained actions, and how these findings can be applied for use in neuroprosthetic devices.;In the first study, I investigated one proposed mechanism to help resolve competition between actions in the BG: feedforward inhibition of striatal medium spiny cells (MSNs) by fast-spiking interneurons (FSIs). I recorded single unit activity from presumed MSNs and FSIs together with motor cortex and globus pallidus (GP), in rats performing a simple choice task. My findings support the idea that FSIs contribute to action selection processes within striatal microcircuits, but suggest that the feedback pathway from GP to FSIs may be particularly important for the suppression of highly trained yet unwanted actions.;In my second study, I examined the role of large neuronal ensembles of the BG and motor cortex during two variations on a simple action selection task. Analysis of local field potential (LFP) oscillations revealed that ∼20Hz rhythms beta20) were prominent during the hold period, but only if subjects were instructed on which direction to move during the hold period. This finding is consistent with the hypothesis that beta20 is involved with withholding specific selected actions, and agrees with pathological observations of increased beta20 in Parkinson's Disease.;In the third study, I examined how action selection circuitry can be exploited to aid in the development of a neuroprosthetic system. This system is one solution to a long standing problem in neural engineering: central nervous system neurons do not regenerate after traumatic injury, and can lead to paralysis. By bypassing injured neurons, we can allow for direct motor control from non-injured neurons. I developed an algorithm that observes the pattern of activity in cortical ensembles and allows both the subjects and control system to co-adapt their behavior to allow naive rats to use a neuroprosthetic device. The results of this study show that subjects can learn to select discrete actions in real-time using the neural activity of the cortex.;By developing a deeper understanding of the mechanisms behind selecting motor actions, we will provide further insight into such neurological diseases as Parkinson's Disease or Tourette Syndrome. In this thesis, I investigate action selection at the single-unit and multi-unit levels, while studying neural ensembles both within and across brain structures. Further knowledge in this field will also yield more sophisticated, yet more natural control of neuroprosthetic devices which will rely on native BG and cortical roles in action selection.