Control of hand movements based on combinations of muscle synergies
Simon Overduin, Emilio Bizzi

We wish to address two questions: what are the organizational principles subserving the control of the hand, and how is this control implemented physiologically? It has been suggested that the complexity of limb control may be simplified by the presence and combination of fixed patterns of muscle activity. By activating muscles as "synergies", the nervous system may effectively reduce the degrees of freedom associated with the musculoskeletal system. Our project extends the synergy hypothesis to the primate hand. We will search for synergies in the patterns of electromyographic (EMG) activity observed in a variety of natural movements of the hand.

Our next goal is to investigate whether the neurons of the hand motor areas of the frontal lobe (especially primary motor cortex, MI) are related to the muscle synergies we have extracted with our computational procedure. To this end, we will utilize three complementary approaches: a) partial inactivation (muscimol) of areas within the MI hand region; b) microstimulation and NMDA iontophoresis of small regions of MI; and c) recording the activity of antidromically identified corticospinal neurons and interneurons from selected areas of MI. These areas will be selected according to the results obtained using muscimol inactivation and/or microstimulation. The issue at hand is whether or not the discharge of corticospinal cells can be represented by the amplitude and time coefficients of the muscle synergies we have extracted.

Cortical correlates of learning in monkeys adapting to a new dynamical environment
Andrew Richardson, Emilio Bizzi

We are recording from monkeys as they execute delayed, visually instructed reaching movements. In our experiments, monkeys learn to adapt to a new dynamic perturbation, namely a force field. By comparing the activity in the movements before and after the adaptation, we can dissociate the kinematics and the dynamics of the movements. Furthermore, we can dissociate the neuronal activity leading to the movement from the effects of the adaptation. Using this paradigm, we have previously studied the activity of neurons in primary motor cortex, the supplementary motor area, and premotor cortex. We found that neurons exhibit a learning-dependent plasticity, evident during both movement planning and execution. We are currently extending this series of investigations to the cingulate motor areas.

Modular control of natural motor behavior
Vincent Cheung, Jinsook Roh, Emilio Bizzi

We address the issue of how the central nervous system may coordinate the many degrees of freedom of the musculoskeletal apparatus to control motor behavior. In particular, we explore the idea that the control of limb movements is organized in a small number of modules that can be flexibly combined. We are testing the hypothesis that the muscle activation patterns observed in natural motor behaviors might be generated by linear combinations of a small number of muscle synergies. In one series of experiments, EMGs are recorded in intact frogs from fourteen leg muscles during different forms of locomotion (swimming, jumping, and walking) and defensive reflexes (wiping and extensor-thrust). We find synergies that are similar across different behaviors and frogs, though a few synergies appear to be recruited only in specific behaviors. Our results support a modular organization of the control of natural motor behavior and suggest that some of the modules may be organized in the spinal cord and shared across behaviors. A factorization algorithm is used to extract a set of synergies whose non-negative combinations can explain the majority of variation in the data.

TMS interference in human motor learning
Simon Overduin, Andrew Richardson, Emilio Bizzi

Investigations from several laboratories have shown that motor learning involves processes parallel to those involved in learning episodic information. For instance, learning a specific motor task is impaired if a second motor task is attempted shortly thereafter (retrograde interference) but not if the second motor task is learned four hours after the first task (motor memory consolidation). While the neuroanatomical basis for episodic memory has been well-delineated, the regions that support motor learning are less clearly defined.

Single cell studies in nonhuman primates adapting to a novel dynamical environment have been performed in our laboratory. These experiments have shown substantial plastic changes in motor and premotor cortex as the monkeys learn a reaching task. While such studies provide a great deal of knowledge concerning the areas involved in motor learning and the neural mechanisms of adaptation, they in general fail to prove the causal relationship of these areas with motor learning. An active role, for instance of M1, would instead be shown by studies using reversible focal modulation. In that respect, the technique of transcranial magnetic stimulation (TMS), allowing for the temporary and non-invasive stimulation of specific regions of the cortex, offers a unique opportunity to address these issues in humans.

Training motor control using a virtual environment
Maureen Holden, Emilio Bizzi

We have developed a new type of motor training system which utilizes a virtual environment (VE) and augmented feedback to enhance rehabilitation of the upper extremities in patients who have suffered neurological injury, such as stroke or traumatic brain injury. Our VE training system is designed to facilitate motor re-learning and motor generalization, and allows quantitative assessment of arm movements in 3-D. Key features of the system include Training Scenes (3-D "pictures" that are designed to elicit movements in a natural way by creating an environmental context and task goal for that movement), a Virtual "Teacher" who shows the correct movement by representing the trajectory of the limb's end point (or entire arm, if desired), an animated display of the Patient's Movement as he/she attempts to "imitate" the teacher in real time, a Scoring System which calculates the degree of "matching" between the teacher and patient trajectories, and multiple additional features which provide augmented feedback during performance.

We have used the VE system to study how patients with stroke and traumatic brain injury learn movements and the degree to which they can generalize what they have learned in VE to real world performance of both trained and untrained movements. To do this, we used scenes designed to elicit specific movements that were then targeted for training in the virtual world. Following VE training of these movements, we tested the patient's ability to perform these movements in both the virtual environment and in the real world. We also tested a variety of untrained movements to assess the amount and type of motor generalization that occurred. We have found the system to be effective for motor learning and generalization, as measured both by quantitative kinematic measures and by clinical measures of motor recovery, upper extremity function, and strength.