paper

Learning to Control a Brain–Machine Interface for Reaching and Grasping by Primates

  • Authors:
Read the paper

📜 Abstract

Reaching and grasping in primates depend on the coordination of neural activity in large frontoparietal ensembles. Here we demonstrate that primates can learn to reach and grasp virtual objects by controlling a robot arm through a closed-loop brain–machine interface (BMIc) that uses multiple mathematical models to extract several motor parameters (i.e., hand position, velocity, gripping force, and the EMGs of multiple arm muscles) from the electrical activity of frontoparietal neuronal ensembles. As single neurons typically contribute to the encoding of several motor parameters, we observed that high BMIc accuracy required recording from large neuronal ensembles. Continuous BMIc operation by monkeys led to significant improvements in both model predictions and behavioral performance. Using visual feedback, monkeys succeeded in producing robot reach-and-grasp movements even when their arms did not move. Learning to operate the BMIc was paralleled by functional reorganization in multiple cortical areas, suggesting that the dynamic properties of the BMIc were incorporated into motor and sensory cortical representations.

✨ Summary

Summary

  • The study demonstrated that two macaque monkeys could use a closed-loop brain–machine interface to control a robotic arm for cursor movement, gripping-force regulation, and coordinated reach-and-grasp behavior.
  • Neural activity was recorded from multiple frontal and parietal cortical areas and decoded with parallel linear models to estimate hand position, hand velocity, gripping force, and arm-muscle electromyography. Large neuronal ensembles provided more accurate predictions than individual neurons or small samples, and multiunit activity was sufficient for effective control.
  • Performance improved with continued training. After the robot was introduced, performance initially declined because the animals had to adapt to the robot’s dynamics, but it subsequently recovered. The monkeys also controlled the system without overt arm movements or corresponding muscle activity.
  • Learning was associated with changes in neural contributions, directional tuning, and correlations among neurons across several cortical areas. The authors interpreted these changes as evidence that the brain incorporated the dynamics of the artificial actuator into motor and sensory representations.
  • Influence: Later reviews of intracortical brain–machine interfaces and neural plasticity cite this study as an early demonstration of ensemble-based control of coordinated robotic reaching and grasping, and as evidence that BMI learning involves distributed cortical adaptation. It is also included among the foundational studies discussed in broader reviews of brain–machine interfaces and neuroprosthetic development. (journals.plos.org)