Toward a Full Prehension Decoding from Dorsomedial Area V6A

Neural prosthetics represent a promising approach to restore movements in patients affected by spinal cord lesions. To drive a full capable, brain controlled, prosthetic arm, reaching and grasping components of prehension have to be accurately reconstructed from neural activity. Neurons in the dorsomedial area V6A of macaque show sensitivity to reaching direction accounting also for depth dimension, thus encoding positions in the entire 3D space. Moreover, many neurons are sensible to grips types and wrist orientations. To assess whether these signals are adequate to drive a full capable neural prosthetic arm, we recorded spiking activity of neurons in area V6A, spike counts were used to train machine learning algorithms to reconstruct reaching and grasping. In a first work, two Macaca fascicularis monkeys were trained to perform an instructed-delay reach-to-grasp task in the dark and in the light toward objects of different shapes. The activity of 89 neurons was used to train and validate a Bayes classifier used for decoding objects and grip types. Recognition rates were well above chance level for all the epochs analyzed in this study. In a second work, monkeys were trained to perform reaches to targets located at various depths and directions and the classifier was tested whether it could correctly predict the reach goal position from V6A signals. The reach goal location was reliably decoded with accuracy close to optimal (>90%) throughout the task. Together these results, show a reliable decoding of hand grips and spatial location of reaching goals in the same area, suggesting that V6A is a suitable site to decode the entire prehension action with obvious advantages in terms of implant invasiveness. This new PPC site useful for decoding both reaching and grasping opens new perspectives in the development of human brain-computer interfaces.