• premotor
• reach and grasp
• neural decoding
• machine learning
• brain machine interface
• BMI
• cortex
• spike trains
• spike sorting
• spike distance
• neuroprosthesis
• neural manifold

In this thesis, neural data recorded from experiments with a non-human primate were used, specifically from a monkey (macaca mulatta) restrained in a primate chair performing reach and grasp (R&G) tasks. Four multielectrode arrays (FMA) were implanted in the monkey's premotor cortex. Useful signals for Brain Machine Interfaces (BMI) were recorded. The aim was to test different neuro-computational decoding methods to predict the intended hand pre-shaping and the planning phases of an upper-limb related movement. These algorithms can be applied in the field of neuroprosthesis to restore upper-limb functions in tetraplegic or amputee patients. Thanks to the locus of implantation, it is possible to access also more cognitive aspects of reach and grasp movements as the intention and planning. A digital pre-processing and a fully automatic spike sorting were applied to the recorded neural data to extract putative single-unit spike trains. Successively the spikes were used to build classifiers using both single neuron or ensemble techniques, above all neural manifolds and summed subpopulations approaches.