Abstract-Dexterousmulti-finger robotics hands/or prosthesis hands are complicated devices to model, control, and to motorize. Modeling involves building coordinated kinematics relations, while the dynamic model involves grasping forces and optimized distributions of forces and torques at grasping locations. Over the last thirty years or more of research, a coordinated control of fingers for such devices was done analytically, however such control issues were facing few number of difficulties. Therefore, the purpose of this paper is to look at novel approach for defining grasping patters from EEG readings, then to learn-mirror such patters into robotics handprosthesis. We shall create an association between fingers motions, forces, and particularly detected EEG brainwaves from human. Such an association is very useful for robotics humanoids, or for prosthesis. The association between human EEG to robotics is modeled here, and it will be used for grasping by system robotic by learning (via training) a robotics multi-finger dexterous hands. In addition, such an association is also useful for controlling a prosthesis for rehabilitations purposes.