This article is looking into some recent development that is taking place and being developed so far, for a thoughtcontrolled robotic hand, and to be used for prosthesis. Force and motion issues of such prosthesis and robotic hands, still remain a crucial problem that is to be looked into. Due to the impossibility of sensing and tactile feedback from a prosthesis to the brain, it is essential to have a local control to within the prosthesis to take care of unanticipated situation a hand should deal. In this sense, a signal generated algorithm will moreover be developed using patterns of hand motions and fingertips to spontaneously prevent a grasp of objects from being accidentally dropped once are disturbed. The article is proposing a learning mechanism for a grasp stabilization and control. The followed approach here is totally based on the use of Principle Components Analysis (PCA) to learn the massive patterns of prosthesis behaviors due to thought signals that are transmitted from the human brain to the hand mechanics.