The current pattern recognition-based EMG control is highly patter-motion related. Once a maping of EMG patterns to desired motions is predefined, it is hard to re-configure and apply again. Considering the human hand grasp is controlled by C.N.S. in a coordinated way (synergy), rather than a one-to-one specific way, I am wondering if this finger coordination can be found in the EMG signals and thus through suitable algorithms these information can be extracted and applied to the prosthetic hand's grasp control.