It's worth noting that Fetz's group (Moritz et al., 2008) has recently described an invasive approach to brain-controlled orthosis conceptually similar to previous attempts based on non-invasive BCI mentioned above. In this experiment, a monkey, paralyzed via a nerve block, can regain control of its forearm by using FES and single cell recordings of the motor cortex. This brings us to an important underlying issue in the development of neuroprosthesis, namely the choice of the kind of mental task to use for control. In most work in non-invasive BCI, people use imagination of different limbs (right/left hand, feet) to deliver different commands to the neuroprosthesis for, say, the right hand. However, it seems more natural to rely on the recognition of different imagined movements of the same limb the neuroprosthesis controls. Initial evidence for such a possibility has been recently provided in an offline study where subjects imagined the execution of different wrist movements (Gu et al., 2009).