Learning and Relearning

Advances in robotics technology have created some new opportunities in neurorehabilitation, as we discuss in our article on page 1 of this issue. Several commercial ventures have already emerged that offer robotic systems that help restore lower-extremity or upper-extremity function following stroke and other neurological disorders.

We expect that there will be new market opportunities in this field for neurorehabilitation firms. But in our view, the real opportunity is not in manufacturing new hardware systems to market to therapists, but rather, devising new training protocols that automated rehabilitation devices can exploit to fine-tune the therapeutic regimen to a specific individual and perhaps even specific tasks.

Though there is considerable evidence that robotic therapy improves recovery from stroke, there is not universal agreement on how this improvement takes place or whether it can be generalized to overall function, as opposed to specific trained tasks. It also appears that the size and location of brain lesions can have a profound effect on the rate and pattern of motor recovery. And in all likelihood different robotic regimens will work best when paired with particular pharmaceutical or neuromodulation therapies.

H.I. Krebs,a pioneer in robotic neurorehabilitation at MIT, has proposed a model of stroke recovery that incorporates elements of learning theory. “Recovery is like motor learning,” he writes in an article in the Journal of NeuroEngineering and Rehabilitation. He proposes that implicit or “procedural” learning—acquisition without the awareness of the learned information—takes place during stroke motor recovery and that there are two phases of procedural learning of a motor task: the unskilled phase in early learning and a skilled phase in later learning. Each phase has different patterns of brain activation. Krebs plans to exploit this model to develop customized training paradigms for stroke recovery.

This philosophy of rehabilitation could lead to a whole new software industry offering training paradigms for robotic devices, with different paradigms offered for different combinations of disorder, lesion size and location, and complementary therapies such as drugs or magnetic stimulation. And the software-hardware metaphor could have application in other areas of neurotechnology besides neurorehabilitation. As NBR senior technical editor Warren Grill points out in his article on page 1 of this issue, a properly designed motor training regimen can dramatically improve the performance of a brain-machine interface.

The incorporation of learning and training concepts into neuro­technology products promises to expand the industry beyond hardware such as electrodes and pulse generators and into software offerings that magnify the capabilities of today’s devices.

James Cavuoto
Editor and Publisher