How to make a brain perceive an exoskeleton as its new body

[rhoenix]

Miguel Nicolelis is a man with a lot of pressure on his shoulders. Since claiming that he will build the robotic exoskeletal suit that enables a paraplegic to perform the opening kickoff during the next world cup, he has been scrambling to make good on his self-imposed mandate. By all measures, he has logged several important advances en route to that goal this year alone. The latest offering from his lab at Duke provides an important link into how an exoskeleton will be incorporated at the cortical level, into the so-called body schema. In other words, how the mind comes to perceive its new self.

The “rubber hand illusion” is a well-known phenomenon which illustrates how plastic our body schema can be. It is best experienced by visually substituting a mannequin arm for your own arm, and applying a physical stimulus like a stroke or prick, to both. In a matter of minutes, your perceptions and concerns begin to morph to what is going on at the mannequin arm. This happens to the point that your brain begins to identify closer with the fake arm. In support of this view, many psychologies like to point to several studies that have shown that the real arm decreases in temperature under these circumstances.

It is conceivable that a robotic exoskeleton could just be programmed to perform a simple kick on its own. If the timetable of Nicolelis does not match up exactly to the World Cup’s, he may not be hung completely out to dry. His new experiments already show that he has reached a new fundamental understanding in how prosthetic devices can be integrated at the neural level. Building on his experience in creating the first successful BMIs, and more recently setting the record for the most simultaneously isolated and recording neurons using microelectrode arrays (up to 2000 by his own estimates), he has now begun to explore what happens in two key areas of the cortex during the integration process.

Nicolelis and his team recorded from cells in the primary sensory and motor cortex of monkeys, while the monkeys were presented with a 3D avatar-like arm in place of their own. In this virtual rubber hand illusion experiment, neurons in both cortical regions initially responded just to the stroking just of the real arm. In time however, they began to respond to the virtual stimulation alone, indicating that some direct neuronal adaptation had occurred here in the brain. This adaptation conceivably reflects the perceptual adaptation that occurs in humans doing an analogous task. These responses tended to occur around 50 to 70ms later than the responses to the real stimuli, which is consistent with the driving inputs coming into these areas from visual areas that are a single synapse away.

These discoveries are very important for Nicolelis’s project because in order to convincingly demonstrate that the wearer is controlling the robotic suit, a sufficiently complex, real-time task needs to be performed. In other words, what a critical audience wants to see is something not only fast and creative, but also something spontaneous, like a reflex. The difficult part is that a reflex generally occurs in the absence of the ponderous thought that goes into the typical slow BMI devices that we may have seen before. They key point here, is that nobody is going to generate a reflex for a temporary prosthetic suit that is not richly interwoven at the deepest, most protected level with our core suite of responses.

Creating this level of perceptual integration involves not just one or two senses gaining control over the muscle outputs used to drive the prosthetic movement, but every sense, and perhaps even some emotional channels as well. It will be of little use if a wearer has to retrain their brain completely every time they remove their limb or suit, essentially beginning again from scratch. The side-effect of this intimacy is that the absence of the machine will be sorely felt, perhaps irreversibly so. If the integration is good enough, one might even expect to experience a full-blown phantom limb effect when a prosthetic is removed.

The good news is that this most recent data indicates that much of the perceptual integration comes for free simply through a little exercise and training. Undoubtedly, regions beneath the cortex play a critical role in shaping these perceptions, and ideally, BMIs will extend into many of these areas as well. I chatted recently with Nicolelis in an open forum, and he gave indication that in just a few weeks, he will be publishing even more revolutionary results in this area. The ability of our brains to map new additions to ourselves is potentially unlimited, whether the manipulation is flesh or machine. Re-mapping a surgically enhanced hand, like that in the picture above, might require more cortical space than we now have at hand. Once the mapping processes are better understood, adding in these modules might eventually be done with ease.

An important advancement in the works not just for the disabled, but for other areas as well.