 |
| Miguel A. L. Nicolelis |
Holy cow, what an amazing talk I just went to. I'm at a conference in Buenos Aires, called "
Engineering in Medicine and Biology". It features many scientific and engineering advances in fields like medical imaging, bioinformatics, and neuroscience. I make it a habit to attend the plenary talks because they're meant for a general audience, and they're often a good way to see a nice overview of a field of research.
This afternoon, I attended a plenary keynote lecture called "Towards a Whole-Body Neuroprosthetic", by
Miguel A. L. Nicolelis of Duke University. What an eye-opener!
The theme behind his research -- implantable arrays that can read electrical activity from thousands of neurons. This activity is used to derive numerical models of behaviour. For example, a monkey is trained to play a little computer game in which it has to move a cursor into a circular region on the screen by manipulating a joystick. While this is going on, an implanted probe in the monkey's motor cortex reads off the neural activity associated with the movements. Later on, the experimenters remove the joystick, and instead derive the INTENDED motions directly from the monkey's neural activity. And it works beautifully! The monkey is able to control the cursor with its mind (similar to classical EEG-based attempts, but no training is needed; the device reads your intentions from your brain's natural internal language).
This technology could be used in corporeal augmentation for the paralyzed (artificial limbs and other actuators could be controlled directly by the brain). That's the first application that comes to mind (no pun intended). But there are other applications, such as remote manipulation... controlling a limb over the internet.
In the talk, he also described experiments where they fed electrical activity INTO the brain. This could be thought of as an aditional form of sensory input. The example he gave was a mouse that lives in a cage with a complex magnetic field. The mouse has a magnetic field sensor that sends information into its brain. It will be interesting to see how the mouse's brain incorporates the new input. Presumably, the mouse will have a completely independent perception of where it is in its cage, so that if you gouged out its eyes and plugged up its ears, it would still be able to navigate in its cage without hesitation. Sounds strange, but you've already got something similar. Most people don't realize that they have more than 5 senses; your inner ear houses your equilibrium, your sense of which way is down. Close your eyes, and stand on one foot. You can still stay upright because of your sense of equilibrium. The mouse's magnetic sense is sorta like that.
Using these neuroprosthetic input and output devices, Prof. Nicolelis' goal is to have a quadraplegic walk onto the field to open the
2014 FIFA World Cup in Brazil (Nicolelis' native country). The person will be strapped into an exoskeleton robot, and implants in their brain will give the robot the complex instructions for walking.
Finally, Prof. Nicolelis urged us to loosen our grip on the idea that cognitive functions are localized in the brain. He said that this division of the brain is a model that we like because its easy to understand; he offered the rather paradoxical quip "our brains don't think the way we think". He also emphasized that the brain is a general-purpose statistical inference machine. No matter what input you feed it, it will start to build the structure of the input into its model of the world. I like the way Nicolelis thinks.
I'm at a conference in Salt Lake City, Utah. The conference is called Computational and Systems Neuroscience (COSYNE).
