Paralysed Let Their Thoughts Wander...

The Naked Scientists spoke to Dr John Donoghue, Brown University
16 July 2006

Interview with 

Dr John Donoghue, Brown University


Chris - Now take a listen to this… that was actually the sound of brain cells, neurons, firing off and talking to each other. It's been done by a man called John Donoghue at Brown University, and he's found a way of making an electrode that can be inserted into the brain of an individual. In this instance they've put this electrode into the brain of a man called Matthew who's been paralysed for the past few years because of a spinal cord injury. What the researchers are trying to do is to record that activity you just heard and then use a computer to decode it so that Matthew can re-establish the ability to move. So using a computer he can control things in the environment and maybe even, in the future, get the computer to do the job of the muscles that have been disconnected from his brain by the paralysis. This is John Donoghue talking about how this research works.

John - The paper is about the technology that we've developed to help a paralysed person communicate with the outside world again, to be able to use their thoughts and for their brain activity to control devices, and mainly what we used was a computer.

Chris - So is this literally recording from individual nerve fibres or is this whole populations of nerve cells that you're listening to?

John - We're recording individual nerve cells, we're recording what's called the spiking activity, which is the language of the brain, and we record many of those at once - dozens to more than 100 - and we transform the pattern of spiking activity into a single control signal.

Chris - So in order to record from the hand area of this person's brain you actually, presumably, have to implant an electrode?

John - An electrode array is implanted. Yes.

Chris - So what does that look like and how does it work?

John - The array is about the size of a baby aspirin and it's implanted on the surface. It's 4 x 4 millimetres and it has 100 hair-like protrusions coming out of it. They go just into the surface of the brain, the cortex is about the thickness of an orange peel and the electrodes go about half that thickness into the brain to pick up the neurons that are just below the surface.

Chris - And then how is that translated by your computer into a meaningful movement? And does the subject have to learn to control this?

John - There's not actually a learning required to control it, the part that has to be learned is the relationship between a complex pattern of activity that's coming out of the brain and a desired motion of the cursor. So we have the patient imagine that they're tracking a cursor on a screen, and by the changes in brain activity that we observe while that patient is observing the cursor motion we build a map that says, here is the pattern of activity and here's the cursor position, and later on, here is another pattern of activity and here's another cursor position, and we try to map the two together. And of course another significant finding is we're able to do that. We do it over a period of about 15 minutes or so. So that's not really learning, it's establishing the mapping function. Once that mapping function is put together in the computer then the patient is able to just think about moving and the cursor will move pretty much in the motion that the hand would take if you were to imagine, say, moving left or right.

Chris - How accurate are the movements that the person who has the implant fitted can make?

John - So the motion of the cursor by thought is I would describe as wobbly and unstable, so right now, if your mouse performed like that it would be rather frustrating. And that's actually been part of our study, is to understand what happens as a person is exposed to that kind of signal? Do they get better and better at controlling it? And so far we haven't seen a major change in their control. And what that means is that at least we haven't found out how to exploit the brain's plasticity, so we need to change the computer to make the control signal better. And we're doing that and actually having some good success.


Add a comment