Brain/Machine Interfaces for the Masses


A while back, I promised to venture boldly into the realm of Brain/Machine interfaces. It’s bold, because it’s a bit outside of my normal expertise. But that never stopped me before. And, as it happens, my wife is a neuroscientist. So this first post will be speculative on my part, drawing on things I’ve read from around the net. Look for a future post or comment to include her corrections and hopefully some deeper thoughts.

The question with Brain/Machine interfaces is not whether they will work — we already have natural biological interfaces from our eyes, ears, skin through the sensorium (those sensory neurons that feed into the brain). We know that sensory input can be mapped from one area of the brain to another, as happens with phantom limb sensations–it just so happens that limb’s sensory signals end up in the brain near those for the face, and the face can easily take over in a pinch. So it’s not inconceivable that the brain could accept a new kind of signal, a properly constructed artificial one, and thereby develop a new pathway to sensation and thought. The question is how much training and development for these new neurological structures is required? How much time and effort?

For example, right now, artificial vision requires implanting electrodes in the visual cortex, where stimulating columns of neurons can cause light or perhaps even colored flashes to be "seen." When properly mapped to compensate for individual differences in brain organization, these can produce something like an image. But implantation of such a device is obviously only for those that absolutely need it. The idea of directly stimulating neurons from outside the brain is still pretty far off.

Or is it? Sony seems to think they’re on their way to simulating sensory stimuli. Of course, if you read to the end of the article, the spokesperson admits it’s blue sky. They don’t know how to do it, which, in my mind, makes this patent an utter fraud (you used to have to at least prototype something to patent it — if all you did was read science fiction, there’s no reason we should grant you a monopoly on the idea). But the point is, someday it may be possible to use sound waves to stimulate other sensory inputs. How? Who knows?

Transcranial Magnetic Stimulation has been shown to be able to induce some sensory hallucinations using strong magnetic fields. So if we could get precise enough control over where and how the machine writes into our brain, we might be able to portray a reasonable virtual world. But that’s a big IF. If we can control magnetic waves passing through random matter so precisely and for millions of points at once, there’s a lot more we could do than just mind control. But it also likely takes more than just EM waves, since I understand that neurons work by both exciting and suppressing other neurons down the line and over-stimulation by radiation can cause a host of problems all its own.

On the other hand, there’s a lot happening on the "output" side of the equation, where brains can directly control computers. This site has a great roundup of recent news. Worthy of note are the number of startups that are getting into the brain/gaming-controller field, with the aim being a console-style controller that can let the user control games. Expect something in 12-18 months, but don’t expect super-fast response times just yet.

Right now, these systems generally work by having you train the computer to recognize arbitrary patterns–e.g., thinking of, say a tiger, that causes a particular gross brainwave pattern the computer can recognize as your intention to go left. That’s cool, but I don’t want to have to think about a tiger every time I want to move. More recent advances allow the computer to more directly record your sensory impulses relating to motion planning and intent, without nearly as much training or thought (it still requires training, but not much conscious thought after that). That’s much more useful for VR, where you may be physically sitting in your easy chair, but virtually playing in the World Cup.

As long as we’re only using our brains to control computers, there’s not too much danger of corruption. The old mind control tin foil hat brigade has worried about the next step since the ideas first became popularized in fiction so many years ago. But their concerns are valid to the extent we wind up with a computer that can write thoughts or other sensory hallucinations into our heads, even if we ask it to.

What happens when such a virtual world is indistinguishable from reality? The "Matrix" is often brought up as an example, but it’s a poor one — dying from disconnecting is just a plot device, as is the idea that given sufficient control of our brains, the AIs couldn’t just lull people into doing whatever they want. The movie "Brainstorm" is much better fodder for discussion of the implications of this technology, still some years out. Whether we can record as well as play back the sensory information, the idea that one could induce schizophrenia or even a psychotic break doesn’t seem nearly as speculative as the technology itself.

But the nearer-term issues aren’t that different from those of video games. What happens when the virtual world is more fulfilling than our actual lives? What happens when we condition ourselves for violent twitch responses? (and don’t get me started on the whole violence in video games issue — video games don’t cause violence, but they do train us for it).. In fact, the issues aren’t that different for projected VEs, HMDs, or brain interfaces, except that with brain interfaces, there’s a whole set of checks and balances that we can bypass, going straight to the cerebral cortex, as they say.

  1. #1 by Ted on May 30, 2006 - 5:45 pm

    Another interesting technique for enabling the blind to see is via stimulating the tongue; see this article. This has also been used to restore a sense of balance to people with inner-ear damage, and there’s work being done to to give sighted people an additional sense, as described here.

  2. #2 by avi on May 30, 2006 - 5:56 pm

    Cool. Thanks, Ted.

(will not be published)