Brain-Machine Interface

We spend a great deal of time in the pages of Science-Based Medicine taking down every form of pseudoscience in medicine. Of course, what we see as pseudoscience, proponents often see as emerging or cutting edge science. They are taking advantage of the fact that there is a great deal of legitimate emerging science, and they hope they can sneak past the gates by cloaking themselves in the trappings of real science (jargon, studies, their own journals, etc.). Emerging science, however, no matter how plausible and earnest, still has yet to prove itself (by definition), and has to go through the rigorous process of scientific evaluation to slowly gain acceptance. That process – sorting out what works from what doesn’t, the real from the fake – is where all the action is in SBM.

It is refreshing sometimes to talk about an emerging field that, while still experimental, is legitimate and has the potential to usher in a genuinely revolutionary treatment.

I have been following the research into brain-machine interfaces for some years, and reporting on many of the significant “baby steps” in the advance of this new technology. A recent study published in The Lancet represents another incremental and encouraging advance. Researchers at the University of Pittsburgh implanted two strips of 96 electrodes into the motor cortex of a 52 year-old woman with tetraplegia. The electrodes are capable of detecting the firing of neurons in the motor cortex and transmitting those signals to an external processor that in turn controls a fairly sophisticated robotic arm. The arm is described as having seven degrees of freedom – three dimensions of translation, three dimensions of orientation, and one dimension of grasping.

After two days the subject was able to move the robotic arm with her thoughts alone. Over the course of the 13 week study she progressively gained control of the arm and eventually was able to feed herself with the arm. While this is still very far from a “cure” for paralysis or a restoration of full function, for someone who is tetraplegic (all four limbs are paralyzed) having any independent function is a huge improvement in quality of life.

So where are we with this technology?

There are two basic approaches to such brain-machine interfaces – using scalp electrodes or using implanted brain surface electrodes. The advantage of the scalp electrodes is that they are non-invasive. The subject has to just where a cap or helmet of electrodes. There is, however, a significant disadvantage to this approach – the skull and other tissue between the brain and the electrodes significantly impedes the electrical signals from brain activity. This reduces the resolution of the information that can be read from brain activity.
Implanted electrodes have much greater resolution in terms of reading brain activity. You can have a greater number of smaller electrodes, like seeing a picture which far smaller pixels and in focus. The huge disadvantage of implanted electrodes is that they are implanted – this is an invasive procedure, with wires exiting the skull to connect to the external computer. The potential for medical complications, therefore, is far greater than for external electrodes. Research, however, seems to be favoring implanted electrodes as the resolution of scalp surface electrodes is just not sufficient.How much of a limitation this will have on the application of this technology remains to be seen. As you can see from the picture above, the subject has a device connected to the top of her head that connects to the implanted electrodes. This study took place in the lab. They hope to attach the robotic arm to her wheelchair so she can use it outside the lab, but I do wonder what the long term safety is of having implanted electrodes connected to an external device.

There are analogies in other medical technologies. Pacemakers are fully implanted electronic devices that function safely long term. Perhaps implanted electrodes could be  fully internal, and communicate to the external device through blue tooth or a similar wireless technology. Then, however (like a pacemaker) the transmitter will have to be self-powered. Advances are being made on batteries and powering small implanted devices so they don’t have to be recharged, so this is plausible, but not yet available. Patient who have ports or other devices implanted that connect to the outside world are at risk of infection. This can be managed, but it does ultimately limit the life expectancy of such devices. Brain infections, however, are extremely serious, and so any such implant will have to be as safe as possible.

Perhaps a permanent brain implant with an external connector can be devised. The signal processor and any external devices could then be plugged or unplugged from the connector (in a way similar to what was portrayed in the movie, The Matrix). In any case, this is a technical hurdle that needs to be solved for such devices to be safe and effective for long term use outside the laboratory setting.

Some advances are incremental and fairly predictable – increased number of electrodes, more sophisticated computer algorithms for control, and more sophisticated robotic designs. Over the last decade these factors have been steadily improving and can be extrapolated from existing technology. It remains to be seen, however, what the limits of this extrapolation are. How many electrodes can we place on the brain for this purpose?

In summary – researchers have already proven the concept of the brain-machine interface, and the technology has already advanced to the point where functional control of a computer or a robotic device is possible. Now it really is just a matter of further refining the technology.

The next step is the machine-brain interface (and of course closing the loop with a brain-machine-brain interface). What this means is providing feedback from an external machine to the brain. This feedback essentially means artificial sensation, such as having sensors on the hand of a robotic arm that can provide feedback as to the strength of the grip. Researchers have been focusing mainly on the proof of concept of such devices, and so far the results have been very encouraging.

Last year researchers published a study in Nature involving implanted electrodes in a monkey subject. The electrodes were on the motor cortex, and the monkey was able to control a virtual robotic arm. In addition they implanted electrodes on the somatosensory cortex and the monkey was also able to learn to distinguish virtual objects by their feel.

There is further research into brain plasticity and the function of making us feel as if we are inside of, own, and control the parts of our body. These sensations, that we take for granted and may not even be aware of, are specific functions of our brains. The brain uses sensory feedback, which it coordinates with motor intention and movement, to create the sensation that we occupy our bodies. Researchers have been able to reliably trick subjects into feeling as if they occupy a virtual body simply by playing with this sensory feedback. Essentially, if you see a body or limb, and you have a physical sensation that correlates with that body being touched, in many cases that is sufficient for your brain’s processing to conclude that you are that body. Further, if that body or limb moves when and how you intend for it to move your brain will create the sensation that you own and control that body part.

What all of this means is that visual and sensory feedback can be coordinated to make the brain’s inherent wiring create the sensation that a person occupies, owns, and controls either a robotic or even a virtual body or limb.  Therefore it is theoretically possible for a robotic arm controlled by implanted electrodes to feel as if it is a person’s natural arm – that it is part of them.

The existing research is very encouraging, indicating that brain-machine-brain interface technology should work well and create direct mental control of virtual or physical external devices, including robotic limbs or even an entire robotic body or exoskeleton. We do not yet know the limits of this technology, however. Brain plasticity – the ability to adapt to new inputs and functions – exists and is crucial for this technology, but it is not unlimited.

Where we are now, therefore, is right on the brink of real-world applications for patients of brain-machine interface devices. Further, by simple extrapolation of existing technology and principles, such devices should become incrementally more powerful and sophisticated. Early research is also promising for truly revolutionary applications of this technology, combined with machine-brain interface technology, to provide neuroprosthetics that can potentially replace lost or paralyzed limbs, and many other potential applications. This is one medical technology that may seem like science fiction, but is rapidly entering the realm of reality.

Posted in: Neuroscience/Mental Health

Leave a Comment (25) ↓

25 thoughts on “Brain-Machine Interface

  1. cervantes says:

    Of course one difficulty with cutting edge technology is that we are going down multiple paths at once, often making very large investments in more than one, and we might get somewhere else first that makes this entire endeavor largely moot. If we are able to reconnect severed spinal cords, or regrow lost limbs, not much use for this technology after all. Indeed, it can never be a good substitute for the real thing.

    Another problem: Is this really the best use of our (now more and more limited) biomedical research dollars? It’s trivially obvious that we could get vastly more benefit from the same investment in preventing diseases that affect literally thousands of times more people. Just something to think about.

  2. The Nature article about virtual sensations in monkey subjects was co-authored by Miguel Nicoleilis, who has written an excellent book about his work with brain-machine interfaces (BMIs): “Beyond boundaries : the new neuroscience of connecting brains with machines—and how it will change our lives.” If you’re interested in this subject, I highly recommend it.

    The only quibble I have with the book is that he got carried away when describing potential future applications of the technology. At a couple of points in the book he speculates that we might one day explore the surface of other planets using BMIs. Unless he’s talking about doing it from orbit around the planets, the speed of light makes that vision impossible.

    In order for the brain to stay in synch with the device you’re controlling, and to have that sense of ownership, the control signals and the feedback (video or direct stimulation of the brain) needs to happen within the same time frame that your nerves would require to pass signals to and from parts of your body — a few hundred milliseconds.

    A signal takes something like 2.4 seconds to make the round trip to the moon. That’s an order of magnitude beyond the maximum time that a BMI link could tolerate, and the moon is the closest place we could go. The lag time to Mars would be measured in minutes. If you’ve ever tried to type when your word processor is lagging and the letters aren’t appearing immediately, you’ll recognize the impossibility of trying to control a robot with a BMI if there’s any significant lag in the communications.

    Other than that minor point, Nicoleilis’ book is excellent.

  3. WilliamLawrenceUtridge says:

    Two points:

    - Can you use magnetic induction to charge a battery through the loops of wire implanted just under the skin?

    Cervantes, your failure to recognize the importance of this advance vis-a-vis giant mecha is disappointing.*

    *Your point via the investment of scarce research dollars and most “bang for buck” is distressingly (from a giant mecha perspective) perspicacious.

  4. Scott says:

    If we are able to reconnect severed spinal cords, or regrow lost limbs, not much use for this technology after all. Indeed, it can never be a good substitute for the real thing.

    Not at all. There’s no particular reason the technology has to be limited to replacing lost function – it’s just that it’s currently too invasive/risky to seriously consider using it to ADD more functionality.

    Doc Ock style extra limbs, potentially stronger or otherwise more capable than those we presently have, are an obvious minor extension to the technology. People are already talking about using it to provide a better computer interface than mouse/keyboard/touchscreen. And I have no doubt other applications will be devised.

    If a means can be devised to provide appropriate-resolution electrodes without major surgery, the possibilities are almost limitless. That’s a major hurdle with no obvious route to an answer today, but I decline to label it impossible.

  5. rork says:

    I am tired of this keyboard and screen, and have been waiting a long time for direct-to-brain.

  6. mousethatroared says:

    Cool! Thanks for the update SN – I enjoy your critical approach to the topic. It’s easy to get caught up in the excitement of new technology and forget that looking at concerns like possibility of infection and power sources are important ways to improve those innovations and make them a real possibility.

    I do wonder about an implantable post that can accommodate a clip on device similar to the BAHA (Bone Anchor Hearing Aid)..But the BAHA post is implanted into the skull to receive vibrations – it’s not implanted into the brain. Obviously I’m in over my head (sorry).

    I heard an interview with Miguel Nicoleilis on the Diane Rehms show awhile back. Very interesting and worth checking out if you want to hear more.

  7. gears says:

    Man, I was going to suggest wirelessly charged batteries, but you beat me to it WLU.

  8. WilliamLawrenceUtridge says:

    But unlike me, you sound like you could build one :)

    Suggestions are easy, execution not so much. I wonder if it could be done without heating the tissue an undue amount?

  9. Terminus says:

    Hey cool, my area of research on SBM!

    Our research group works with surface electrodes, but we of course follow all the implanted electrode work. They are definitely the future, being able to detect the activity of individual neurons rather than millions or billions makes a huge difference. However they are still a way away from permanent, everyday use implants, even for those who really need them like tetraplegics. Let alone amputees, or even normal folk who don’t like moving to change the channel! When they leave the laboratory, I imagine they will be wireless to reduce the risk of infection. I see surface electrode systems being used sooner, and then slowly being replaced by implanted systems, for less and less serious disorders, until we eventually buy implants like a new mobile phone.

    I am particularly fascinated by this technologies ability to treat a wide spectrum of disorders in the same way. We are often capturing movement at the ‘intent to move’ stage, so if anything in between this brain system and actual movement of the body is impaired – we can just bypass it all with the same device and restore some quality of life. This means the same treatment can be used for paralysis, amputation, muscular dystrophy, stroke, tremors, muscle weakness, cerebral palsy etc…

    This ‘intent to move’ even creates some interesting ethical dilemmas. Since suppression of a movement can take place at a later stage, if a patient pulls a trigger – did they actually mean to do it? Or did they only imagine it? Or would normally have stopped themselves?

    We also have the potential to expose useful brain signals that are normally contained entirely within the brain. For example there is a distinct signal when we observe an error – and it is different if we make the error, or somebody else makes the error. Preparing safety systems like brakes for your car? Or improving the accuracy of speech to text perhaps?

  10. gears says:


    Naw dude, that was just some serious technobabble, and it looks like you fell for it ;)

    But seriously. Like you said, the implementation is the hard part. Hence the dearth of giant mechas.

  11. elburto says:

    Hmm. Perhaps my robotic exoskeleton could be a reality some day?

    As someone with impaired proprioception, and multifocal neuropathy that’s rendered my sensory nerves virtually useless, this field is incredibly exciting.

    I am almost as intelligent as a monkey, so who knows? Perhaps the future has a glimmer of. hope.

  12. mousethatroared says:

    “These sensations, that we take for granted and may not even be aware of, are specific functions of our brains. The brain uses sensory feedback, which it coordinates with motor intention and movement, to create the sensation that we occupy our bodies.”

    Maybe this is a weird leap, but – if we use sensory feedback to train the brain to recognize an object as part of the body, can we use sensory feedback to train to brain to stop recognizing a particular sensation?

    I’m thinking of things like phantom limb and paresthesia.

  13. WilliamLawrenceUtridge says:

    It’s nice to see that people are starting to focus on the important parts of this topic – giant mecha.

    Do I wish were born ten years later, so I stood a better chance of benefiting from this kind of advance? Or do I appreciate when I was actually born because I saw what the world was like before the massive leaps forward in computing?

    It’s nice to live in exciting technological (wow, check out this touch screen phone!) rather than Chinese proverb (wow, check out how many of my fellow farmers died because of civil war) times.

  14. mousethatroared says:

    WLU – I’m pretty sure I’d be excited about gaint mecha if I knew what it was…

    no, no, don’t tell me. I prefer to enjoy the mystery.

  15. bhami says:

    Key application areas for these technologies will be industrial control — situations such as manipulating robots in Chernobyl or Dai Ichi Fukushima or in a space walk, where direct human presence is dangerous, due to radiation or other hazards, but where the human controller can be within a few hundred yards or a few miles, so that light speed delays are not an issue.

  16. Purenoiz says:


    Can I come work with you in about 6-12 years? We really need to get the Giant Mechs prepared. I’m just not sure if WLU is thinking Robotech style mechs or MechWarrior style.
    Either way, where did I park my 2000 ton mech?

  17. Lytrigian says:

    I’m sure these aren’t as “high resolution” as the manufacturers claim, and since they appear to attach over hair as well as skin they’re probably even less sensitive than more carefully placed scalp electrodes, but I thought I’d drop this here anyway as possibly relevant:

    There do not as yet seem to be any applications available for giant mecha control.

  18. agitato says:


    I didn’t know what Giant Mecha was either but I do now. (google images: giant mecha)

    Good grief. I then went straight to…google images: cute kittens

  19. WilliamLawrenceUtridge says:


    Is there really any bad type of giant mecha? Though I personally lean towards Glitterboy – it stretches the definition of both “giant” and “mecha”, but you do get a boom gun and you don’t have to get out to poop or eat.

  20. elburto says:

    WLU – I’m with you. For me, being a BOOM!bot with an electronic body, inbuilt weaponry, and maybe some cool lighting effects, is win/win.

    Don’t get me wrong, my powerchair is great, but it lacks a certain oomph, or. visual impact.

    mouse -

    can we use sensory feedback to train to brain to stop recognizing a particular sensation?

    I’m thinking of things like phantom limb and paresthesia.

    I’ve used ad hoc CBT on myself to help with paraesthesia, and some new, but surprisingly simple methods, have been aiding amputees with PLS:

    The answer is mirrors+CBT. I have an acquaintance born with a missing foot who was plagued by PLS, and they’ve undergone a trial of mirror training, and experienced reduced pain and negative feedback.

    As it seems like the ‘map’ to the body is hard-coded, maybe the sensory training mentioned above could, as you suggested, be developed to assist with extreme paraesthesia and PLS.

  21. Kandle says:

    There is already a bionic implant which goes inside the skull (cochlear implant). I imagine that a neural mechanical interface could use similar technologies for power and communications.

  22. 2Healthy says:

    Certainly the “open circuit brain” isn’t a good idea, very prone to very dangerous infections sooner or later.
    I agree, radio,bluetooh,wifi or some wireless communication system should be used. A battery installed inside may be charged by inducction. Moreover they could be charged by movement, like some watches.
    The power for so close range antenna woudn’t be to high, 10 mW?, moreless like a LED.
    Besides I bet that resolution from surface readings may seem “too wide” but with proper deconvolution and personalized machine learning you could probably get the same resolution as in intracraneal arrays.
    Also if instead of a single simple arm controler they used something more computer-like with a cam they could get the arm to do much of the hard work by itself. Mind it a motor movement is more exact than ours.

    For funding, research and peer finding please refer to the non-profit Aging Portfolio.

  23. Narad says:

    A battery installed inside may be charged by inducction. Moreover they could be charged by movement, like some watches.

    This is getting silly. Putting an inductive coil in someone’s head presents certain practical obstacles, if you will. Self-winding piezoelectric clockworks are just apropos of nothing.

  24. Pingback: get bigger boobs
  25. HopDavid says:

    I’m wondering why they put implants on the surface of the brain. If the implants communicate with a robotic arm, why wouldn’t they be placed on afferent and efferent nerves going to the arm?

    Happy to see a skeptic enthused by this technology. My son-in-law was recently paralyzed. There are a lot of very optimistic claims, often being made by scammers preying on people who are already suffering terribly. A lot of hype and false promises going on. But also some legitimate reasons to be hopeful.

    Buying the Miguel Nicoleilis book as per Hilary Mark Nelson’s suggestion.

Comments are closed.