I, Robot: The Past, Present, and Future of Patient-Controlled Prosthetics


This week’s question comes from my dad, who asks, “I watched a 60 Minutes segment once on prosthetics people could control with their thoughts. How do those work exactly?” The 60 Minutes segment he is referring to, to catch everyone up, can be found here. As a short run down for the lazy (aka just about everyone), it shows two different types of patient-controllable prosthetic limbs. The first involves a patient with a condition that causes total loss of muscle control, making her effectively paralyzed save for her ability to speak, breathe, move her head, and talk. They implanted electrodes directly into her brain that were able to control an external arm. With some practice, she was able to do complex movements like feed herself and give a handshake with the device solely by virtue of her thoughts. The second involved a man who lost his lower arm and hand. His prosthetic was directly connected to his stump and was able to interpret the existing nerve signals traveling through the upper arm into a usable artificial limb.

Some background on the recent evolution of prosthetic limbs: As we (hopefully) all know, the wars in Iraq and Afghanistan have created thousands of wounded veterans, largely due to the use of improvised explosive devices (IEDs). Many have lost one or more limbs due to these combat injuries. As time passed, it became strikingly clear that the existing prosthetic limb technology could not fully rise to the occasion of rehabilitating these wounded warriors, especially those who were otherwise healthy save for a missing limb. Technology progressed significantly due to this demand, even creating prosthetics durable and functional enough to allow some to return to full active duty. Yet, these prosthetics are still dumb, disconnected from the conscious will of the user they were created to help. However, as the 60 Minutes segment displayed, we have reached the point where prosthetics no longer have to be dumb. While still very experimental, we have conclusively shown we can bridge the gap between the nervous system and external electronics. Now, let’s explore a little of how the two types work.

The first one shown in the segment involves direct brain control of an external machine. Use of the machine is made possible by the adaptive nature of the brain; so long as the part of the brain responsible for generating movement signals remains intact, even otherwise paralyzed people can theoretically utilize this technology as if their appendage still worked. The insertion of the electrodes directly into the brain bypasses the otherwise useless nervous system that does not transmit these signals to the rest of the body.

The area of the brain involved in this process is known as the primary motor cortex. The primary motor cortex is the part of the brain involved in creating the impulses that eventually lead to voluntary muscle movement. The cortex itself is arranged in a predictable manner that effectively replicates the anatomy of the body (though distorted), which is known as a homunculus. Below I provided a picture of where the primary motor cortex is and how the homunculus is organized across it. From there, electrodes are implanted on or within the cortex itself within the relevant area (ex. to control an arm, electrodes would be placed in the part of the cortex involved directly in arm movement). Schwartz, et al., describe this process in much more detail in their 2006 article “Brain-Controlled Interfaces: Movement Restoration with Neural Prosthetics.” The differences between the implants are not particularly important for our purposes (though it is an interesting read). What is important to understand is that these implants are able to read the electrical signals put out by relevant brain cells and, through a computer, translate them to produce the intended movement in an external robot.


Now, you may be wondering how the brain could possibly do this if it is cut off from the sensations it would otherwise receive from its original limb. In other words, in the absence of feedback, how does the brain know what to do? The answer is, the brain is incredibly adaptable and is capable of utilizing other senses, primarily vision, to try to bridge the gap. But, as one might imagine and the video shows, this process is far from perfect. Even with tremendous practice and constant coaching, the brain-machine interface can still be very clumsy. But why is this the case if we’re directly tapping into the area of the brain responsible for movement?

Refer back to the photo of the primary motor cortex for a moment and notice what lies right beside it. Yes, the somatosensory cortex, which is the area where your brain processes the physical sensations it receives back from nerves throughout the body. We may be reading the signals the brain is generating for movement, but we are currently giving it nothing back about the results of that movement. Thus, our movements are restricted to a certain extent because we’re operating a numb arm. If you’ve ever tried to move anything numb, you probably quickly realized you can’t accurately discern how hard to grip something, the existence of small objects, or how to execute fine actions. The brain needs feedback in order to best instruct movement. Given that we can currently read and translate motor cortex signals, it is only logical that the next step is to complete the circuit and return sensation signals from the prosthetic back to the sensory cortex. If we can accomplish this, the capabilities of the external prosthetics should dramatically improve.

But, what about people who are otherwise fine but are just missing part of a limb? Should they be forced to endure a serious brain procedure in order to reap the benefits? The answer, as I mentioned earlier, is no. These sorts of prosthetics, which have much wider applicability than the brain implants, utilize a totally different technology to achieve a similar goal.

When a limb is amputated, the nerves serving that portion of the limb still exist above the amputation, which means that the signals originating from the brain and going back to the brain still make a full transit. The signal just doesn’t travel past where the limb no longer exists. But, this does not mean that the signal originating from the brain does not contain information capable of moving said non-existent limb (the homunculus in the motor and sensory cortices don’t normally rearrange in response to an amputation). Should something else be there to capture and utilize the information, it could be moved just the same as if the original limb remained.

Unlike with the brain implants, these types of prosthetics are non-invasive. Instead, they read the nerve impulses via surface electromyography. Electromyography reads the electrical impulses generated by individual nerves by examining the firing of reference muscles. Surface refers to reading these signals through the skin, rather than directly in muscle tissue using needles. These signals are then interpreted by computers attached to the prosthetic device and translated into the appropriate movement. This method is extremely versatile and has been applied widely across different types of limbs with promising results.

Likewise with the brain implants, sensory feedback is proving to be an essential component of improving the functionality of these devices. Current systems use vibration against the existing skin proportional to the amount of force being applied by the prosthetic. While this has helped, there has, as of yet, not been a significant breakthrough in transmitting accurate sensory data through the skin in a non-invasive way.

In summary, either type of prosthetic achieves its goal by reading nerve impulses and translating them into an external action. This area of biotechnology remains an exciting and quickly advancing field. It is well within our grasp to soon have the ability to effectively return amputees to their every day lives with minimal limitation.

Hope this helps, dad! Thank you for the question. As always, anyone can submit a question either directly through this site, directly to me, or via Facebook.

Thanks to everyone for reading and keeping up with this. We’re closing in on 1000 views! Crazy.

Till next time, that’s all she wrote.



Cipriani, et al. “On the Shared Control of an EMG-Controlled Prosthetic Hand: Analysis of User–Prosthesis Interaction.” IEEE Transactions on Robotics. February 2008.

Georgopoulos, et al. “Neuronal population coding of movement direction.” Science. September 1986.

Schwartz, et al. “Brain-Controlled Interfaces: Movement Restoration with Neural Prosthetics.: Neuron. October 2006.


  1. justinandrew88

    Cody posted via Facebook: “Interesting. I was just researching the same thing as well as why severed nerves are so difficult to reattach in combination with how to innervate prostheses with nerve bundles.”

  2. justinandrew88

    My response to Cody via Facebook: “It would appear that the ideal would be to attach severed nerve endings directly to the prothesis so that it could have direct access to both the motor and sensory signals. However, I think we’re a ways away from that, especially since the transcutaneous approaches still have significant gains to make. As I alluded to in the article, sensory activity is key to making these things truly work the way we want them to. I think that is the next step in the process, but the best method for providing that kind of feedback isn’t clear. Curious to see how far we go with this and what ends up bearing the most fruit.”

  3. justinandrew88

    Matt B. posted via Facebook: “Cool man! The motor cortex and somatosensory cortex are definitely the biggest players in voluntary movement. It’s why an infarct to that area will cause paresis. However, there are also many other parts of the brain that go in to movement. When initiating movement, the basal ganglia serves to modulate inputs so they arent too strong or too weak (dysfunction here can cause Parkinson’s or Huntington’s). After movement, sensation comes back up through the spinal cord- some fibers go to the cerebellum for proprioception (position sense) to check if the movement was balanced/what was desired (i.e. you have to stop yourself from going to far and knocking over a glass when reaching for it). The cerebellum actually prunes those pathways in response, thus allowing for what is effectively “muscle memory”. Other fibers allow pain, temperature, vibration and position sense to go to the thalamus, and then the somatosensory cortex. It isn’t enough to just relay electrical impulses from the cortex because those signals are unfiltered. For full control, one would need to combine the modulatory effects of the basal ganglia, the thalamus, and the cerebellum. And don’t forget reflexes which are controlled by the spinal cord at the level of the primary sensory output from the limb. I think the biggest problem facing these future prosthetics would be the position sense. The prosthetic would become much harder to use if you couldn’t see it- even if you could feel temperature, “pain” or vibration. For movement it’s not just what IS there, it’s also what ISNT there.”

  4. justinandrew88

    My response to Matt B. via Facebook: “Most definitely man, clearly bypassing the brain’s movement and coordination systems is not ideal. But I look at it from a different direction. Given that areas like the basal ganglia, thalamus, and cerebellum are pretty difficult to mess around with (I’d love to meet the neurosurgeon cavalier enough to snake a probe wire from the thalamus to the skull haha), I think there’s an alternative. You mention that connecting to the somatosensory cortex would provide unfiltered signals. But why must this be true? Could we not create external computers responsible for modulating movement? While obviously not perfect, I think externalizing this activity could be effective enough to provide the type of modulation that allows for more normalized function. Though it wouldn’t be a true replacement (especially absent positional sense and the spinal reflexes), it would be a great advance. Of course, this technology is in its early infancy. It is entirely possible we will figure out how to generate a sensory signal that is directed through the spinal cord rather than brain that can utilize all of these endogenous systems. But, for now, we will have to settle for a totally paralyzed woman simply feeding herself with a robotic arm connected to her brain by wires :).”

    • justinandrew88

      Matt B.’s response via Facebook: “Yeah you’d ultimately need the prosthetic to modulate for you. It certainly seems possible. If you also linked undesired responses to punishment or reward then it might make up for some of the cerebellum’s functions. The brain is so magnificently complicated.”

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s