Direct neural interface: Hopefully coming soon to a brain near you

27 October 2015

Direct neural interface has long been a dream and fantasy of tech geeks like myself who grew up reading science fiction. Slap an electrode net on your head (or screw a cable into an implanted jack) and there you are, controlling a computer with the same ease that you'd walk down the street or bend a paperclip with your fingers. If nothing else, those of us who battle the spectre of carpal tunnel syndrome constantly know that our careers have a shelf life, and at some point we're going to be out of action more or less permanently. So we are constantly on the lookout for ways to not wind up on permanent disability because we can't work anymore.

Or maybe you just found out way more about me than you really needed to know. Let's move along, shall we?

Bits and pieces of brain/computer interface technology have been around for years: The electroencephalogram is a non-invasive sensing technology for picking up the electrical activity of the brain, and relatively inexpensive open source eegs like the OpenBCI exist for people hacking around. Microprocessors are now fast and powerful enough to crunch EEG data in realtime for very little money and the most unusual hardware can be repurposed for getting the data into your laptop. You can purchase reusable EEG electrodes on Amazon for very little money to ensure that you get the highest quality signals (or you can make your own). TMS (transcranial magnetic stimulation) has been around since the mid-1990's, when only a dedicated subculture of body hackers and modification enthusiasts were winding their own electromagnets and seeing what would happen when they were placed on different areas of their skulls. But what would it take to put all of this together to transfer information from one person's brain to that of another person?

The answer is: Not much, really. A research team at the University of Washington have published the results of experments they've conducted that accomplished just that. One test subject was wired up to an EEG monitoring their cortical electrical activity; the EEG was interfaced with a computer plugged into their local area network where it transmitted the data to another computer. In another lab about a mile down the road, a second test subject with a TMS unit strapped to the back of their head that was interfaced with a second computer receiving the EEG data from the network. The TMS was positioned over the primary visual cortex. When the TMS was energized the resulting magnetic field caused phosphenes to appear in the subject's field of vision (if you want to replicate this at home, close your eyes and gently press on your eyelids; what you see are phosphenes triggered by the pressure stimulating your retinas, which send signals down your optic nerves into your visual cortices). The first test subject viewed a static image; the second test subject used some software to ask the first yes-or-no questions about the image, which was answered by thinking "Yes" or "No" very hard. The second test subject detecting a strong phosphene interpreted it as a "Yes" response. When the experiments were done, the numbers were crunched and it was found that five pairs of test subjects playing twenty games, where half were controls and half were real games showed a success rate of 72%. In a separate control group, their success rate was only 18%, which is significantly below that of the experimental group. If you've a mind to, their peer-reviewed paper is available at PLOS ONE. There are other forms of DNI research taking place right now. Possibly one of the busiest ones is mitigation for amyotrophic lateral sclerosis, better known as Lou Gehrig's Disease, is a disease in which motor neurons in the brain and leading from the spinal cord to the rest of the body progressively fail and die, resulting in near complete paralysis. In recent years a technology called Braingate was developed, which allows people who are paralyzed to interface with computers directly and carry out relatively simple tasks. Braingate technology continues to advance with the development of Braingate2, which is more precise and thus more useful. Braingate2 is surgically installed the user's brain into the part of the brain which controls the user's hands. Software samples and interprets activity from the brain and moves a cursor on the screen, in much the same way that moving a mouse or using a touchpad would do the same thing. The control software recently underwent a hefty rewrite to optimize it; turnaround time from thought to action went from several seconds to about 20 milliseconds, which is a much more reasonable delay. Better signal processing to filter out noise was implemented because the environment is saturated with electromagnetic interference, from the 60 Hz hum from power lines in the walls to overhead lighting. These fixes lead to the test subjects being able to type about six words per minute, which is a significant improvement over Braingate v1 and the older control software. The upshot of this is that ALS patients who are largely unable to interact with the outside world due to the progression of the disease are now able to communicate via text while connected to the interpreting computer. Early stage technology or not you have to admit that this is a significant breakthrough.

A big problem with DNI is that very electrodes often have to be installed directly into the brain. Generally speaking, organic bodies aren't very sympatico with metal bits being stuck in them for very long, highly delicate neurons most of all. The cell membranes tend to become thicker to push the electrodes away, which breaks the connections until you can't get a clear signal. It's also very tricky to prevent the stoma from becoming infected; unfortunately, we still don't have the organic/inorganic interface seals down, either. So, research into making electrodes which are made of materials that organic bodies don't interpret as hostile continues. At Lund University a research team has figured out how to grow semiconducting gallium phosphide wires on the nanoscale that human neurons interpret as bio-neutral, which seems to encourage them to grow toward and around them. Neurons treat them as native attachment points and glial cells grow toward and around them as support structures. In a slightly more unusual variation, a research team at UCal Berkeley published a paper on Arxiv where they postulate using something called neural dust, particles of piezoelectric material about 100 micrometers in diameter (a little thinner than a human hair, or half the size of your average dust mite). The idea is that the neural dust could pick up the electrical activity of individual neurons and emit ultrasound which freely passes through the skull and skin, where it could then be picked up and decoded; precisely modulated ultrasound could then be used to transmit information back into the brain. Unfortunately, there appear to be enough mechanical engineering problems with this technique that it may never take off, such as heat dissipation inside the skull. Another, more feasible possibility might be using syringe injectable circuitry designed on a similar scale to neural dust which spontaneously unfolds to lay itself over the cerebral cortex which neurons will accept as their own and grow around and through. This so-called neural lace circuitry is already being fabricated and tested in laboratory mice with a 90% success rate. After the neural lace is injected into the skulls of the mice, the interface between circuitry and gray matter occurred spontaneously without harm to the mice. Additionally, the mice are connected to computers which are monitoring the electrical activity of their brains on an on-going basis... maybe we're finally getting somewhere.