What Do Player Pianos, IBM Punchcards, Telepathy and Elon Muskâs Neuralink Have In Common?
This is a player piano punch sheet or score taken from a rolled up drum. The holes make the keys move. It is a very simple pattern that translates into musical notes.
This is a decoding of âneural spikesâ according to scientists at Elon Muskâs brain computer company Neuralink. The spikes are pictured in single boxes, and many spikes when they fire make a pattern. From that pattern you can begin to decode thought. According to Neuralink, âEverything you hear or think is all action potentials, its action spikes and it feels so real, it feels very real.â
This is an early IBM computer punch card, and the basic pattern recognition of the presence or absence of holes in predefined patterns led to everything we know today about computers. Though IBM introduced the first cards in 1928, they had already been in use to âprogramâ cloth in the Jacquard loom in 1804, making gorgeous silks and tapestries.
I think we are basically back in the same position as we were when IBM first made the computer punch card in 1928 in terms of brain research and decoding thought. Neuralink is putting together elements in brain research that 100 years from now, or even 50 years from now will have tremendous repercussions. The company did not invent all of the aspects of technology it is using to put things together, but they did improve upon existing tech and brought it all together.Â
The Neuralink robot doing an implantation
Neuralink works by extracting electrical signals from neurons. In order to do that it has to implant extremely thin wires in the brain to access and read those signals onto a nanochip. Before it can do that, it has to make a precision surgical robot machine operated by neuroscientists to insert the wires - so it did. Apparently there was some DARPA money thrown into the mix somewhere along the way. The robotic surgeon has to implant tiny wires in between blood vessels and neurons, not on them. This can only be achieved through microscopes and nanometer precision. The implant needle is 24 microns small. Tiny threads are about 1/10 of a human hair, which is about the same size as a neuron. The needle to implant the wire is 24 microns small.  You can open the skull, insert the threads, put in a tiny chip, and then glue the skull shut. The chip functions as a wireless bluetooth signal.
This picture shows thin nano threads pasted onto a fingertip. The thinness of this wire is the big issue DARPA was trying to solve with Moldavian wire from Paradromics, that I previously blogged about in 2017, but Musk beat them to the punch. He wrote about his breakthroughs here.
This is the different new design of the threads called a âlinear edgeâ, which are made of layered polymers. They are so are super duper thin they canât bee seen with a human eye, so they need the special robot that inserts the threads onto the surface of the brain. Another reason is the brainâs surface moves with inhalation and exhalation, and the robot can account for this natural movement. The wires and chip have to record the output from neurons. They are micro fabricated as precisely as the size of an electron beam. It is important to separate the signal to noise ratio in the chips, as they work with nanometer sizes of light. A new design. 350 nanometers,is smaller than visible light.
This image is the crux of how Neuralink acquires the signal. Here you see the neuron sending out an electrical spike and a thread next to it picking up the impulse. The electrical spike is the bulls eye. The copper colored needle is actually the implanted wire thread. You need to be 60 microns away or less to read the signals, so you really need to be under the skull. This is a graphic representation
This is a photo of the real thing, with lots of wires precision implanted into brain tissue. If you look really closely you will see they skirt around the blood vessels and neuron branches, but donât touch any of them. Its sort of like an amazing game of darts, but the goal is to miss the bullâs eye of the vessels.
This is a short history of brain research about making chips to implant in the brain. At this point the Utah array is still the most used implant chip in academia. Neuralinks chips are way, way faster and smaller. Their research builds on a century of neuroscience research and a decade of neuro engineering research. More advanced applications with advanced innovations will follow.
These are early iterations of chips and devices made by Neurlink. Looks like Arduino 101, more or less.
The N1 sensor - the beginning of the sophisticated bean sized implant that goes into the skull and contains the chips.
The actual chips are made on super thin nano wafers in a hermetic environment so there is no dust. They are also super duper thin. A 4 x 4 millimeter chip has a thousand electrodes,and implanting up to 10 chips is feasible. At this point the best FDA approved chip implant for Parkinsons Disease only has 10 electrodes. The Neuralink chips read and write, and are 1000 times more powerful than what is publicly approved. They will get better with newer versions.
Shades of Cyborg Neil Harbisson and his Eyeborg! The Neuralink implant with four chips receive information from the threads, and sends them to an output area for batteries and firmware.Â
This is the size of the output piece, like an earphone.You can upgrade the firmware on the pod on the ear, it is not the actual implant but connects to it. It will be controlled through an iPhone app. Probably Android as well.
Waves of neural spikes from an array from implanted threads that are being read out on a computer. monitor The color screen shows the brain at work, and traces of electrodes from single threads. Each trace is a voltage waveform in time. If you focus on one trace, it shows voltage deflections, or spikes per wave. It occurs when a neuron has an action potential, because that is the core information that is recorded. Then the algorithm is decoded, which means capturing the intended information. You just have to think about something and build up the decoding data from that information or thought, and you can begin to interpret movement, memory, and many other different types of experiences.
This is a basic diagram of signal processing from the chip - how they get the stuff out of the wires and threads to actually show the spikes. Neuralink said âEverything we care about is contained in the statistics of spikes (inside the brain).â So it goes back to the basic IBM punchcards, or the player piano drum roll - decipher the pattern and you decipher the thought.
Here is their basic logic analogue to digital conversion that is necessary to change action potential spikes to computer code..Calling Alvin Lucier, John Cage, Nam June Paik, David Rosenboom, Richard Teitelbaum - or Duh, this has been going on with the brain, changing analog to digital since 1965 in the music world.
Spike rasters in the brain - The top is the brainwave spikes pointed at by the blue arrows, and the bottom is the beginning of pattern recognition of individual neurons. There should be one pixel per electrode. They are on-chip spike detectors. The methods for detection are thresholding signals or directly characterizing the shape. The Neuralink scientists claim they can identify different neurons from same electrode based on their shape. The engineers had to modify the algorithms and scale them to compress neural data up to 200 times. It takes only 900 nano seconds  to compute the signal, faster than the brain knows that the signal even occurred. They can also stimulate any combination of up to 64 channels
One good use of this in the next few years is creating visual feedback for the blind by targeting the visual cortex to create an image better than a dot matrix image - or computer vision basics meets brain wetware. The scientists want to not only read out, but read into the brain. You can read into the brain by passing a current in the electrode. This causes the cell to fire an action potential, like for cochlear implants, or a way for the eye to restore vision. You can also use this technique in the brain to restore the sense of touch or vision. The visual cortex has maps, a spatial map (orange section of the brain in graphic). If you stimulate a point in that area, a blind person sees a point of light or phosphene. The idea is you can stimulate areas of the brain in the visual cortex to resemble a dot matrix level of the world. There are also parts of the brain that control orientation, color, size and speed of moving objects, and once you figure out what they are, and where to stimulate them, you can generate a more comprehensive image that a blind person can experience. Neuralink wants a device with electrodes that are small enough, but with high density that can do better than a dot matrix image.
The first iteration of their implant will have three different types that can go from a mobile device to a mouse or keyboard on a Bluetooth signal. Neuralink needs to get FDA approval. Right now they are working on patients with complete paralysis so it is for serious neurological needs, and idea is to make it really safe.Â
Who needs Google maps when you can tap into someoneâs brain algorithm in their Hippocampus which contains spatial orientation. Here is a rendition of someone who really knows San Francisco, and they can send you pattern signals as you wander through the park and direct you telepathically - voila!
Musk gave a number of reasons for his very public video presentation. The first he shamelessly admitted, was to recruit for new talent to Neuralink. He then framed the motivation of his company as wanting to solve brain ailments, spinal disorders, or catastrophic injuries like a broken neck or spine,. He admitted it wonât happen quickly, and kept mentioning the need for FDA approval. He wants to make his devices as cheap and accessible as a Lasik-like device.
âHopefully,â he said, âAI is a benign scenario. You can chose to have a neural implant if you want, it is not a mandatory thing... We already have digital super intelligence - our access to a computer and a smart phone. The input speed in the human brain is fast, due to our vision, but the output speed is slow because we have to type information into a computer. We are constrained by human bandwidth and mechanics. âÂ
But then he dropped the Neuralink AI bomb -Â that he believes that we âultimately (will) achieve a symbiosis with artificial intelligenceâ (if you want it at a civilization level scale). Then he added, âtwo people with Neuralink could have telepathy, a new kind of communication, conceptual telepathy, it has to be consensualâ.
So the new âme tooâ movement will concern having consensual telepathy #metooconsensualtelepathy.
 *Screen shot photos all taken from publicly available Neuralink YouTube video here.