Questions About Brain Computer Interfaces

[RobotTemplar]

I have been really curious about BCIs for a while now, and I have read a lot about them. One of the coolest technologies I could see coming from this kind of research is a spinal cord interface, ala the Matrix, except that the major application would probably be for people with spinal cord injuries (or disembodied brains...mwahahaa). I have been thinking how such an interface could be implemented (and whether it will even be feasible at all in the near future, without radical bio- and genetic engineering advancements) but I find that I am hindered by a degree of ignorance about the physiology of the nervous system that cannot easily be rectified by surfing the internet or reading Wikipedia. It seems like asking a knowledgeable person (what I am doing on this forum) might be a lot faster than searching through textbooks.

 

So, my questions.

How does the brain distinguish between an input from, say, your right index finger versus your right middle finger? Is this like vision, where you have individual cells at some level that correspond with parts of the visual field? I read that axons from cells in your motor cortex actually extend down to the base of your spinal cord...so do cells from the sensory part of the NS (like the part that senses touch to your index finger) similarly have axons that extend upward through the spinal cord, perhaps meeting the motor cortex outgrowths at the top? I am very unclear about how this works. Would a cross section of the spinal cord then have axons from from every part of your "tactile field" below that section of the spinal cord? I have read things that indicate that some complex signaling mechanism and neurotransmitters may play a role. Please help ease my confusion.

 

Another question, much simpler, if you can answer it: would a spinal interface consisting solely of small enough, biocompatible electrodes be enough to transmit/receive signals (I assume not)? Or would it be necessary to have some kind of mechanism that could feed out neurotransmitters and and accept them as inputs?

Edited April 19, 2011 by RobotTemplar

[SMF]

There are topographic maps in the cortex that correspond to touch in somatosensory cortex, and motor units (a specific group of muscle cells) within various muscles in motor cortex (look up motor and somatosensory homunculus). The position within these maps is a primary part of the code that distinguishes different body regions. For somatosensory cortex, the highest resolution for discrimination between two points is called the two point threshold which can be pretty fine in areas like the tips of the fingers, lips, and tongue.

 

The spinal cord carries information for the body, but not the head, and because motor axons from motor cortex peel off at their approximate appropriate level and sensory information enter the cord at approximately adjacent levels (look up dermatome map), your supposition regarding what is above and below any cord level is correct and this is why serious spinal cord damage causes loss of sensation and motor control below the damage. The motor and sensory pathways in the spinal cord are separate and take very different routes and, except for spinal reflexes, interact only very indirectly.

 

Also, ascending and descending axons in the spinal cord are laid down in an orderly topographical manner. For example, a small cut on the surface of the somatic touch pathway at a specific level will predict a specific region of sensory loss, while small penetrating wound at the same level will result in a different region of loss.

 

It is possible to put long term microelectrodes in nervous tissue (e.g. look up cochlear implants) that can record electrical activity from, or stimulate, small regions. I know that there are problems of formation of scar-like tissue that can cause problems, but scientists are working on it. There has been quite a bit of experimentation with a brain interface via macro electrodes that can just be fastened to the scalp, but this reduces the number of electrodes because the areas of brain they can record from are so large. SM

Edited April 19, 2011 by SMF

[RobotTemplar]

There are topographic maps in the cortex that correspond to touch in somatosensory cortex, and motor units (a specific group of muscle cells) within various muscles in motor cortex (look up motor and somatosensory homunculus). The position within these maps is a primary part of the code that distinguishes different body regions. For somatosensory cortex, the highest resolution for discrimination between two points is called the two point threshold which can be pretty fine in areas like the tips of the fingers, lips, and tongue.

Could you (or any knowledgeable person) perhaps expand on the statement I have in bold? Does this mean that it is not the case that there are cortical cells at some level that have a one to one correspondence with sensory neurons?

 

The spinal cord carries information for the body, but not the head, and because motor axons from motor cortex peel off at their approximate appropriate level and sensory information enter the cord at approximately adjacent levels (look up dermatome map), your supposition regarding what is above and below any cord level is correct and this is why serious spinal cord damage causes loss of sensation and motor control below the damage. The motor and sensory pathways in the spinal cord are separate and take very different routes and, except for spinal reflexes, interact only very indirectly.

Thank you for this edifying segment...

 

Also, ascending and descending axons in the spinal cord are laid down in an orderly topographical manner. For example, a small cut on the surface of the somatic touch pathway at a specific level will predict a specific region of sensory loss, while small penetrating wound at the same level will result in a different region of loss.

Aha, this is exactly what I wanted to know.

 

So, does the spinal cord consist primarily (or entirely) of nerve axons, then? Really really long nerve axons?

 

 

It is possible to put long term microelectrodes in nervous tissue (e.g. look up cochlear implants) that can record electrical activity from, or stimulate, small regions. I know that there are problems of formation of scar-like tissue that can cause problems, but scientists are working on it. There has been quite a bit of experimentation with a brain interface via macro electrodes that can just be fastened to the scalp, but this reduces the number of electrodes because the areas of brain they can record from are so large. SM

That's cool, I never made the cochlear implant connection. From reading parts of the Wikipedia article, my guess would be because nerve activation is achieved using an electromagnetic field, and the implant never actually comes in contact with the nerve. The BCIs I have read about (not including stuff like artificial retinas) have always been for reading from the brain, not to it.

 

 

Also...

I have a question about spinal nerves. Does a spinal nerve as it nears the spinal cord also, like the spinal cord, contain individual nerves that correspond with tactile inputs at that level (so in a cross section you might be able to identify, "oh, this nerve is for this point on the forefinger, and this one is for...")? Finally, just to make sure that I'm getting this right, the nerves coming out of the spinal cord in the image below are in fact the spinal nerves, correct?

 

http://upload.wikime...tem_diagram.png

[SMF]

Re the topographic map on the cortex: If you were to electrically stimulate a small region anywhere on the body surface, there would be a corresponding set of cortical neurons within the map (homunculus) of the somatosensory cortex. You could record this activity in just the right spot. It is not a one to one connection, but it is pretty close. If you stimulated this cortical spot in a person, they would report touch sensation in the corresponding spot on the body. Note that the sensory pathways pass through the thalamus for processing prior to reaching the sensory cortex.

 

If you were to electrically stimulate just a few cortical neurons in the primary motor cortex, mapped for a finger, there would be a corresponding twitch in a few motor units within one of the muscles in the forearm that is involved in controlling the specific finger, and there would be a corresponding very small movement of the finger. Location in the brain corresponds to location on the body. Is this what you wanted to know?

 

Re the spinal cord constituents: The cord carries a variety of long (and some shorter) nerve tracts consisting of bundles of axons, but at each segment that is associated with a specific dermatome there will be neuron cell bodies. Sensory neuron cells are in or near the dorsal horn, and motor neurons are in the ventral horn. These neurons project axons out to the periphery in the peripheral nerves, while others send (sensory) or receive (motor) activity to/from the brain via long tracts of axons in the cord. The spinal cord actually does some neural processing, so there are quite a few interneuron circuits in the dorsal and ventral horns.

 

Re current noninvasive microstimulation and recording technology: I haven't really kept up with this. So, if you find my information useful I would, in exchange, appreciate a short piece in this thread regarding what you find out about this. Once you get your search terms down, be sure to use Google Scholar to find the real science (e.g. primary literature and scientific reviews).

 

Re spinal nerves: First, be careful with terminology. A "nerve" is a component of the peripheral nervous system that consists of a bundle of individual "axons" contained in a connective tissue sheath. There is a tendency to use nerve and axon interchangeably, but this can cause confusion. A bundle of axons in the central nervous system (CNS= brain and spinal cord) is a "tract," or "commissure" when it crosses the midline (there are other descriptions, but this is safe). Axons projecting to the periphery from the dorsal and ventral horns collect together outside of the cord into a single nerve. Sensory and motor axons branch into smaller and smaller nerves that are all named and their terminal fields are pretty well known, although there is quite a surprising amount of variability. I am pretty sure that axons that are going to branch together stay together. Isn't that nice? However, the path that axons take within a peripheral nerve tends to weave about, so that any topography that might exist would be hard to reliably demonstrate.

 

I hope this helps. SM

Edited April 20, 2011 by SMF

[RobotTemplar]

Re spinal nerves: First, be careful with terminology. A "nerve" is a component of the peripheral nervous system that consists of a bundle of individual "axons" contained in a connective tissue sheath. There is a tendency to use nerve and axon interchangeably, but this can cause confusion. A bundle of axons in the central nervous system (CNS= brain and spinal cord) is a "tract," or "commissure" when it crosses the midline (there are other descriptions, but this is safe). Axons projecting to the periphery from the dorsal and ventral horns collect together outside of the cord into a single nerve. Sensory and motor axons branch into smaller and smaller nerves that are all named and their terminal fields are pretty well known, although there is quite a surprising amount of variability. I am pretty sure that axons that are going to branch together stay together. Isn't that nice? However, the path that axons take within a peripheral nerve tends to weave about, so that any topography that might exist would be hard to reliably demonstrate.

 

I hope this helps. SM

That does help, thank you. The main part of the confusion that I had coming into this thread is resolved. Though really, I'm just going to have to do some hard studying for the level of detail I am interested in. Questioning an expert in real time, I think, is a very efficient learning strategy and an art form, but it doesn't translate so well into a forum medium.

 

Re current noninvasive microstimulation and recording technology: I haven't really kept up with this. So, if you find my information useful I would, in exchange, appreciate a short piece in this thread regarding what you find out about this. Once you get your search terms down, be sure to use Google Scholar to find the real science (e.g. primary literature and scientific reviews).

I will give a shot at this, since you took the time to answer my questions.

 

I will add this in in another post by, let's say, 6pm tomorrow. I intend for my major to be in either electrical engineering or bioengineering (plus I want to learn about BCIs regardless), so it is easy to justify this time expenditure.

[SMF]

RobotTemplar, I am glad to have helped. Because you are a serious student I would like to caution you that one of the biggest challenges of teaching science is to simplify complex information without distorting it too much. For example, further study will reveal to you that there are several axon tracts in the spinal cord that carry information other than simple touch and motor control and that both sensory and motor information are processed within relay stations and parallel systems, not just in the cortex. The nervous system she is not so simple. SM

 

 

[RobotTemplar]

Okay, my report is a little late.

 

First of all, thank you for introducing me to google scholar. Although I can't read any of the articles (wants $), the place is a treasure trove of information. I suppose I am accustomed to being able to find any information I want digitally (or thinking that I am able to do this), but perhaps this is not really the case for science.

 

----------------------------------

 

I spent a couple hours searching around, but unfortunately I have not been able to really add on to my basic understanding. Electrodes can get smaller, and circuitry or programming can get smarter, but it seems like the basic mechanism for a nerve interface is pretty uniform. With a brain computer interface, you have an electrode that is placed close enough to pick up electrical signal residue from neurons (as to how this works in detail, I do not know - electricity is still mostly a mystery to me). Nerve stimulation is the same concept in reverse. Apparently you may also be able to achieve stimulation using magnetism - but I am not yet an electrical engineer ... I can't explain to you how this works.

 

The capabilities of noninvasive external (outside the skull) electrodes are very limited because the skull blocks signals, and for some reason you are only able to get a reading of more global brain activity with these electrodes (I think read this in a book about frontiers in brain science, called "The Scientific American: Brave New Brain"...I may have also read something to this effect in "More Than Human" by Ramez Naam).

 

I am assuming you have heard about the monkey controlling a robotic arm with a BCI. That is what got me really interested in this. Basically this sort of thing works because there is a lot of redundancy in the brain (protection against injury, and/or somehow related to neuroplasticity?). So implanted BCIs can be pretty powerful. But of course a robotic arm with only a couple of actuators is very different from a human or animal arm, which has probably thousands (hundreds of thousands or millions?) of muscle fibers, allowing for much finer control.

 

--------------------------

 

As for sources, I mentioned the two books I have read about this. Most of anything else I have said will also be covered in the following article:

 

http://computer.hows...r-interface.htm

 

If this site is not trustworthy, then I am pretty much screwed, because 90% of my technical education up to this point has come from working through its articles....

 

 

I am sorry if I didn't tell you anything new. The above pretty much summarizes my understanding of this subject.

Edited April 22, 2011 by RobotTemplar

[SMF]

RobotTemplar, you can always come back here when you are the expert. More about Google Scholar- I have found that trying several searches with different key words often brings up more information. Watch for the "PDF" at the right of each result item because this usually indicates a full text article, and sometimes "HTML" in this position does as well. When the article you want is behind a paywall click "All (some number) Versions", bottom right on each search result item, because there may be one that isn't restricted. Failing that I have sometimes found a full text article or a good description by pasting its whole title into a regular Google search. For example, the article may be up in a preprint form on the web page of one of the authors or the paper is in a book of collected papers and the bookseller allows you to read excerpts.

 

Obviously a good way to follow up on a topic is to use the cited references in an article to see what has been done before that is closely related to an article, but notice "Cited by" (some number) lower left by each search result item. This provides newer research that is citing the one you are looking at and expands your search to a relatively restricted group of papers. Also the "Related articles" and the "Advanced Scholar Search" links are also useful. SM

Edited April 22, 2011 by SMF

[tar]

RobotTemplar and SMF,

 

Nice discussion. Thank you. My interest in neuroscience is an attempt at understanding the physical structure of our brains and body, in relation to language. That is, I have an interest in understanding the meaning behind words. And given the fact that we communicate the best with other humans, I jump to the conclusion that that is because we come with the same equipment, and hence "already know" something about the other mind we are communicating with. So in learning what has been discovered about how we are put together, I learn a lot about what it is we share, meaningwise, when we communicate.

 

Along the lines of this thread however, I do have a thought, and a question. The "plasticity" that was mentioned, I take means that one "setup" in the system, might be repurposed, if need be, and I was wondering if this ability was an important thing to understand, in the evolution of our system, in the growth of our system and in the functioning of our system. For instance, given a set of sensory neurons, located in a particular location of the body, say a finger, and a set of motor neurons that can fire and move that finger, and a brain that can remember what is sensed as a result of a particular firing order and timing, there is a certain amount of "learning" that goes on. I heard we have a "predictive motor simulator" that rehearses the firing order and timing "before" actually firing the motor neurons that will activate various muscles resulting in coordinated motion. So my question is, how plastic IS the system? For instance, if you took some "close point" area of skin surface and attached a device that could deliver variable pressure and heat to each "point", and controlled the heat and pressure based on the frequency and intensity of focused light hitting the device(which would have a sensor system layed out with a one to one correspondence with the skin "points"), could you learn to "see" your environment as you pointed the device around, and "learned" the patterns of sensation?

 

Regards, TAR2