Tag Archives: Uploading Brain

197. Brain and Computers Compared (Revised)

28 Dec

Brain-Vs-Computer-Pic2This essay is an addendum to my previous blog-193 “Brain uploading ideas are nonsense.”  It also provides my understanding of what is known about human brain functioning.

Those interested in downloading a human brain into a computer, and have it function, such as answering questions, should consider this. Although the human brain can be called a computer, it is profoundly different from a man-made computer. THE FIGURE above illustrates the fixed regularity of a computer (upper box) with every component specified and with definite location. The Brain below shows a changing and branching structure, where the actual array of components is the “program.”  Every human experience makes at least a slight structural change. Converting from an entire brain configuration with quadrillions of control points,  to an equivalent computer representation, I assert is impossible. (More explanation below.)

BRAIN AND MAN-MADE COMPUTERS ARE VERY DIFFERENT

Man-made computers have a FIXED structure of electronic components. Modern microchips have thousands of transistors and other components in a well-defined, set pattern. Computer programs, cause the chips to function. When a computer is activated by a program, it makes use of many of the components, such as microscopic transistors, so that something meaningful can go to output devices like printers or screens. In a word, the electronics are fixed and the program varies. Note: a “computer program” is a set of instructions that can be written down or typed into a file that can be “read” by a computer. It tells the computer how to utilize its well-defined components in performing certain tasks. There can be many programs to provide for a wide variety of computer functions.

The human brain is nothing like this. A live brain in a body, is constantly changing. Instead of using an external set of programming statements to control the action, the control components themselves (such as transistors) quickly vary to produce a certain output. These variations include additional nerve cells, additional connections, and each connection (a synapse or gap junction) can undergo substantial change. A synapse, based upon its chemistry and inter-connections, can be more excitable, blocked, or anything in between. The nerve cells, axons, and dendrites form branching structures that provide many paths for its operation.

THE BRAIN FUNCTIONS BY CHANGING ITS STRUCTURE

If you look at a magnified computer microchip you will see regular rows of components. The components will vary in different places, but within each part, there is exact duplication.

But in a brain, you will see great variation in size, shape, and connectivity. The “program” is built into the “brain electronics” and varies as the individual is exposed to many different environmental factors, such as learning, imitation, memorization, food supply, social features, etc.

THE BRAIN IS LIKE A CITY, WITH MANY ROADS AND PATHS

Here is a helpful analogy. Think of the brain as a city with millions of roads and pathways. The sensory systems, like vision, hearing, and touch are roads leading into the city — and there are roads going outbound causing the movement of muscles, activating vocal cords, affecting glands, etc. Between the inputs and outputs are millions of pathways with wide streets and narrow paths, free-flowing and constricted in various ways, all to provide something meaningful. At millions of junctions (synapses, etc.) there are traffic cops that speed things up or block the passage. If someone pinches your arm, action potentials flow down wide roads (axons) leading to the brain and spinal cord. Within these structures there is a maze (inter-neurons) that quickly processes the input information and activates outbound pathways, or simply stores info. (This maze is where complex thinking takes place.) The outbound paths (motor nerves) lead to many muscles that can move your arm, body, and make you say “ouch.” If you are pinched over and over again, the relevant pathways tend to widen, causing increased flow and a more efficient reflex. In our city (brain) there are also archive centers that store memories, and other mechanisms that enhance the whole process.

The ability of a brain to function in this way developed over several billions of years. It is a marvelous structure and its exact functioning at the nerve cell level is still unknown. But we can make good educated guesses as to how much of it may work. We don’t know exactly how words and ideas are stored. And we don’t know exactly how millions of memories are accessed during a wide range of operations, and how a brain creates an essay or scientific theory. Yet, the fact that there are many trillions of “control points” (like transistors) suggests that amazing things can result.

A DEAD ISOLATED BRAIN CANNOT PERFORM IN ANY WAY

I might add that a disembodied dead brain can never function like it did live in a body even if fully preserved. Cutting out a brain removes inputs, outputs, the spinal cord, and feedback loops that are critical to functioning. And would involve damage to the brain itself. A removed brain can never be the same as a live intact brain so any hope of living-on in a computer representation is impossible.  Any sectioning of a brain would destroy many interconnections.

Consider this fantasy. You transfer a brain into a computer, and it can talk. Most likely it would scream, “I can’t see, I can’t feel my legs or arms, I can’t hear, I can’t touch  anything,  where am I, HELP!” How could you talk to this isolated brain that has no ears. So you would really have to save a whole head and even the spinal cord. As you contemplate all of these difficulties, it is clear that trying to save a human brain, which can function, is nonsense. Also, for all the reasons outlined above, it is not possible to “read out” specific thoughts or ideas, by any brain recording method.  (Note: recording sub-threshold “thoughts” from vocal cords is not a brain recording.) 

I might add that although uploading a useful brain, or reading specific thoughts, is not possible, what is remotely possible is a good understanding as to how a brain functions at all levels. This could take decades or even centuries, but it is worth discussing and exploring.

 

193. Brain Uploading Ideas are Nonsense

26 Nov

Several writers have suggested that “someday” technology will advance to the point where a brain can be “uploaded” to a man-made computer, where it will be able to function and communicate with us. In some cases, these writers seem desperate to live on after death, by transferring themselves into a computer where they will continue their life. Here are the names of some brain-computer “speculators:” Ray Kurzweil (newsletter), Michael S.A. Graziano (TED talk Oct 2019), Greg Gage (TED talk June 2018) and several others. Most of the “speculators” are brilliant people, who are simply overly optimistic. But a company called Nectome will preserve your brain (“backup your mind”) so that it can be scanned hundreds (?) of years from now (deposit is $10,000). See MIT Tech Rev, Mar 13, 2018.

I am writing on this subject not only to challenge extreme speculation, but because this discussion serves to illustrate the complex nature of our brains and certain technology limitations.

My position is that this thinking belongs to science fiction because there are insurmountable obstacles for this endeavor. It is true that we can explore smaller and smaller biological objects. And computer memories continue to increase in miniaturization and capacity. But there actually are limitations because “matter” has certain limiting features. For example, our units for computer memory are close to the molecular level in size and will soon reach a limit. The speed of computers is limited to the speed of light. Even if we can ultimately reach these limits, it is likely that reliability will decrease. Certain “cosmic rays” and other factors will play a role. I maintain that the laws of physics and chemistry make real limits to tech advancement, that can never be overcome.

I think writers should not bother us with extremely unlikely speculation unless they actually can propose some ways to overcome obvious limits. I could say that someday we will travel to and explore, the inside of “black holes,” or travel to the center of the earth. But there is no value to idle speculation and it only creates confusion for the general public. Companies like Nectome (see above) will collect your $10,000 for something not possible.

Below are some relevant details.

We start with the dificulties in brain scanning. For helpful information on brain control points, see Mark Mayford et al, Cold Spring Harbor Perspectives in Biology, “Synapses and Memory Storage”, June 2012. For a simpler explanation, see Deborah Halber, Brain Facts.Org, Storing Memories in your synapses, Oct 11, 2018. Also very helpful and fascinating are Google images. Search under neurons, gap junctions, synapses, etc. and add the word “type” like “neuron types” to see more variation. As I was writing this, I looked at “gap junction types” and was amazed by the recent advances in this area.

Here are a few brief definitions:
Control point: either a synapse or gap-junction. It is like a transistor in a computer.
Synapse: a connection between two nerve cells, using chemical transmitters, which cross a tiny gap.
Gap-Junction: a connection between two nerve cells, where the cells actually touch.
Action potential: an all-or-none digital signal that transmits info up to a distance of several feet.
Axon: a long extension of a nerve cell (neuron) that carries forward info in the form of an action potential. The axon is like the wire connecting two telephones.
Dendrites: a short extension of a neuron that receives an action potential from an axon.
Nerve cell body: The cell body is supportive and provides nutrients for the whole cell. It may or may not directly contribute to info passage. The cell body has projections called axons and dendrites, which carry info in the form of action potentials.

The first question is whether you would be working with a living or dead brain. Let us start with a dead brain. In order to extract the brain itself, you would need to cut numerous inputs and outputs. The most important structure removed would be the spinal cord, which is considered a part of the central nervous system. Everything cut away from the brain destroys important feedback systems because the brain is highly interactive with the rest of the body. It is possible that even advanced thinking involves feedback loops with the eyes, nose, vocal cords, major muscles, skin receptors, etc. A cutaway dead brain might say nothing but “ouch.” Any type of brain cutting or sectioning would destroy important connections. Any type of deeply penetrating electrodes or laser beams would be partially destructive and location would be likely impossible. Many systems such as circulation and nutrition would not function and could not be fully studied. The brain emits hormones and is affected by chemicals in its blood supply. Most of these chemicals function with feedback loops and could be vital for  understanding. Scanning a dead brain might be possible, but much of the vital biochemical info would be unavailable. Even if all the nerve cells were available for some type of non-destructive scanning, the extremely tiny control points (like transistors), such as synapses would only be superficially informative. To summarize, a dead brain makes unavailable much vital info regarding its functioning and would be useless to transition to something functional.

A live brain would have all the relevant chemicals and structures, but presents another set of impossible difficulties. Who would sit still for months with millions of holes in his scull for electrodes. Or sit still for laser scanning for months to get all the connections and their efficacy, if this were even  possible. Any deeply penetrating laser beam would cause some damage or produce limited results. How would a laser beam record the efficacy of synapses or gap junctions. Also extremely important is the dynamic nature of the brain. Synapses are constantly changing in efficacy and number, and there is internal movement caused by the circulatory and nutritional systems. There will never be a way to reliably record from this impossible dynamic system with extremely tiny and numerous structures. And even if you could do some good recording from a limited area, covering the whole brain and spinal cord could take hundreds of years.

Here is a summary of the complications of brain scanning:

1. There is a very large number of components, ultimately hundreds of trillions, and sufficient information from all the components would require perhaps a thousand bits of info. The final count for all the info necessary for understanding a brain would be well into the quadrillions.
2. In addition, there are numerous  analog aspects, which trigger action potentials (digital). A synapse will only produce an action potential if it reaches a certain (analog) voltage value.
3. The living brain is a dynamic system in which the control points are constantly changing, due to use and disuse, nutritive and other factors.
4. One would have to work with tiny structures, close to the molecular level. Any prolonged scanning could and would often fail because of spurious movement. A bus driving by, or minor earthquakes, or a scientists steps could disrupt scanning. Blood flow could also disrupt.
5. If you are recording analog info at a synapse, for example, you might need to record for several minutes or longer to understand its action. Even a one-second recording for a trillion synapses would take centuries.
6. Scanning with glass or metal electrodes would cause damage to the brain cells. You might try some kind of laser system, but to go deep into the brain would require a powerful beam and  would cause damage.

For each of, say, 500 trillion control points, you would need the following data to fully understand how it functions.
1. Its exact 3-D location relative to some reference point, like 4.4342343 inches left, 3.4544546 inches right, and 3.66767667 inches deep.
2. A complete description of all its connections to other control points. Could be a thousand of such connections and the nature of each connection must be described, including its location, synapse or gap junction, neurotransmitter and any modifications to synapse efficacy.
3. Each synapse has a pre-synaptic and post-synaptic (receptor area) characteristic, which would take a 1000 or more bits to describe.
4. Gap junctions must be described by their exact location along axons and the efficacy of their connection, which would depend on the area in contact, and other factors.
5. Making sense of all the interconnections, we are definitely looking at many quadrillions (a quadrillion is 1 with 15 zeros) of data bits, probably several hundred quadrillions. It would take centuries to record all of the information.
6. The amount of data would require, probably hundreds of hard drives. How could this mass of data ever be organized and converted into a functioning “computer brain.”

.     THE HUMAN BRAIN IS A VERY UNIQUE TYPE OF COMPUTER

But, let us say that after several centuries you are able to overcome all of these problems (not likely) and you complete a scan of the brain, spinal cord (part of the CNS) and have info on inputs and outputs; all stored on a thousand (or more) giant external drives. How in the world would you be able to turn this into something useful. The human brain is a computer, but it is profoundly different from man-made computers, which are entirely digital and have a fixed structure. The human brain is a complex combination of digital, analog, and chemical operations, which is constantly changing whether sleeping or waking. It has a unique branching (tree-like) structure unlike anything man-made.

You might ask, could we someday construct a computer with the same electronic characteristics as the brain to make transfers easier. I think it is possible to someday build such a computer, but uploading info from a living brain to this computer would not be possible, because connections in this type of computer are developed through experiences (a huge variety of data inputs). For example, in a younger person, every time you learn something, there can be new nerve cells, new axons, and a variety of changes to many synapses. In an older person there may be a few new nerve cells, but there are many changes in synapses and probably other structures.

At this point, you might say, so what; after a thousand centuries, we will be able to overcome all of these problems. Which brings us to another point: Does it make sense to make theories about things that we will never be able to prove. P. W. Bridgman (Nobel Prize winner in physics) says don’t waste your time. Maybe we should leave “uploading a human brain” to the science fiction writers, particularly if you have no special facts or methods to make radical ideas more acceptable.

One more point, on a related issue. Although I assert that complete scanning or uploading a brain is likely not possible, a fairly complete understanding of how the brain works (at the level of synapses) may be possible someday.