Is the Brain Equivalent to a Turing Machine? By ZeuS572 in Technology Mon Mar 17, 2003 at 02:00:56 PM EST Tags: Science (all tags)
	From the NewScientist.com: "The world's first brain prosthesis - an artificial hippocampus - is about to be tested in California. Unlike devices like cochlear implants, which merely stimulate brain activity, this silicon chip implant will perform the same processes as the damaged part of the brain it is replacing. The prosthesis will first be tested on tissue from rats' brains, and then on live animals. If all goes well, it will then be tested as a way to help people who have suffered brain damage due to stroke, epilepsy or Alzheimer's disease."

This brings up the key question that this article will focus on - Is the Brain a Turing Complete Machine? Description of a Turing Machine Alan Turing was a mathematician and logician in 20th century. He is often considered the pioneer of computer science as we know it today. He helped the British Government crack German codes during World War II. His world-changing contribution to society was a Turing Machine. For a history about Alan Turing, the man, see: http://www.turing.org.uk/turing/index.html Introduced by Alan Turing in 1936, a Turing machine is a very simple kind of computer whose operations are limited to reading and writing symbols on a tape, or moving along the tape to the left or right. This tape is divided into squares, any square of which may contain a symbol from a finite alphabet, with the restriction that there can be only finitely many non-blank squares on the tape. The Turing machine can only read or write on one of these squares at once--the square located directly below its "read/write" head. In addition to a finite alphabet, the Turing machine has finite states. The machine is guided by a set of instructions that define the following: Given the current state and the current symbol, write x new state and y new symbol, and move the head either left or right. The state table defines these instructions. Turing machines are important because they represent what computers can and cannot do. Any deterministic algorithm can be mapped to a Turing machine algorithm. If an algorithm is described as "Turing Complete," then it meets the requirements of a Turing Machine. The Implications If our brain or a self-contained part of the brain could be mapped to a Turing machine, what would that signify? There are some significant developments that could come out of such a discovery. We could start off by the creation of Artificial Intelligence (Artificial only in the sense that it's not in a living human. It would otherwise be a copy of intelligence from a real brain). If we could map the operations of the brain to a Turing machine, then we could replicate this behavior in a machine. We would have a non-organic human mind, living in a computer. Not only will it be fascinating to have such a mind, but it could serve several useful purposes. For example, brains could be sustained and utilized with less resources than that necessary for traditional humans to function. We could have three brains operating for the price of human one. Secondly, we could modify certain functions and see how the overall model of the human brain was affected. The discoveries coming from this research could help in medical discoveries to aid real humans. For example, we could replicate an Alzheimer's brain, and keep attempting to modify it until it no longer suffered from the ailment, and then replicate the steps in true humans suffering. We could use these models to find enhancements that can be done to the human brain (outside of simply curing diseases). And finally, we could find out what a human could be like with more processing power--we could speed up a human brain and find out how much more it can process, and if it changes it's personality and/or inherent nature. Besides the discoveries in computation, we could also learn to modify the state tables of humans. Perhaps we could implant memories in humans (ala the movie "Total Recall"). Perhaps every child would be implanted with the amount of knowledge that current adults have, and they would start off with this base knowledge at an early age. Clearly, the possibilities are endless (within the limits of human imagination!). The Counter Argument - Non-Computational States A showstopper argument often given is that the brain has states that can't be measured finitely, or that follow patterns that cannot be measured mathematically. If this was the case, we wouldn't be able to map the brain to a Turing machine, and thus wouldn't be able to map it to a computer in the way that we currently create algorithms and programs. The existence of non-mathematically measurable patterns within math was first proved by Kurt Godel in Godel's Theorem, in 'response' to Bertrand Russell's Principia Mathematica. One well known author Douglas Hofstader writes in his beautiful book, <u>Godel, Escher, Bach: An Eternal Golden Braid</u> that these brain patterns are from self-referencing loops (what he calls "strange loops"). One example he gives of a self-referencing statement that creates a loop is to evaluate whether the following MEANINGFUL statement (could be an axiom) is true or false-- "This Statement is False." The statement simply makes an assertion; though because of its self referencing nature, it creates a state which is neither true nor false (or possibly both true and false). If it was a true statement, it would be false, since it proclaims to be false (and it's true what it proclaims); if it is false, then it must be a true since it tells the truth of itself. Hofstader believes that consciousness is comprised of such self-referencing loops. Also, consider this--If the brain modifies itself, then we could possibly never map it to a Turing machine, because the instruction table would keep changing. The self-modification aspect would be hard to replicate. What does the reported discovery promise? Though this discovery in itself is not enough to say whether the brain is or isn't a Turing machine, it does bring it once step closer to it being a mapping to a Turing machine (albeit a very small step). Though they couldn't understand how it worked, they could copy it (e.g. mimic the state table and the instruction set). It's important to keep this in perspective: They haven't yet tested it fully, but they are as closer to this than any has been before. That's what makes this quite exciting. The road to replicating AI is much longer. We would have to finish successfully replicating the hippocampus (which is the easiest part), and then move on to the more complicated parts. Presumably, if we could find a method to replicate without understanding, we should be able to use this method in any part of the brain. "No one understands how the hippocampus encodes information. So the team simply copied its behaviour. Slices of rat hippocampus were stimulated with electrical signals, millions of times over, until they could be sure which electrical input produces a corresponding output. Putting the information from various slices together gave the team a mathematical model of the entire hippocampus." Also, of interest is the following fact: The brain processes information in parallel, whereas a Turing machine is ultimately serial. In order to mimic a parallel machine in a Turing machine, you would have to have an exponentially larger number of states. And yet they did it with computer technology available today! What about consciousness? The researchers haven't yet gotten that far. The experiment currently focuses on the Hippocampus. But they will try: "The researchers developing the brain prosthesis see it as a test case. `If you can't do it with the hippocampus you can't do it with anything,' says team leader Theodore Berger of the University of Southern California in Los Angeles. The hippocampus is the most ordered and structured part of the brain, and one of the most studied. Importantly, it is also relatively easy to test its function." If they can do it to the hippocampus without understanding its functions, who knows what will happen with deeper levels! Part 2: Ethics What does this mean with regards to the value of human life? If we can be easily replicated, copied, or replaced, what will be left in our inherent humanness? Would you support behavior modification? Would you support implanted memories? (ala "Total Recall") One other important ethical point is that of the "life" of the machine that was created (via replication). Even if the machine is not living in the traditional sense, it still is aware of itself and has the functions of a human. Would it be ethical to then modify this behavior? Could the machine feel pain as we do? If we create weird states for it, who experiences these states? And if you argue about protecting this machine, then what about animal research today? Do you support that? Some More Reading About Turing Machines: http://www.ams.org/new-in-math/cover/turing.html http://www-csli.stanford.edu/hp/Turing1.html http://grail.cba.csuohio.edu/~somos/bb.html The Myth of the Turing Machine An Alternative View of CS:Quantum Computing AI- Is the Brain a Turing Machine? A Google Query Godel, Escher, Bach: An Eternal Golden Braid The Singularity Institute Creating Friendly AI

Related Links

Google http://www .turing.org.uk/turing/index.html http://www .ams.org/new-in-math/cover/turing.html http://www -csli.stanford.edu/hp/Turing1.html http://gra il.cba.csuohio.edu/~somos/bb.html The Myth of the Turing Machine Quantum Computing A Google Query Godel, Escher, Bach: An Eternal Golden Braid The Singularity Institute Creating Friendly AI Also by ZeuS572