X-Message-Number: 15466 References: <> Date: Sun, 28 Jan 2001 13:59:28 +0000 From: "Joseph Kehoe" <> Subject: proof of brain speed >Message #15461 >Date: Sat, 27 Jan 2001 08:06:44 -0500 >From: Thomas Donaldson <> >Subject: answers to 2 questions > >Hi everyone! > >2. Once more on neurons, brains, and computers: basically we do not > have a situation in which we have some large subset of our neurons > which connect totally to one another. Our connections take up only > a subset of those possible. Just how big a subset and just which > neurons connect to others has some commonality between us, but > when we look at details we are all different. Anyone who reads > PERIASTRON and thinks about what its articles say will know this. > > Moreover, our fine connections are not stable (recent discoveries > verify this). That is why the number of possible connections is > closer to 2^N and than N^2... though the number N here is hardly > well known, as yet. Yes, at any instant Mike Perry or Thomas > Donaldson will have a given set of connections between a given > set of neurons, and less than N^2 of them, too ... but since our > experiences will cause new connections and wipe out some older > ones, this fact hardly seems important. If we wanted to make > an instant copy of Mike Perry, for preservation, his exact set > of connections WOULD become important, but only for that purpose. > Once we revive him, his set of connections will start changing. > > The value of parallel computing here is that living in the real > world we must do several different "mental" processes all at the > same time. Not interleaving them, but really at the same time. > Most of these are unconscious, but they're hardly less important > because of that. I will mention breathing and heartbeat as two > examples, but hardly a complete set. There are enough of each > that we cannot really expect to work well with a single computer > at all ... which does not mean we might not be able to THINK > well, but instead that we could not SURVIVE well at all. I have no doubt that parallel computing will be used as the solution of choice to simulate brains (if we ever get that far). single processor or parrallel makes no difference to me as long as it works. > I hope this point explains why I think numbers like N^2 mean > very little if we want to find out how brains work and preserve > them. As for parallel computing, even in computing, its value > comes because it lets us do things which otherwise would take > far too long ... in some cases, even now, things which would > literally take millions of years, when done on a single fast > sequential computer. And that's why I think Turing neglected > timing, and timing is important enough that it should NOT be > neglected. Turing machines do not mention timing at all because the amount of time required to do a calculation plays no part in determining if it can be implemented by an algorithm. For practical uses timing is very important but that is not what Turing machines are about. They are a model of " computable algorithms". The thesis is anything that can be done by a Turing machine is a computable algorithm AND any computable algorithm can be implemented as a Turing machine. From a theoretic perspective all timing is a non issue as to whether Turing machines can simulate us. If TM's can simulate us it means we are computable algorithms and this would be the discovery of the millenium if proved. Whether we can ever get computers to work fast enough to do so in real time is a seperate practical issue. So we have two problems 1. Are we computable (by Turing machines - the best model we have of computable algorithms)? 2. Assuming we are computable will we ever have the power to compute one of us in real time? I am only arguing that the answer to one is yes whereas I suspect you are only arguing that the answer to 2 is no. If we restrict ourselves to 2. We assume we are computable, but you say that the computation is so complex we will never do it in real time. For this to be the case the computation is NP-complete (or better yet NP-hard if I remember my terminology correctly) and the computation speed required is X. We will never reach X, ever, ever. Now for this to be the case X would have to be extremely high (duh) and moores law would have to meet some universal law that stopped it in its tracks. This law would prob. involve the speed of light c and Shannons Information theory. You need to find 1. This law which gives us the upper bound of computation, say N 2. The processing required by one of us, say X 3. Show that N < X before you do that however... Assume 3 is True. But in my cranium I have a computation machine runs at X, as does everyone on this list. Therefore either 3 is not true or our assumption was wrong. Our only assumption is that we are computable. Therefore if we are computable speed is not an issue. If you only wish to argue that it will never be done on a single processor then I won't argue as it is irrelevent how we actually do it as long as we can do it. Remember a Turing machine is not a single processor, it is not even a processor just a model of algorithmic computation. With this in mind the only question is: Are we Turing computable? If we are not then either We are unable to model neuron We are unable to model neuron connections I cannot see how either of those could be true unless 1. Some mysterious quantum effect is involved - of which we have no evidence 2. We have duality ( a la Descartes) of brain and mind. This seems to be pure superstitious nonsense to me but ... Joseph. Rate This Message: http://www.cryonet.org/cgi-bin/rate.cgi?msg=15466