MECHANICAL LEARNING What is mechanical learning? A machine is any system of objects interacting via cause and effect across a space time distance. Mechanical learning is any learning that takes place when one object B receives an effect from another object A separated from B by a space/time distance. In the absence of B receiving an effect from A there is no learning possible about A, so we can conclude that the only way that B can learn about A is if A has some effect on B. That effect IS the learning that B has of A. From that effect B will then *INFER* other things about A. This is called inferential learning. It is also called: Learning by being an effect rather than learning by being a cause. Learning by looking at effects rather than learning by looking at causes directly. Learning by looking at symbols rather than learning by looking at referents. In the above case B is trying to learn about A by receiving an effect in itself from A. A is the original referent, and the change in the state of B is a symbol once removed from A. B thinks: "Well since I am receiving photons of 5000 angstroms from A, A must be painted red." All machanical learning is inferential learning, and all inferential learning is mechanical learning. Of course B can't know that the photons are coming from A, God could be arbitrarily injecting them into the space time stream just to annoy B, or B's equipment could be broken, thinking it is seeing photon's when it isn't, or thinking it is seeing 5000 Angstrom photon's when they are in fact 50 Angstrom photons. Another thing to notice, once the photon's reach B, A could be completely gone, because it takes TIME for that causal messenger wave carrying information about A to get to B. In fact, even if A is 'still there', the original A that emitted the photon at B *IS* gone, to be replaced by another A slightly later in time by the time B gets that photon. So B is never at any time in direct contact with A, the original referent, but only in contact with effects in itself allegedly emmanated by A in the past. Present time B can never know about present time A across a non zero space time distance. The best B can do is know about a past time A across a space time distance, and that knowing is very indirect because B has no direct contact with A but only contact with itself and the changes in itself allegedly caused by A. So at best B can receive an effect and infer it is from A, but attaining certainty that it came from A or what the true nature of A is, is fact certainly impossible. Mechanical or inferential learning can never produce a perfect certainty, but only a theory, a probability, a good bet about the nature of A. 100 percent certainty is not just another higher kind of probability. 100 percent certainty is not merely a quantitative step away from 99 percent probability, it is an *INFINITE* unbridgeable QUALITATIVE step away from 99 percent probability. 100 percent certainty and 99 percent probability are not even the same order of thing. 100 percent certainty isn't even on the scale of what mechanical learning can produce. 100 percent certainty is certainty of TRUTH. 99 percent probability is probability of dependable followingness, it has nothing to do at all with truth. One's theories can work very well most of the time and be completely wrong. The scientific reasons for these things are a bit steep and will take some thought. Mechanical learning takes place in B, by B changing state, as an alleged result of A. But, B's present state does not imply a change in state, i.e. a different prior state. There is nothing in your present state that proves you were in a different prior state. So after B has changed state, it can't know that it has changed state! After B has changed state, all B has is its (allegedly new) present state, which state does not contain the information that it was in a different prior state. There is no possible state in B that absolutely guarantees that B has changed state, i.e. was in a prior different state. Thus B can never know it has changed state for sure, and thus can never prove there is change or time for sure. It B can't prove that it has changed state it certainly can not prove that it received an effect from A! Even if B had a 'picture' of its prior state, and compared it to its present state and they 'looked' different, any part of B's circuitry could be misbehaving to make either the memory picture wrong or the comparison process wrong. Mechanical 'memory' and human memory are both fallible for the exact same reason, anything at all can get in there and mess up the recording and leave behind no trace that it has. Now this is diffcult but very important. Suppose we build the following simple machine. The machine has two bits, either bit can be zero or one at any time. Bit 2 will be used as a memory picture of bit 1. Bit 1 is a bi stable flip flop. Everytime a photon hits it, if it was 0 it changes to a 1, and if it was 1, it changes to a 0. Once the flip flop has changed state it locks in its new state until freed by the machine to change again when the next photon comes in. Photons that come in during the locked period are ignored. Every second the machine does the following: 1.) It compares bit 2 and bit 1, if they are same it does nothing and sleeps until the next second. 2.) If the two bits are different, the machine prints 'DIFFERENT' on a paper printer, then it copies bit 1 over to bit 2 so they are the same again. The machine then unlocks bit 1 so it is free to change again and goes back to sleep for another second. Bit 1 is open to the environment to receive influences and causal waves that might hit it. Every time a cause wave hits it, bit 1 changes from 0 to 1 or from 1 to 0, unless of course it is locked from having already been hit once, and not yet unlocked. Bit 2 is heavily protected against the environment, it is used as the the 'memory' of bit 1. We don't want ANYTHING to change bit 2 except the copy operation from bit 1 every time bit one changes state. Say both bit 2 and bit 1 are zero and the machine is sitting there 'listening' with bit 1 open to the universe. A photon comes in and hits bit 1 changing it from zero to 1. A short while later the machine wakes up and compares bit 1 to bit 2 and sees that they are different. The machines prints 'DIFFERENT' on the paper, copies bit 1 over to bit 2 so both are now 1, unlocks bit 1 and goes back to sleep. So every time bit 1 changes, a second or so later this is shown on the paper by the printed line. During this experiment you get to build the machine and test it out originally, but once the experiment is going, you only get to observe what is printed on the paper, and not what is happening to bit 1. Now let's ask some questions. Can we be absolutely perfectly 100 percent certain that every time the word DIFFERENT was printed on the paper that bit one had actually changed state? Can we be absolutely perfectly 100 percent certain that all the times the word was NOT printed on the paper that bit 1 had not changed state anyhow? Would you be willing to bet your ETERNITY IN HELL that the paper record matched the changes in bit 1 *PERFECTLY*. Can you think of ANY circuitry (chain of cause and effect) that would make the correlation between the paper output and bit 1 to be 100 percent certain? Well more of the same produces more of the same, no matter how many circuits you add, or how complex you make the machine, it is always learning by being an effect, by cause and effect sequences that in the end only follow each other *THEORETICALLY* and not certainly. So no machine can ever produce absolute certainty. If B trusts its own circuitry, then B may conclude that it has changed state and get on with theorizing about A. But its only trust, there is no absolute way for B to know for certain that it really has changed state because state does not prove prior change in state. Since B can't know if it changed state, how can it know it received an effect from A? No certainty of receiving an effect means no certainty that A exists at all, let alone that it emmanted the effect in the first place. So we put a guy in a tank with the TV set connected to the outside video camera. Can the guy be absolutely certain that what he sees on the TV set is actually out there feeding photons to the video camaer? No of course not, any part of the circuitry could be bad, firing when it isn't supposed to, or not firing when it is. Can the guy put MORE video cameras on the outside and the inside so that the circuity can look at itself and compare itself against known circuit diagrams of how its all supposed to look, in order to verify with absolute certainty that nothing has been changed, altered or is non functional? This is called the self verifying machine problem. It's a deep problem in mechanics and although putting in redundant self monitoring circuitry will decrease the probability of a malfunction from going unnoticed, it can be proven mathematically that perfect certainty of circuit integrity is impossible, no matter how many video cameras you have checking other video cameras. Pefect mechanical self awareness is impossible. Beyond reasonable doubt is a practical issue of dependable followingness, are we willing to bet on it if we have to, it has nothing to do with perfect certainty. Notice this all arises because everything is trying to learn about other things by looking at itself. You can never learn with certainty about A by looking at B even if you ARE B. Let's say this again. Mechanical learning means learning about A by being B and looking at B. B can never prove A because of the third party law. The third party law states that if B follows A, then either A may have caused B to follow, or some third party C caused both A and B to happen in such a way that they look like they caused each other to happen. A movie projector is a third party to a movie picture of a ball bouncing off a wall. In the case of a movie, there is no cause between the ball and wall, even though it looks like there is. Thus mechanical learning can never produce an absolute certainty. The video cameras that are looking at the other video cameras are actually looking at effects in THEMSELVES to determine the states of the other video cameras they are worried about. Thus they are looking at a different object (themselves) in order to learn about the other cameras. Consciousness on the other hand can produce a perfect certainty, The following experiment depends on how conscious you are. Those that aren't very conscious won't be able to do the experiment properly. Mockup two different colored squares and put them 3 feet in front of you. Verify that you did so and THAT THEY ARE TWO DIFFERENT COLORS. Do you see the perfect certainty? Find two different colored objects and put them in front of you and close your eyes. Now open your eyes and ask 'Are two different colors still there?' See the two different colors. Verify over and over again that the answer is "Yes I see two different colors FOR SURE, no way I can be wrong, this is not a theory, this is not an inference, I am in direct contact with the original referent (my conscious picture) right there in front of my nose." You are learning about them by being in direct contact with THEM, not by looking at some change in state in yourself the looker and inferring back that the looked-at must THEREFORE be different. This ability to learn by looking directly at the cause can only take place if there is no space or time between the looker and the looked-at. Space and time separating two objects makes them TWO DIFFERENT OBJECTS, and then we are back to being one of them and trying to learn about the other by looking at changes in ourselves. That's mechanical learning. Space/time -> TWO DIFFERENT OBJECTS -> MECHANICAL LEARNING But when we see the two different colored mockups, we are looking AT THEM, directly, something a machanical system can not do. Certainty of our own mockups thus implies an absence of space/time between the looker and the looked-at, no matter how much the mockups look like they are OUT THERE, they aren't. Within consciousness at least, space/time is an illusion. If they were out there, you wouldn't be able to see them at all. You would only be able to infer they might be there from some effect you saw in yourself, and by the time the effect got to you, the cause of the effect out there might have gone away completely so all you could ever glean was an inference about the past. Since you can actually SEE the mockups out there, and you KNOW THEY ARE RIGHT THERE RIGHT NOW, that means not only is there no space between the looker and the looked at, but also no time, the seeing is instantaneous, consciousness is always seeing itself in absolute present time. No machine can do that. No machine can use any part of its circuity to inspect any other part of its circuitry and observe its state as it is in present time. The inspecting circuitry can only see a past rendering, rendition of, the inspected circuitry, never the thing itself, and never as it is now. Thus in an absence of space/time between looker and looked-at, we must conclude that consciousness is a zero dimensional scalar operating actuality, and that the process of perfect certainty is one where cause and effect are the same event, i.e the learner and the learned about, are all one and the same event. Mechanical learning always implies two events, the learned about out there emitting a causal messenger wave towards the learner, and then *LATER* somewhere *DISTANT* the learner receiving the messenger wave, changing state, trusting that it has changed state, and then infering by deduction that because it has changed state here and all effects are caused, that there must have been a cause out there. That's not the process of seeing a mockup and learning the mockup has two different colors. Intensive and honest study of how a machine with video cameras for eyes learns that an external object has two different colors, and how a conscious unit learns that two of its mockups has two different colors, will go a long ways to clarifying the point that learning in consciousness is *QUALITATIVELY DIFFERENT* than mechanical learning, and can produce a perfect certainty where mechanical learning can not. Ramifications on the zero dimensional scalar nature of consciousness can then be drawn, with further study directed at how one might interface a zero dimensional consciousness with a 4 dimensional brain. (3 space and 1 time). This material is weirder than Special Relativity, don't expect to get it in a day. Homer ------------------------------------------------------------------------ Homer Wilson Smith The Paths of Lovers Art Matrix - Lightlink (607) 277-0959 KC2ITF Cross Internet Access, Ithaca NY homer@lightlink.com In the Line of Duty http://www.lightlink.com Sun Jan 1 20:41:01 EST 2006