fransvo@htsa.uucp (Frans van Otten) (04/07/89)
A lot of comp.ai writers seem to misunderstand the difference between "reality" and "simulation". Actually, both words are pointers to some actual process, which has no name, I'm afraid. Reality and simulation are "relativity" concepts. When we simulate a flying plain, it is (within the simulator) really flying (in so far as the simulator simulates flying). To us, it is a simulation of a flying plane. That is, in our reality the plane does not fly, but in the reality inside the simulator it does fly ! Let's assume we made a computer system which can calculate the physics of a person. Let's assume we can communicate with this simulation. Now to us he is a simulation. But if we asked him, he would say "Me ? A simulation ? You must be out of your mind, I do really exist !" Again, in our reality the person is a simulation, but in the reality within the computer system the person is real. Our reality is not a "universal" reality, either. The rules for our existence are the chemical and physical rules, which are being executed by the molecules. The rules for the existence of molecules are being executed by the constituting atoms. The rules for the existence of atoms are being executed by the constituting atoms. Et cetera. If we would execute the rules for the molecules on a computer, we would not notice any difference except that nuclear fusion would not be possible. The only problem is that a simulated reality is not physically in contact with the simulating reality. So I could never feed myself with a simulated meal. But that is also a very nice property of simulations: whatever crashes a simulated plane makes, no real person dies ! And nobody cares about all those killed simulated people, since we didn't simulate their family and friends. -- Frans van Otten Algemene Hogeschool Amsterdam Technische en Maritieme Faculteit fransvo@htsa.uucp
ssingh@watdcsu.waterloo.edu ( SINGH S - INDEPENDENT STUDIES ) (04/09/89)
In article <827@htsa.uucp> fransvo@htsa.UUCP (Frans van Otten) writes: > >Our reality is not a "universal" reality, either. The rules for our >existence are the chemical and physical rules, which are being executed >by the molecules. The rules for the existence of molecules are being >executed by the constituting atoms. The rules for the existence of >atoms are being executed by the constituting atoms. Et cetera. If we >would execute the rules for the molecules on a computer, we would not >notice any difference except that nuclear fusion would not be possible. > > Frans van Otten > Algemene Hogeschool Amsterdam > Technische en Maritieme Faculteit > fransvo@htsa.uucp Very thought provoking!!! Since processes in the brain are chemical reactions, I am wondering what models are the best we have to capture the rules of chemical-chemical interaction that allows information processing of an intelligent sort to take place. Can the properties of the chemicals be abstracted into something as imaginary as cellular automata? How WOULD we analyse the brain in such a way to be able to decipher the dance of chemicals that is the esscence of being human? ssingh@watdcsu.waterloo.edu
lammens@sunybcs.uucp (Jo Lammens) (04/09/89)
In article <5790@watdcsu.waterloo.edu> ssingh@watdcsu.waterloo.edu ( SINGH S - INDEPENDENT STUDIES ) writes: >Very thought provoking!!! Since processes in the brain are chemical >reactions, I am wondering what models are the best we have to capture >the rules of chemical-chemical interaction that allows information >processing of an intelligent sort to take place. Can the properties >of the chemicals be abstracted into something as imaginary as >cellular automata? How WOULD we analyse the brain in such a way to >be able to decipher the dance of chemicals that is the esscence of >being human? > I wonder if there's any point in doing that. If you want to understand how a car works, you don't analyze it in terms of chemistry and quantum mechanics, although it is probably possible to do so - at least in principle. If you want to achieve high-level understanding, you should probably use high-level descriptions. That's basically why nobody wants to use machine code any more to write a complex program, although it is possible to do so - at least in principle. Jo Lammens BITNET: lammens@sunybcs.BITNET Internet: lammens@cs.Buffalo.EDU UUCP: ...!{watmath,boulder,decvax,rutgers}!sunybcs!lammens
ssingh@watdcsu.waterloo.edu ( SINGH S - INDEPENDENT STUDIES ) (04/09/89)
In article <5106@cs.Buffalo.EDU> lammens@sunybcs.UUCP (Jo Lammens) writes: > >I wonder if there's any point in doing that. If you want to understand >how a car works, you don't analyze it in terms of chemistry and quantum >mechanics, although it is probably possible to do so - at least in >principle. If you want to achieve high-level understanding, you should >probably use high-level descriptions. That's basically why nobody >wants to use machine code any more to write a complex program, >although it is possible to do so - at least in principle. > You're right. There is no reason to analyse a car in terms of quantum mechanics. That is carrying reductionaism too far. We know that in the construction of the car, there are parts. It makes sense to analyse this system of parts for a true understanding of what is going on. The basic unit of information processing in the brain is the neuron. Clearly tons of neurons put together will give us some form of behaviour. It makes sense to abstract the properties of the neurons into some sort of precise model. Then do simulations (soft or hardware) that show us a system of N neurons. But the model MUST be consistent with the actual properties of biological neurons. Other simulations could be constructed with different unit properties. >BITNET: lammens@sunybcs.BITNET Internet: lammens@cs.Buffalo.EDU >UUCP: ...!{watmath,boulder,decvax,rutgers}!sunybcs!lammens
lammens@sunybcs.uucp (Jo Lammens) (04/09/89)
In article <5791@watdcsu.waterloo.edu> ssingh@watdcsu.waterloo.edu ( SINGH S - INDEPENDENT STUDIES ) writes: >The basic unit of information processing in the brain is the neuron. >Clearly tons of neurons put together will give us some form of >behaviour. It makes sense to abstract the properties of the neurons >into some sort of precise model. Then do simulations (soft or hardware) >that show us a system of N neurons. But the model MUST be consistent >with the actual properties of biological neurons. [...] I think not even all connectionists would agree with your last sentence. What I was really hinting at before is that a description at the neural level *MAY* be too low to yield significant understanding, even though it is tempting and certainly fascinating. It's a bit like trying to understand how an operating system works by analyzing the function of all the transistors in the machine it's running on. They can be modeled more or less precisely, and for sure tons of them can do some amazing things. But I doubt that this kind of analysis will yield any significant understanding. Please convince me of the opposite. Jo Lammens BITNET: lammens@sunybcs.BITNET Internet: lammens@cs.Buffalo.EDU UUCP: ...!{watmath,boulder,decvax,rutgers}!sunybcs!lammens
ssingh@watdcsu.waterloo.edu ( SINGH S - INDEPENDENT STUDIES ) (04/10/89)
In article <5125@cs.Buffalo.EDU> lammens@sunybcs.UUCP (Jo Lammens) writes: > >I think not even all connectionists would agree with your last >sentence. What I was really hinting at before is that a description at >the neural level *MAY* be too low to yield significant understanding, >even though it is tempting and certainly fascinating. It's a bit like >trying to understand how an operating system works by analyzing the >function of all the transistors in the machine it's running on. They >can be modeled more or less precisely, and for sure tons of them can >do some amazing things. But I doubt that this kind of analysis will >yield any significant understanding. Please convince me of the >opposite. > You analogy seems correct, but what alternative is there? We could study simple brains, but we lose the idea of "emergence." There is not all THAT much difference between our brain and a monkey's. Why are we so much more complex??? You could use automata with local communications properties that you define yourself and run a simulation, but these are pure abstractions; there does not seem to be a hope of physically realizing it, just as it was not possible to physically create Von Neumann's self-reproducing automaton machine. BY THE WAY, THE WORLD IS DISCRETE, NOT CONTINUOUS (in response to past postings). Now, about the operating system parallel. To me, it is more important to study the structure of the computer running the operating system we call a MIND. How does a neuron react to stimuli, how does it connect to others, how does it change itself? If you want to lift the "algorithms of thought" from a mind and transplant it elsewhere, GOOD LUCK. What do you mean when you say "...any significant understanding?" ssingh@watdscu.waterloo.edu ... a mind forever voyaging ...
arm@ihlpb.ATT.COM (Macalalad) (04/11/89)
In article <827@htsa.uucp> fransvo@htsa.UUCP (Frans van Otten) writes: >A lot of comp.ai writers seem to misunderstand the difference between >"reality" and "simulation". Actually, both words are pointers to some >actual process, which has no name, I'm afraid. Reality and simulation >are "relativity" concepts. I always thought of those "actual processes" AS REALITY. Whose concept of relativity are you using? >When we simulate a flying plain, it is (within the simulator) really >flying (in so far as the simulator simulates flying). To us, it is a >simulation of a flying plane. That is, in our reality the plane does >not fly, but in the reality inside the simulator it does fly ! No, there is only ONE reality, not many separate realities. The concept behind relativity is that the MEASUREMENTS taken of reality, such as time and distance, are relative to the observer. This is quite different from claiming that each observer is taking measurements of a different reality. Thus, it doesn't even make sense to talk of a separate reality within a simulator. Where is the observer within the simulator? All of the measuring and collapsing of quantum states take place at our level, not the simulated level. Sure, we can redefine what we mean by flying to encompass the concept of flying on the simulator level, but remember that we are still talking about two different concepts of flying. (I think you do acknowledge as much implicitly in your discussion of two "realities.") Putting two concepts under the same label does not make them the same. I think that strong AI is making a stronger claim: to create an intelligent machine not in some "simulated reality" but in our reality, fully interacting with us and our environment. We may be in the grips of an ideology, but let's not lose our grip on reality. -Alex
rapaport@sunybcs.uucp (William J. Rapaport) (04/13/89)
In article <827@htsa.uucp> fransvo@htsa.UUCP (Frans van Otten) writes: > >When we simulate a flying plain, it is (within the simulator) really >flying (in so far as the simulator simulates flying). To us, it is a >simulation of a flying plane. That is, in our reality the plane does >not fly, but in the reality inside the simulator it does fly ! And I quote: "It is often suggested that a simulation of a phenomenon is not an instance of the p[henomenon being simulated. For example, simulated hurricanes are not real hurricanes. After all, as people often point out, simulated hurricanes don't get you wet. I think it is wrong to suppose that this shows that simulated Xs aren't Xs. Simulated huricanes won't get _you_ wet, but they _will_ get a simulated you simulatedly wet; if they didn't they wouldn't be very good simulations of hurricanes. The proper way to look at it is that both simulated and real hurricanes are implementations of an abstract notion of hurricane; in this way, they can both legitimately be said to be hurricanes. But suppose that it is the case, _in general_, that simulated Xs aren't Xs. Still, there might be _some_ values of X for which simulated Xs _are_ Xs. In particular, simulated mentality seems to me to be a good candidate for such an X. If I have a conversation with a computer that passes the Turing Test, it might very well be the case that I could learn something from it. To use the hurricane metaphor, it might indeed get me "wet": it might give me information in much the same way that a Xerox copy of Searle's book can give me the same information that an actual copy would. Such a copy to perhaps all but certain book-collectors, is the book itself." From Rapaport, William J. (1988), ``To Think or Not to Think'' (critical study of Searle, _Minds, Brains & Science_), Nous 22: 585-609.
dmocsny@uceng.UC.EDU (daniel mocsny) (04/13/89)
In article <5227@cs.Buffalo.EDU>, rapaport@sunybcs.uucp (William J. Rapaport) writes: > And I quote: > > "It is often suggested that a simulation of a phenomenon is not an > instance of the p[henomenon being simulated. For example, simulated > hurricanes are not real hurricanes. After all, as people often point > out, simulated hurricanes don't get you wet. Why doesn't, or couldn't, a simulated hurricane get you wet? Because we aren't so good at simulating things yet, or because that would be "cheating?" For us to detect the presence of a hurricane, we must absorb some sort of sensory data (raindrops on skin, the view of the ground rushing by as the gale carries us along, or perhaps we detect the hurricane indirectly by reading reports from instruments). To detect the presence of a _simulation_, we must also absorb some sort of sensory data. This means that the simulating device must have some link with the physical world for it to convey messages to our sensory organs. If a simulation can't deliver some sort of sensory experience to an observer, then in what sense does it exist? For a simulation to be a simulation, it must first exist...one hand clapping, that sort of thing... Thus any device that simulates things must have two essential parts: the computing engine that performs what we ordinarily think of as a simulation, and the I/O subsystem that pumps energy into the physical world in a way that provides a convincing sensory experience to an observer. The I/O subsystem can be as crude as a line printer that hammers out line after line of numbers. It can be a high-fidelity digital audio system that faithfully simulates the sounds of an orchestra (...heard any CD's lately? Simulated hurricanes don't get you wet yet, but simulated orchestras can move one to tears). It can be the future successor to NASA's VIVED (VIrtual Visual Environment Display) with stereoscopic LCD goggles and head-motion cancelling panning, that delivers a convincing panoramic view of simulated surroundings with depth cues. Someday it may be neural-interface implants that directly induce signals on our sensory nerves indistinguishable from the "real" thing. A really good simulation should saturate the observer's sensory bandwidth, just like the "real" thing does. Certainly a hurricane simulator could include a fancy device to generate raindrops and impart the correct velocity vectors to them. Spraying water as an output is not different in principle from pounding ink into a page--the simulating device has an expendable store of material that it imparts energy to and exhausts over the course of the simulation. Dan Mocsny dmocsny@uceng.uc.edu
rapaport@sunybcs.uucp (William J. Rapaport) (04/13/89)
In article <864@uceng.UC.EDU> dmocsny@uceng.UC.EDU (daniel mocsny) writes: > >Why doesn't, or couldn't, a simulated hurricane get you wet? Because >we aren't so good at simulating things yet, or because that would >be "cheating?" I was thinking of the sort of computer simulation of a hurricane that the weather service might construct, not a machine that produces a hurricane artificially. The latter ought to get you wet; the former wouldn't.
lee@uhccux.uhcc.hawaii.edu (Greg Lee) (04/13/89)
From article <864@uceng.UC.EDU>, by dmocsny@uceng.UC.EDU (daniel mocsny): " ... For a simulation to be a simulation, it must " first exist...one hand clapping, that sort of thing... Next you'll be asking what is the sound of one symbol clashing. Greg, lee@uhccux.uhcc.hawaii.edu
weltyc@cs.rpi.edu (Christopher A. Welty) (04/13/89)
In article <5106@cs.Buffalo.EDU> lammens@sunybcs.UUCP (Jo Lammens) writes: >In article <5790@watdcsu.waterloo.edu> ssingh@watdcsu.waterloo.edu ( SINGH S - INDEPENDENT STUDIES ) writes: >>Very thought provoking!!! Since processes in the brain are chemical >>reactions, I am wondering what models are the best we have to capture >>the rules of chemical-chemical interaction that allows information >>processing of an intelligent sort to take place.... > >I wonder if there's any point in doing that. If you want to understand >how a car works, you don't analyze it in terms of chemistry and quantum >mechanics, although it is probably possible to do so - at least in >principle. If you want to achieve high-level understanding, you should >probably use high-level descriptions. Perhaps I missed something by jumping in the middle of this, but it seems to me that all the articles posted on this Simulation vs Reality argument are talking about two fundamentally different concepts as if they were the same. Understanding and simulation are not the same thing. I don't know very many people, in fact I don't know any, who could accurately simluate a car, although I do know many who understand understand how it works. I would argue that accurate simulation DOES require a model from as low a level as possible in order to behave exactly as the real thing being simulated. Typically `high level' descriptions of functional groups of low level objects are mere generalizations of the function of the group, and thus only incorporate the default knowledge of that function. Christopher Welty --- Asst. Director, RPI CS Labs | "Porsche: Fahren in weltyc@cs.rpi.edu ...!njin!nyser!weltyc | seiner schoensten Form"
dmocsny@uceng.UC.EDU (daniel mocsny) (04/14/89)
In article <5244@cs.Buffalo.EDU>, rapaport@sunybcs.uucp (William J. Rapaport) writes: > I was thinking of the sort of computer simulation of a hurricane that > the weather service might construct, not a machine that produces a > hurricane artificially. The latter ought to get you wet; the former > wouldn't. The hurricane simulation that the weather service might construct outputs a list of numbers representing the velocity field of the hurricane. The velocity field is not the hurricane, but a formal description of the hurricane that we can obtain directly by planting a grid of instruments in the hurricane's path. So strictly speaking, I would not call the weather service's simulation a simulation of THE hurricane, but rather a simulation of some abstract measurements we might take of the hurricane. Also, the artificial hurricane I suggested need not reproduce every single effect of the hurricane on the real world, but only those that lead directly to sensory experience of the observer. If the observer inconveniently wants to walk around and poke through the wreckage, then we have to include more of those effects. If the observer is very persistent, we have to generate a full-blown artificial hurricane. I don't see a sharp distinction between a "simulation" and "a machine that produces a hurricane artificially," but rather that they lie on a continuum. Can we agree that a physical system has no reality outside the sensory experience it conveys to an observer? If so, then a "simulation" is the action of some artificial device that reproduces some or all of the sensory experience we associate with the "real" thing. If the simulation leaves something out, then it is incomplete. If it doesn't leave anything out, then the observer cannot tell it apart from the "real" thing. From the observer's standpoint, it IS the real thing. Every simulation must effect some change in the real world, because the observer exists in the real world and can only detect things in the real world. Two of our most important senses--vision and audition--respond only to energy fluxes from the real world. The other senses respond to both energy and material fluxes (or perhaps I should say, energy fluxes mediated by material fluxes). Perhaps the distinction we are trying to draw is that "simulating X" delivers strictly energy-mediated sensations, while "reproducing X artificially" includes whatever material fluxes are necessary to deliver the full range of sensations. But I don't see this as being a particularly natural division. Since every simulation must have some detectable real-world effect, what is the difference between an energy-only effect and an energy+material effect? They differ only in the complexity of the output devices required. Dan Mocsny dmocsny@uceng.uc.edu
bwk@mbunix.mitre.org (Barry W. Kort) (04/15/89)
In article <5244@cs.Buffalo.EDU> rapaport@sunybcs.UUCP (William J. Rapaport) writes: > I was thinking of the sort of computer simulation of a hurricane that > the weather service might construct, not a machine that produces a > hurricane artificially. The latter ought to get you wet; the former > wouldn't. And then there are the occasional storms projected by the weathermen which fail to materialize. In that case, one could say it's the National Weather Service that gets wet. :-) --Barry Kort
bwk@mbunix.mitre.org (Barry W. Kort) (04/15/89)
In article <3752@uhccux.uhcc.hawaii.edu> lee@uhccux.uhcc.hawaii.edu (Greg Lee) writes: > Next you'll be asking what is the sound of one symbol clashing. Of course you meant "one cymbal crashing". ;-) --Barry Kort
sarima@gryphon.COM (Stan Friesen) (04/15/89)
In article <5791@watdcsu.waterloo.edu> ssingh@watdcsu.waterloo.edu ( SINGH S - INDEPENDENT STUDIES ) writes: >You're right. There is no reason to analyse a car in terms of quantum >mechanics. That is carrying reductionaism too far. We know that in >the construction of the car, there are parts. It makes sense to analyse >this system of parts for a true understanding of what is going on. > >The basic unit of information processing in the brain is the neuron. >Clearly tons of neurons put together will give us some form of >behaviour. It makes sense to abstract the properties of the neurons >into some sort of precise model. Then do simulations (soft or hardware) >that show us a system of N neurons. But the model MUST be consistent >with the actual properties of biological neurons. Other simulations >could be constructed with different unit properties. > Good start. At least for a very low level understanding of mental processes. Such simulations could well give us a model of clusters of neurons. This knowledge could then be used to generate a simulation in which the "parts" are tightly coupled clusters of neurons. The point? Neurons are the basic unit of processing in the brain only at one level, a rather low level at that. It could be called the mental equivalent of assembly language. Most information processing in the brain is performed by co-ordinated sets of neurons, called nuclei, tracts, and columns. For AI purposes I think this level might be more useful, since it deals with coded information. By the way, there is also a lower level than the neuron, the synapse. A remarkable amount of processing is done at the individual synapse. And even more is done in the dense clusters of interlocking synapses called glomeruli. We are talking about a remarkably complex structure when we talk about the brain -- Sarima Cardolandion sarima@gryphon.CTS.COM aka Stanley Friesen rutgers!marque!gryphon!sarima Sherman Oaks, CA
jwi@lzfme.att.com (Jim Winer @ AT&T, Middletown, NJ) (04/18/89)
In article <50068@linus.UUCP>, bwk@mbunix.mitre.org (Barry W. Kort) writes: > In article <3752@uhccux.uhcc.hawaii.edu> lee@uhccux.uhcc.hawaii.edu > (Greg Lee) writes: > > > Next you'll be asking what is the sound of one symbol clashing. > > Of course you meant "one cymbal crashing". ;-) > > --Barry Kort The sound of one symbol clashing seems both more likely (since it was stated), and more interesting. After all, everyone's heard one cymbal crashing -- it happens all the time. You only get to hear one symbol clashing if you subscribe to comp.ai. Jim Winer ..!lzfme!jwi I believe in absolute freedom of the press. I believe that freedom of the press is the only protection we have from the abuses of power of the church, from the abuses of power of the state, from the abuses of power of the corporate body, and from the abuses of power of the press itself. Those persons who advocate censorship offend my religion.